Bernard Williams is one of the most influential British philosophers, and deservedly so, considering his brilliance, his erudition, his wit, and his involvement in public affairs. In Ethics and the Limits of Philosophy and Descartes: The Project of Pure Enquiry1 he tries to win us over to a metaphysically materialist view of the world, while recognizing that the sort of reduction of the intentional to the non-intentional that Fodor and Millikan hope to achieve is impossible. At the same time he tries to achieve a reconciliation of relativism and a weak form of cognitivism,2 by exploiting the Kuhnian idea of incommensurability.3 Given the ambition of these two books and the distinction of the author, I believe that it will be well worth while to examine his doctrines here.
Although I shall be interested in seeing how a successor to the fact-value dichotomy, if not the traditional fact-value dichotomy itself, makes an appearance in Bernard Williams’ philosophy, I shall begin with Williams’ metaphysics, because that metaphysics—and especially his notion of an “absolute conception of the world”—is at the base of all of his arguments.
The Absolute Conception and the Metaphor of Perspectives
The notion of an “absolute conception of the world” as something that science ideally converges on (and ethics does not, even ideally) is employed in the closing chapters of Ethics and the Limits of Philosophy.4 Thus, Williams writes,
The basic idea behind the distinction between the scientific and the ethical, expressed in terms of convergence, is very simple. In a scientific inquiry there should ideally be convergence on an answer, where the best explanation of the convergence involves the idea that the answer represents how things are; in the area of the ethical, at least at a high level of generality, there is no such coherent hope. The distinction does not turn on any difference in whether convergence will actually occur, and it is important that this is not what the argument is about. The point of the contrast is that, even if this happens, it will not be correct to think it has come about because convergence has been guided by how things actually are, whereas convergence in the sciences might be explained in that way if it does happen, (p. 136)
In reflecting on the world that is there anyway, independent of our experience, we must concentrate not in the first instance on what our beliefs are about, but on how they represent what they are about. We can select among our beliefs and features of our world pictures some that we can reasonably claim to represent the world in a way to the maximum degree independent of our perspective and its peculiarities. The resultant picture of things, if we can carry through this task, can be called the “absolute conception” of the world. In terms of that conception, we may hope to explain the possibility of attaining the conception itself, and also the possibility of other, perspectival, conceptions.
The notion of an absolute conception can serve to make effective a distinction between “the world as it is independent of our experience” and “the world as it seems to us.” It does this by understanding “the world as it seems to us” as “the world as it seems peculiarly to us”; the absolute conception will, correspondingly, be a conception of the world that might be arrived at by any investigators, even if they were very different from us … The aim is to outline the possibility of a convergence characteristic of science, one that could meaningfully be said to be a convergence on how things (anyway) are. (pp. 138–139)
The notion is explained more completely in Williams’ earlier Descartes.5 Like Descartes, Williams accepts a sharp distinction between “primary” qualities (such as extension, but in presentday physics also mass, charge, the various field intensities and so on) and “secondary” qualities (such as heat and green). In principle, what Williams calls an “absolute” description (or “conception”) of the world can be given using terms for the primary qualities alone.6 And only such a description describes the world as it is independent of our experience, as opposed to describing how it is in relation to one or another tribe or species of observers.
In Descartes, Williams introduces these ideas by means of a thought experiment. We are invited to consider how we would describe the world before there were observers, or, alternatively, to imagine the world as it might be if there were not any observers.7 In a colloquial description, we might, of course, include secondary qualities in our account, speaking of green grass or a hot day, but we are aware, or can easily be brought to be aware (Williams claims), that in so doing we are simply describing how the grass would have appeared to observers with our sort of visual apparatus if they had been present, or how the day would have felt to observers with our sort of body. This information is not really necessary, nor is it appropriate if we want to describe the world without observers as it is independent of our experience. Thus we come to see that this world, the world without observers, can be described (as it is independent of our experience) in terms of primary qualities alone.
Williams’ motive for introducing the absolute conception in this way is the following, I believe: the world we live in, the world with observers, evolved out of a world without observers. And the laws that govern our world are just the laws that governed the world without observers. So a description of our world using just the primary qualities that we need to describe the world without observers must be possible. The seductive appeal of this thought is undeniable. But is the thought really right?
Consider the following: it is true that with the evolution of animals and human beings and the development of human (and animal) societies no new physical laws came into existence. Very few people any longer suppose that living things violate any laws of physics (as some thinkers supposed as late as the nineteenth century), or that human beings have immaterial souls which cause them to move in ways that violate the laws governing the conservation of momentum,8 as Descartes’ model required. Physics can, in principle, predict the probability with which a human body will follow any given trajectory, given appropriate initial data, as it can predict the probability with which anything else will follow any given trajectory. But such predictions are of phenomena which are described in the language of physics, not the language of biology, psychology, or economics. With the appearance of living things and societies new laws do appear; not laws which contradict the laws of physics, but laws which apply to things under descriptions which are not available in physics. The laws of economics are a good example. “Supply” and “demand” cannot even be defined in the language of physics; so the question of “explaining the relation between supply and demand” is not a meaningful question to address to a physicist.
Williams is perfectly well aware of this consideration, and his reply involves an extended use of a particular metaphor—a metaphor which appears repeatedly in both of the books I alluded to. This is the metaphor of perspective. In optics, it is clear what we would mean by an “objective description” of a scene: namely, a description of the sizes and shapes and locations of all the bodies in the scene and of the sources of light. Using this objective description, and the laws of optics, which we may think of as a part of the objective description, we can predict how the scene will look “in perspective” from any particular given point of view, such as why Ermintrude looks tiny to Elmer when Elmer is looking down at her from the top of a hill, and why Elmer looks tiny to Ermintrude at the same moment. In this sense, the objective description explains all the “local perspectives”.
Williams’ leading idea is that the descriptions we give using the vocabularies of the special sciences, using terms for secondary qualities, using ordinary-language intentional terms,9 and so forth, can in a strictly analogous way be explained using the “absolute conception”; using, that is to say, not present-day physics, but an envisaged perfected analogue of present-day physics.10
The metaphor is carried even further. In the optical case, the objective description is one that the various observers can agree on (if they make the appropriate measurements, confirm the laws of optics, and so on), even if their perspectival descriptions appear to disagree. In a strictly analogous way, Williams believes that any conceivable species of intelligent beings (if they frame hypotheses properly, perform the appropriate experiments) can “converge” toward agreement on the laws of the ideal physics, in the fashion first envisaged by C. S. Peirce.11 (This does not mean that they can ever be sure that they have converged on the absolute conception, however; this is not required for us to be able to employ the absolute conception, Williams assures us.)12 Williams allows that the optical case is disanalogous in only one way; he believes that humans can, when we get sufficiently close to it, use the absolute conception to explain the possibility of our local conceptions (including our grasp of the absolute conception itself), but, puzzlingly, he does not insist that all the other species of observers must be able to do this as well.13 The reasons for this breakdown of the analogy will be be discussed later.
The Entanglement of Fact and Value
It must be emphasized that, in Williams’ view, to deny that something is part of the “absolute conception of the world” is not necessarily to deny that it is true or that it can be knowledge. It is true that the grass is green, and we know that the grass is green; but this, although genuine knowledge, is not knowledge of how things are “independent of perspective”—rather, it is knowledge of how things are in our local perspective. And the same goes for ethical statements. Williams does not deny that ethical utterances have a truth value, nor does he deny that there can be ethical knowledge. (In the jargon of contemporary philosophy, Williams is not a “non-cognitivist”.) I am not entirely sure of the reasons for Williams’ rejection of non-cognitivism, but one reason is surely an increased appreciation of what might be called the entanglement of fact and value.14 That entanglement was a constant theme in the writings of John Dewey, whose views I will turn to later. But this aspect of pragmatism was neglected in Anglo-American philosophy after Dewey's death, in spite of Morton White's valiant effort to keep it alive in Reunion in Philosophy, and it was, perhaps, Iris Murdoch who reopened the theme in a very different way.
Three essays by Murdoch, published together as The Sovereignty of “Good”, contain a large number of valuable insights and remarks: two have proved especially influential. Murdoch was one of the first English-speaking philosophers15 to emphasize that languages have two very different sorts of ethical concepts: abstract ethical concepts (Williams calls them “thin” ethical concepts), such as good and right, and more descriptive, less abstract concepts (Williams calls them “thick” ethical concepts), such as cruel, pert, inconsiderate, chaste. Murdoch (and later, and in a more spelled-out way, John McDowell)16 argued that there is no way of saying what the “descriptive component” of the meaning of a word like “cruel” or “inconsiderate” is without using a word of the same kind; as McDowell put the argument, a word has to be connected to a certain set of “evaluative interests” in order to function the way such a thick ethical word functions, and the speaker has to be aware of those interests and be able to identify imaginatively with them in order to apply the word to novel cases or circumstances in the way a sophisticated speaker of the language would.
For example, someone who has studied how the word “cruel” is used without performing such an act of imaginative identification could predict that the word would be used in certain obvious cases, for instance, torture. But such a person would be baffled by the fact that some cases which seemed (from the same external point of view) to be cases of “kindness” would be described by us as “subtle forms of cruelty”, and by the fact that some cases of what he or she would describe as cruelty would be described by us as “not cruel at all under the circumstances”. The attempt of non-cognitivists to split words like “cruel” into a “descriptive meaning component” and a “prescriptive meaning component” founders on the impossibility of saying what the “descriptive meaning” is without using the word “cruel” itself, or a synonym.
Murdoch also emphasized that when we are actually confronted with situations requiring ethical evaluation, whether or not they also require some action on our part, the sorts of descriptions that we need—descriptions of the motives and character of human beings, above all—are in the language of a “sensitive novelist”, not in scientistic or bureaucratic jargon. When a situation or a person or a motive is appropriately described, the decision as to whether something is good or bad or right or wrong frequently follows automatically. For example, our evaluation of a person's moral stature may critically depend on whether we describe the person as “impertinent” or “unstuffy”. Our life-world, Murdoch is telling us, does not factor neatly into “facts” and “values”; we live in a messy human world in which seeing reality with all its nuances—seeing it as George Eliot, or Flaubert, or Henry James, or Murdoch herself, can to some extent teach us to see it—and making appropriate “value judgments” are simply not separable abilities.
Williams not only considers these views correct, or likely to be correct, but in fact elaborates on them in an interesting way. Under the influence of quasi-social-scientific, quasi-philosophical talk about “facts” and “value judgements” many people simply assume that if we praise somebody we must, implicitly at least, avail ourselves of some “thin” (highly abstract) ethical notion such as good/bad or right/wrong. Thus, if we learn that, in a traditional society, Elder Jones criticized Matilda by saying she was not chaste, we are likely to analyze what Elder Jones was saying as follows:
- Matilda is not [insert a “value-neutral” word with the same descriptive meaning as “chaste”] (factual claim).
- It is bad not to be [insert the same “value-neutral” term here] (value judgment).
- Matilda is bad (unstated value conclusion).
According to Williams, however, it is quite unwarranted to attribute our own abstract ethical notions in this way to every society on the face of the earth.17 Williams performs the thought experiment of imagining a “hypertraditional society” which lacks our thin ethical concepts altogether, but which still possesses thick ethical concepts18 (imagine Elder Jones and Matilda live in such a hypertraditional society). Williams believes that a thick ethical concept like “chaste” can function as both a description and a value judgment, and it is a fallacy of division to suppose that the whole speech act must be divisible into a descriptive claim (which could, in principle, be expressed in value-neutral language) and a value judgment (which could, in principle, be expressed using thin ethical concepts).19
Murdoch's second claim, that we cannot in fact do without our thick ethical language and the point of view that goes with it, is one that Williams would heartily endorse. (He would rephrase it by saying that we cannot do without our “local perspective”.)
Relativism and the Fact-Value Dichotomy
According to Bernard Williams, a properly worked-out “non-objectivism”, by which he means a position which rejects the characteristic thesis of relativism (that propositions are “true for” or “false for” one or another culture, rather than simply true or false) while finding a “truth in relativism”, can do justice to the way in which fact and value can be inseparable—do justice to the way some statements which are both descriptive and true (“Caligula was a mad tyrant”) can also be value judgments. It can do this by exploiting the distinction between truth and absoluteness.
Williams nowhere explains what he understands truth to be, but he obviously intends it to include right-assertability in the local language game, so that if the practices and shared values of a culture determine an established use for the word “chaste” (a use which is sufficiently definite to permit speakers to come to agreement on someone's chastity or lack of chastity), then it can be simply true that a person in the culture is chaste. Of course, if we do not belong to the culture in question and do not share the relevant evaluative interests, then we will not describe the person in question as chaste, even if we know that that is a correct thing to say in that culture; we will be “disbarred” from using the word, as Williams puts it. As he also puts it (with deliberate paradox), that So-and-so is chaste is possible knowledge for someone in the culture, but not possible knowledge for us. Note that Williams does not say that “So-and-so is chaste” is true for Elder Jones’ and Matilda's culture but not true for, say, the culture of some Berkeley Free Love Movement. Williams argues that that sort of relativism quickly leads to absurdity (“I know where you're coming from, but, you know, relativism just isn't true for me”).20
Nevertheless, whether it is possible knowledge for us or not, on Williams’ view, “Matilda is chaste” can be true in the very same sense in which “snow is white” is true, while still being an ethical utterance. But there is an insight in non-cognitivism, Williams claims, even if non-cognitivism was mistaken in what it took to be its essential thesis, that ethical sentences are not capable of truth (or, alternatively, that an ethical sentence has a distinct “value component”, and this “value component” is not capable of truth). But what was the insight that the factvalue distinction tried to capture?
According to Williams, there are truths and truths. If I say that grass is green, for example, I certainly speak the truth; but I do not speak what he calls the absolute truth. I do not describe the world as it is “anyway”, independent of any and every “perspective”. As I have explained, Williams thinks that the concept “green” (and possibly the concept “grass” as well) are not concepts that finished science would use to describe the properties that things have apart from any “local perspective”. Martians or Alpha Centaurians, for example, might not have the sorts of eyes we have. They would not recognize any such property as “green”, and “grass” may be too unscientific a classification to appear in their finished science. Only concepts that would appear in the (final) description of the world that any species of determined natural researchers is destined to converge on belong to the “absolute conception of the world”. The philosophically important point here is that while value judgments containing thick ethical concepts can be true, they cannot be absolute. The world, as it is “independent of our experience”, is cold. Values (like colors) are projected onto the world, not discovered in it.
A second important point is that, on Williams’ view, values are even worse off than colors in this respect, for the discovery that green is a secondary quality has not undermined our ability to use the word. Although we no longer think that colors are non-dispositional properties of external things, this in no way affects the utility of color classification. But the realization that value attributes, even “thick” ones (“chaste”, “cruel”, “pert”), are projections has a tendency to cause us to lose our ability to use those terms. If we become reflective to too great a degree, if we absorb too much of the absolute conception of the world, we will no longer be able to employ our ethical concepts. The realization that ethical concepts are projections places us in a ticklish position: we cannot stop being reflective, but we cannot afford to be too reflective. We are in an unstable equilibrium.
The reason for this difference between ordinary secondary qualities like green and thick ethical attributes like chastity, according to Williams, is that the interests which color classification subserves are fixed by our biology (according to Williams) and thus are universal among human beings, whereas the interests that thick ethical concepts subserve are the interests of one human community (one “social world”) or another. But the interests which define one social world may be in conflict with the interests which define a different social world. Realizing that our ethical descriptions are in this way parochial (however “true” they may also be) can unsettle us, weaken our confidence in our own ethical language.
Williams believes that coming to realize just how far ethical description misses describing the world as it is “absolutely” must affect our first-order ethical judgments. According to him, moral praise or condemnation of another way of life loses all point when that other way of life is too distant from ours (too distant in the sense that neither way of life is a live option for the other). It makes no sense to try to evaluate the way of life of a medieval Samurai, or of a bronze age society. To ask whether their ways of life were right or their judgments true is (or should be) impossible for us; “the question of appraisal does not genuinely arise”,21 once we understand the non-absoluteness of ethical discourse. That the question lapses is “the truth in relativism”. (Accepting this view is what Williams calls “taking a relativist view” of such conflicts; note that this is different from accepting the kind of relativism that I discussed earlier.)
This is undoubtedly an ambitious and interesting set of views. Let us now examine them critically, beginning with the notion of “absoluteness”.
What Is “Absoluteness”?
Earlier I quoted two representative passages from Ethics and the Limits of Philosphy in which Williams explains what he means by the “absolute conception”, as well as briefly summarizing his explanations in Descartes. In all these explanations, three features recur, with different emphasis in different places:
- The primary qualities (which describe the world as it is “anyway”, independent of experience) do not depend on our constitution, whereas the secondary qualities “depend on psychological factors” (Descartes, p. 241).
- The absolute conception (which is equated with the complete description of the world in terms of primary qualities) explains the possibility of the local conceptions, and its own possibility. It (ideally) includes “a theory of knowledge and of error” (Descartes, p. 246).
- If science succeeds in converging on the absolute conception, that convergence is explained by the fact that that is “how things (anyway) are”.
The first feature, the “dependence on psychological factors” of the secondary qualities, seems obvious to Williams, although it is nowhere explained very clearly. Here is one discussion:
A familiar line is to treat “________is green” as in fact relational, though complexly and hypothetically so, equivalent roughly to “________ is of such a nature as to look green to standard observers in standard circumstances”. Under such an analysis, ascriptions of secondary qualities will in fact mention human relativities … it will be clear both why and how secondary qualities should be laid aside in giving the conception of the world as it is without observers. However this relational way of analysing secondary qualities … may well not be correct. For one thing, it leaves us with the discouraging task of explaining “_______ looks green” in some way which does not presuppose any prior understanding of “_______ is green”. How the relational pattern of analysis might possibly be replaced is part of a larger question, how the partial views and local experiences are themselves to be related to the world as conceived in independence of them. (Descartes, p. 244)
Unfortunately, far from giving us a reason for agreeing that secondary qualities should be “laid aside”, this passage rejects one of the common reasons for thinking so. Williams also tells us that “the traditional arguments bring out the ways in which the secondary qualities depend on psychological factors” (p. 241), by explaining why things should seem one color to one person and a different color to another. But we have, in fact, perfectly good ways of deciding what the color of, say, a surface is (for example, by comparing it with color samples); and if a red surface looks gray to a color-blind person (or “monochrome” to a species without color vision), that does not show that the color is relative to the viewer, but only that some viewers make mistakes (and some species lack color vision). Indeed, the tendency to confuse a secondary quality with the sensation of the secondary quality appears throughout these pages (237ff). Thus, Williams quotes with approval the traditional argument that heat cannot be physical because it shades insensibly into pain (p. 237), an argument which transparently conflates heat (temperature) and the sensation of heat.
The example of heat is interesting because its scientific counterpart, temperature,22 has been identified by physics with an objective property of things, albeit a highly disjunctive one, one which figures in the most fundamental physical theories (for instance, in quantum field theory, the notion of the temperature of the field is an important one). Perhaps Williams’ point is that an ordinary-language predicate like “hot” presupposes a standard which is based partly on our physiology and partly on contextual interests, in short on what Democritus called “convention”. This may be granted, and it may be granted that, for that reason, a precise description which abstracts from that physiology and those interests would replace predicates like “hot”, “lukewarm”, and “cold” with numerically specified temperatures; but that does not mean that “heat” (temperature) is not a feature of the objective world. Moreover, in the case of color, Williams’ picture is clearly that there is no such thing in the world (Williams speaks of “such large-scale mistakes as that the world is itself colored”, Descartes, p. 249). Does Williams also think that it is a “large-scale mistake” to think that some parts of the world are hotter than other parts?
Perhaps Williams thinks that the notion of heat in our common language (the language that Wilfrid Sellars referred to as “the manifest image of the world”) just doesn't refer to the same thing as the scientific notion of temperature. But this “two worlds” view is hardly forced on us by science itself; why shouldn't we say that science has discovered what the layman is talking about, discovered what being hotter and colder is, rather than saying that science has discovered that there are no hot or cold bodies in the world?
We would, indeed, have a good reason for saying that the layman is wrong about the world if we were to attribute to the layman the idea that his sensation of heat (or “idea” of heat, in seventeenth-century philosophical parlance) “resembles” the heat in the world; but this argument would equally support the idea that there is no distance in the world: the layman's sensation (or “idea”) of distance hardly “resembles” the distance talked about by the relativistic physicist or the supergravitation theorist. Thomas Reid already objected to attributing to laymen this confusion between heat and the sensation of heat, and Williams himself criticizes some seventeenth-century thinkers for arguing in this way (pp. 240–241).
Can we say, at least, that the sensation of heat cannot be identified with anything in an objective physical description of the world? To do so would require arguing that sensations cannot be identical with brain processes (which are part of the absolute conception of the world), and this is not a task which Williams anywhere undertakes.
Finally, Williams might fall back on the claim that the predicates “hot”, “green”, and so on are at least not synonymous with any terms in the absolute conception (even if they share reference with them); but this would hardly be a very interesting claim, even if we assume that the notion of synonymy is sufficiently clear. Donald Davidson has suggested in a famous essay that a very similar claim is not sufficiently clear: “Suppose that in my office of Minister of Scientific Language I want the new man to stop using words that refer, say, to feelings, thoughts, and intentions and to talk instead of the physiological states and happenings that are assumed to be more or less identical with the mental riff and raff. How do I tell if my advice has been heeded if the new man speaks a new language? For all I know, the shiny new phrases, although stolen from the old language in which they refer to physiological stirrings, may in his mouth play the role of the messy old mental concepts” (Inquiries into Truth and Interpretation, p. 188).
Nor is the situation much better if we consider color rather than heat. Although the philosopher Larry Hardin23 has argued on the basis of contemporary theories that color is “subjective”, I believe that most researchers hold that color is an objective property, at least in the case of surfaces. For example, the neurobiologist Jerome Lettvin has pointed out24 that the eye corrects for enormous differences in the chromatic composition of the light shining on objects in the course of the day, and, in doing so, it functions as a “computer” (he has analyzed the “computation” it carries out). It turns out that what the eye provides is a very good estimate (under a wide range of conditions) of certain objective potentialities (reflectancies) of the surfaces in question. Color vision is not a mere reaction of our physiology, uncorrelated to anything objective in the surface scanned.
One argument used by Hardin for regarding color as subjective is that the physical property of the surface is highly disjunctive; many different combinations of reflectancies will produce, say, a green surface. But, as Jonathan Westphal points out in a brilliant book on this topic,25 the disjunction admits of a fairly simple approximate characterization: a surface is green just in case it refuses to reflect a significant percentage of red light relative to light of the other colors, including green. The characterization certainly leaves the boundaries fuzzy, but it explains why many different combinations of reflectancies will result in a green surface. Moreover, the argument from disjunctiveness is unpersuasive; temperature is certainly disjunctively defined in present-day physics, but no physicist would regard temperature as “not in the world”. Indeed, since any non-disjunctive predicate can be rendered disjunctive (and vice versa) by changing which terms we take as primitive,26 is it clear that regarding disjunctiveness as a proof that a predicate is not really an objective “universal”, as some philosophers do, is not importing a “local perspective”, the perspective of a familiar set of linguistic habits, into questions of metaphysics?
Another argument used by Hardin is that judgments as to what is a standard green (as opposed to, say, a yellowish green or a bluish green) are not intersubjectively stable. This certainly does show that, unless we adopt a convention, such as a color chart which makes such determinations, there is no fact of the matter as to the “center” of the green part of the spectrum; but it hardly follows that there are no clear cases of green or no clear cases of non-green.
These questions are obviously too hard to settle now. But it should be clear that they are extremely complicated, and the view that green is a perfectly good property of things, one which is relational in the sense of involving the relations of the surface to light, but not relational in the sense of involving the relations of the surface to people, is still alive and well. Of course, the boundaries of “green” need to be suitably legislated if we are to use it in a precise description of the world without observers; but, as Wheeler et al. point out,27 “one second” (substitute your favorite unit of time) is not perfectly precisely defined, and indeed never will be. Williams would presumably not regard the fact that the time words of ordinary language have to be made more precise if they are to be used in science as showing that there is a “large-scale mistake” in supposing that events take time; by parity of reasoning, I do not see that the fact that the color words of ordinary language have to be made more precise if they are to be used in science shows that it is a “largescale mistake” to suppose that the world is colored.
Westphal's explanation of the nature of green, which I have just summarized, identifies green with a dispositional property—the disposition to reflect very little red light relative to the proportion of light of other colors which is reflected. It has occurred to me that Williams might now argue that talk of dispositions is also too “perspectival” to figure in the absolute conception. My own arguments in the previous chapters might be used to support such a claim. But again the issues quickly become murky. What I argued is that counterfactuals (and dispositional statements) presuppose a distinction between hypothetical situations which are relevant to the statement's truth or falsity and possible worlds which are not, and such a distinction cannot be drawn in non-normative terms. But to conclude that dispositions are, therefore, “perspectival” would be to assume as a premise just what Williams wants to derive as the conclusion of his discussion: that normative notions (such as “relevance”) cannot appear in the absolute conception of the world.
Williams argues that “in understanding, even sketchily, at a general and reflective level, why things appear variously colored to various observers, we shall find that we have left behind any idea that, in some way which transcends those facts, they ‘really’ have one color rather than another”. Lettvin's work, and the work that Westphal discusses, may refute this claim.
Williams might reply that even if the color of a surface did turn out to be an objective reflectance property, this still would not effect the contrast he wants to draw between color properties and value properties. But it does not require Williams’ elaborate metaphysical story to point out that valuation does not arise from an organ like the eye. (John Dewey, to whose philosophy I turn in the final chapter, would urge that we see valuation as arising from the criticism of modes of problem solving.)28
The second feature of Williams’ argument—that the absolute conception (which is equated with the complete description of the world in terms of primary qualities) explains the possibility of the local conceptions and its own possibility by (ideally) including “a theory of knowledge and of error”—is qualified virtually out of existence in Williams’ text almost as soon as it is advanced. For this reason, this feature cannot be the defining feature of what Williams calls “absoluteness”, although it is absolutely essential to the metaphor of “perspectives”.
J. L. Austin once said of the ways of philosophers, “There's the bit where he says it and the bit where he takes it back.” Here's the bit where Williams says it:
This conception of the world must make it possible to explain how it can exist. This conception is not something transcendental, but is an historical product of consciousness in the world, and it must at least yield a comprehension of men and other rational creatures as capable of achieving that conception. (Descartes, p. 246)
And here's the bit where he takes it back:
The demand [that we show the possibility of explanations of the place in the world of “cultural phenomena such as the local non-absolute conceptions of the world, and of the absolute conception itself, including in that the possibility of physical science”] may be much harder to meet, or to evade. The requirement was that we should be able to overcome relativism in our view of reality through having a view of the world (or at least the coherent conception of such a view) which contains a theory of error; which can explain the existence of rival views, and of itself. But this conception involves a dimension not just of physical explanation, but of social explanation as well. We have to explain the emergence of physical science as something which is indeed knowledge. This entails, if we are to sustain the realist outlook which is essential to the idea of the absolute conception, that physical theory and the interpretation of nature not suffer from the same indeterminacy that may affect translation and the interpretation of the mental. We have to explain, further, how psychological, social and other theories, and also less theoretical views of the world, can be related to the world as we understand it in terms of physical theory. In these philosophical and social scientific tasks, we are not only explaining, but ourselves giving examples of theories which … deal with just the kind of subject matter which may be subject to radical indeterminacy of interpretation.
In face of such considerations, the most ambitious ideas that have been entertained of the absolute conception must fail: the idea, for example, of a convergent, self-vindicating unified science of man and nature. How much less than this positivist fantasy [sic] will do? What is the minimum? Perhaps just this: that we should be able to make sense of how natural science can be absolute knowledge of reality, and of why we cannot even agree how much else is absolute knowledge of reality. (Descartes, pp. 301–302)
The problem Williams faces is quite simple: the only sense in which the absolute conception, which we are to think of on the model of “physics and natural science”, can “explain the possibility of the local conceptions, and its own possibility” is that it can predict that certain marks and noises will occur. But this is not to explain how those marks and noises are conceptions or to explain how they describe anything, even “perspectivally”.
The problem Williams faces is similar to the problem Wilfrid Sellars, whose views in many ways anticipated Williams’ present views, notoriously faced. Like Williams, Sellars did not wish to regard any semantic relations connecting words and sentences with things and states of affairs as part of the ideal scientific conceptual scheme. But, again like Williams, Sellars wished to say that our ordinary-language schemes somehow reflect (are perspectives on) the objective world. So Sellars introduced a relation he called “picturing”, to be distinguished from truth and reference, and said that our ordinary-language schemes (our local perspectives) “picture” the world more or less adequately (some of our schemes more adequately and some less), even if they do not bear an objective semantic relation to it. The difficulty, which has split Sellars’ students and followers into two camps,29 is that Sellars owes us at least a sketch of how picturing can be defined in the ideal scientific scheme. Where Williams differs from Sellars is apparently in not recognizing the problem. Yet without giving sense to the notion that the marks and noises which constitute our various “perspectives”, and also the marks and noises which constitute the absolute conception of the world, somehow picture the world, Williams cannot sustain the idea that these marks and noises are, respectively, our local perspectives and an objective description. To be a “perspective”, marks and noises must picture something; to be an objective description, marks and noises must describe.
I was, perhaps, unfair to accuse Williams of not seeing the problem at all. It is, very likely, because he faces this problem that he replaces talk of the absolute conception's “explaining the possibility” of the local conceptions with talk of our being able to explain the local conceptions with its aid. Thus, in one passage,30 Williams says of the absolute conception, “it will also help to explain to us, though not necessarily to those alien investigators, such things as our capacity to grasp that conception” (emphasis added). In other words, the “theory of error” will not be provided by the absolute conception, but will be part of a “local perspective”, albeit one that is informed by the absolute conception (Williams repeatedly mentions neurobiology). But such a theory will be a theory whose central notion, the notion of picturing or describing the world (perspectivally or non-perspectivally) does not belong to the absolute conception. Is Williams saying that it is just our local perspective that there is an absolute conception? Even Richard Rorty might agree with that.
The third feature of the absolute conception is that the convergence of all species of researchers (ourselves and alien species) on the absolute conception of the world will, if it comes about, be explained by the fact that that is “how things (anyway) are”, or how things are independent of experience. This presupposes both the notion of convergence, that is to say, agreement in belief, and the notion that, given a belief, one can say that what the belief says (not the marks and noises used to express the belief) “is how things are”—that is, the notion that a belief describes (or at least “pictures”) something outside itself. But Williams takes seriously the Quinian idea of the indeterminacy of reference: “If the various sorts of considerations which have been summarily sketched here are correct, then we have to give up not just dualism but the belief in the determinacy of the mental. These considerations converge on the conclusion that there are no fully determinate contents of the world which are its psychological contents” (Descartes, p. 300).
Williams might, of course, have chosen to regard the notion of the “content” of a belief as one which will be available in the absolute conception of the world. But in that case he would have had either to show that that notion can be reduced to the notions of physics, which he is persuaded by Quine and Davidson is not a promising line to take, or to follow Fodor in trying to reduce it to notions from the special sciences and/or the ordinary-language “thick” notion of causation (which, I believe, he would regard as too “local”), or to take it as primitive. If he had followed the last course, he would have had to admit that we do not have a sketch of the “absolute conception of the world” in present-day physics; that, indeed, the notion of an absolute conception of the world is either just the ordinary notion of an objective description or a we-know-not-what. In either case, his insistence that normative notions cannot appear in the absolute conception becomes groundless; for that insistence was based from beginning to end on the idea that only primary qualities can appear in the absolute conception.
Indeed, it is precisely because Williams refuses to admit normative notions into the absolute conception that the indeterminacy of translation appears to be such a threat. For an indeterminacy of translation that affects only the interpretation of subsentential parts, however perplexing, and whatever we may decide about it, will not prevent us from seeing our sentences as related to an objective world as long as sentences as wholes have determinate conditions of warranted assertability. It is striking that when Williams describes the indeterminacy he is talking about, he describes it by saying that a sentence may have two correct translations which are not themselves “equivalent”.31
It is, of course, true that this can happen in cases of ordinary vagueness or ambiguity. If S itself is ambiguous, and A and B are the alternative ways of disambiguating it, then it is not surprising that A and B may be equally plausible but non-equivalent translations, and if the speaker was unclear as to what he or she meant to convey, A and B may be equally correct. But no one would call this an example of the “indeterminacy of translation” in Quine's sense. The kind of indeterminacy Williams refers to is a serious threat only if there is no fact of the matter as to what the conditions of warranted assertibility of even an unambiguous sentence are.32 But to believe this is to believe that there is no fact of the matter about the normative—and not just about that small part of the normative which we call the ethical.
If Williams were to drop his endorsement of the indeterminacy thesis, and to admit that we can make no sense of an adequate conception of the world (whether we call it “absolute” or not) that does not make room for normative notions, I would regard the position as greatly improved. But I would still urge him to recognize that there are many true descriptions of the world in many different vocabularies, without trying to privilege one of those descriptions as the “absolute” one. This idea will play a central role when I come to examine Nelson Goodman's position.
The “Truth in Relativism”
Williams’ discovery of a “truth in relativism” does not appear to me to be any more coherent than the “absolute conception of the world”. First, let me point out something that I have not had occasion to remark up to now: Williams is extremely cavalier about the notion of truth. Sometimes the truth is what is “tracked” by the practices of a linguistic community; but at other times Williams treats truth simply “disquotationally”, that is, he applies the principle that if we assert S then we must say that S is true, without invoking any considerations of “tracking”. This is, for example, the way Williams treats the use of the word “true” in what he calls “real confrontations”. It is striking that the same cavalier attitude is characteristic of a philosopher Williams regards as an opponent—Richard Rorty. Like Williams, Rorty sometimes treats truth as right assertibility in a local language game, and sometimes treats it disquotationally, or as a “compliment” we pay to opinions we like. Moreover, again like Williams, Rorty is fond of insisting that we are related to the world causally but not semantically.33 It is true that Rorty rejects the idea of an “absolute conception of the world”—but if it were made clear to Rorty that the conception in question is “absolute” only from our local perspective, would he still feel it necessary to reject it? Could it be that, even if they do not recognize it, Rorty and Williams are in complete agreement in metaphysics? For the rest, I shall content myself with two observations.
- The distinction between “real” and “notional” confrontations is unfortunately drawn, even for Williams’ own purposes. As it stands, the confrontation between the Jews and the Nazis would not count as a real confrontation by Williams’ definition—because “going over to” the point of view of the other was not a “real option” for either the Jews or the Nazis. Yet Williams would not deny that it is appropriate to use the language of good and bad, right and wrong, in such a conflict. How Williams’ distinction might be repaired to remove this inconsistency is something I shall not speculate about.
- Williams says that in a notional conflict “the question of truth does not genuinely arise”. Yet he also says that the members of the other community, the one from which I am relatively distant, have ethical knowledge,34 and that their beliefs (when they deploy their concepts carefully) are true.35 This is a blatant contradiction, and I do not see how to remove it.36 For example, it will not do to to reply as follows:37 when I want to talk about one of the concepts of the distant community, and to reject the values embedded in its thick concepts, I can say “I agree that doing such-and-such is unchaste, but I don't regard chastity as a virtue” (that is, conflict is registered using the “thin” concepts such as good, right or virtue; this would still allow us to say that the distant community's thick judgements are true). First of all, there are many things that traditional communities regard as unchaste (such as, in traditional Muslim communities, women appearing in public without veils, or, in ultra-Orthodox Jewish communities, a woman being alone with a man in a room with a closed door, if the man is not her husband or father)38 which we do not regard as unchaste. We would not say “we agree that her being alone in the office with her employer is unchaste but we don't think chastity is a virtue”, for even those of us who do think chastity is a virtue don't regard that as unchaste. Second, this reply would contradict Williams’ explicit statement39 that only in the case of real conflict can the language of appraisal (including “true” and “false”) be used. It is a flat-out contradiction to say that the judgment that the act is unchaste is true (although we are “disbarred” from making it) and also to say that we can't say that the judgment is either true or false.
Metaphysics and Entanglement
What led Williams to defend this complicated metaphysical theory was the desire to assert a “truth in relativism” while resisting relativism in science. But in the process of building up this intricate construction with its two kinds of truth (ordinary and “absolute”), its perspectivalism about secondary qualities and ethics (and, oddly, also about the intentional) and its anti-perspectivalism about physics, he often ignores the entanglement of the factual and the ethical—although he himself stresses that entanglement at other points in his discussion. Consider, for example, the question as to whether we can condemn the Aztec way of life, or, more specifically, the human sacrifice that the Aztecs engaged in. On Williams’ view, the Aztec belief that there were supernatural beings who would be angry with the Aztecs if they did not perform the sacrifices was, as a matter of scientific fact, wrong. This belief we can evaluate. It is simply false; and the absolute conception of the world, to the extent we can now approximate it, tells us that it is false. But we cannot say that the Aztec way of life was wrong. Yet the feature of the Aztecs’ way of life that troubles us (the human sacrifice) and their belief about the world that conflicts with science were interdependent. If we can say that the Aztec belief about the gods was false, why can we not say that the practice to which it led was wrong (although, to be sure, understandable given the false factual belief)? If we are not allowed to call the practice wrong, why are we allowed to call the belief false? The so-called “absolute” and the ethical are just as entangled as the “factual” and the ethical.
For a very different sort of example, consider the admiration we sometimes feel for the Amish (traditional Mennonite) way of life. Even atheists sometimes admire the community solidarity, the helpfulness, and the simplicity of the Amish way. If sophisticated atheists who felt this way were asked why they admire the Amish, they might reply something like this: “I am not necessarily saying we should give up our individualism altogether. But the kind of individualism and competitiveness which has brought so much scientific and economic progress also brings with it egotism, arrogance, selfishness, and down-right cruelty. Even if the Amish way of life rests on what I regard as false beliefs, it does show some of the possibilities of a less competitive, less individualistic form of life; and perhaps we can learn about these possibilities from the Amish without adopting their religion.” Now, Williams does not deny that we can say things like this—that we can learn from cultures to which we stand in the relation he calls “the relativity of distance”, cultures which are not “real options” for us. But how does this differ from saying “some of the Amish beliefs are false, but other of their beliefs may be true”? Williams’ examples load the dice in favor of relativism by taking science to consist of individual judgments which may be called true or false, while taking “cultures” to offer only “take it as a whole or reject it as a whole” options.40
The problem with the whole enterprise lies right here: Williams wants to acknowledge the entanglement of fact and value and hold on to the “absolute” character of (ideal) scientific knowledge at the same time. But there is no way to do this. It cannot be the case that scientific knowledge (future fundamental physics) is absolute and nothing else is; for fundamental physics cannot explain the possibility of referring to or stating anything, including fundamental physics itself. So, if everything that is not physics is “perspectival”, then the notion of the “absolute” is itself hopelessly perspectival. And the idea of a “relativism of distance” which applies to ethics but not to science also fails, because ethics and science are as entangled as ethics and “fact”. What we have in Ethics and the Limits of Philosophy is, in fact, not a serious argument for ethical “non-objectivism”, but rather the expression of a mood. In Ethics and the Limits of Philosophy, we are told that ethical “non-objectivism” is the “contemporary” point of view, and what is being offered is a sophisticated reflection on the consequences of this presupposition. The sophistication is undeniable, and many wonderful observations and arguments occur in the course of the reflection. But the presupposition itself does not stand up to any kind of examination—or at least, the ways in which Williams defends the presupposition crumble when one subjects them to careful examination.
Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard University Press, 1985): Descartes: The Project of Pure Enquiry (Harmondsworth, Middlesex: Penguin Books, 1978). See also Williams’ Moral Luck (Cambridge: Cambridge University Press, 1981), esp. chap. 11, “The Truth in Relativism”.
“Cognitivism” in ethical theory is the view that ethical judgments can be true and false.
Williams rejects the Kuhnian notion of incommensurability in science, but he finds it useful in ethics. See Ethics, pp. 157–158.
Ibid., chaps. 8 and 9.
Descartes, pp. 236–249. See also pp. 298–303.
Of course, one is also allowed to use terms defined with the aid of the primary qualities, as, for example, “momentum” is defined with the aid of “mass” and “velocity”, and “velocity” is defined in terms of “time” and “position”.
Descartes, pp. 243–246.
Descartes claimed that interaction between the body and the soul, which he placed in the pineal gland, would preserve the conservation of total momentum, but he was unaware that the laws of physics require also that momentum in each direction be conserved, and this must be violated if the interaction is to divert the body from the trajectory it would have followed anyway.
Under the influence of Quine and Davidson, Williams expresses scepticism as to whether intentional content can be part of the absolute description. “If the various sorts of considerations which have been summarily sketched here are correct, then we have to give up not just dualism but the belief in the determinacy of the mental. These considerations converge on the conclusion that there are no fully determinate contents of the world which are its psychological contents” (Descartes, p. 300).
Ibid., p. 247.
See “The Fixation of Belief,” in Charles Hartshorne and Paul Weiss, eds., Collected Papers of Charles Sanders Peirce, vol. 5, Pragmatism and Pragmaticism, pp. 223–247 (Cambridge, Mass.: Harvard University Press, 1965).
Descartes, pp. 302–303.
The absolute conception “will also help to explain to us, though not necessarily to those alien investigators, such things as our capacity to grasp that conception” (Ethics, p. 140; emphasis added).
I discuss this at more length in “Objectivity and the Science/Ethics Distinction,” reprinted in Realism with a Human Face (Cambridge, Mass.: Harvard University Press, 1990). I have adapted a few paragraphs from that earlier essay here.
The Spanish philosopher Ortega y Gassett had pointed this out in 1923. See “Introducción a la estimativa,” in his Obras Completas (Madrid: Revista de Occidente, 1958), vol. 6, pp. 317 and 320–321.
John McDowell, “Are Moral Requirements Hypothetical Imperatives,” Proceedings of the Aristotelian Society, suppl. vol. 52 (1978) and “Virtue and Reason,” Monist 62 (1979).
Ethics, pp. 145–146.
Ibid., p. 142.
I would agree with Williams that utterances in a hypertraditional society cannot be analyzed as conjunctions of value-neutral descriptive claims and claims involving the thin ethical concepts right and wrong, but I have criticized the claim that such utterances do not even imply any claims involving our thin ethical concepts in “Can Ethics Be Ahistorical? The French Revolution and the Holocaust,” in Culture and Modernity: The Authority of the Past, Proceedings of the Sixth East-West Philosophers’ Conference, edited by Len Goodman.
Witty saying credit: Alan Garfinkle.
Moral Luck. p. 141.
In some contexts, the counterpart of ordinary-language “heat” is quantity of heat rather than temperature; but the seventeenth-century discussion usually turns around comparisons of temperature.
See his Color for Philosophers (Indianapolis, Ind.: Hackett, 1986).
“XIV. Neurology,” in Quarterly Progress Report no. 87 (Research Laboratory of Electronics, M.I.T.: 1967). A similar view is presented by David Hilbert in Color and Color Perception: A Study in Anthropocentric Realism (Stanford: Center for the Study of Language and Information, 1990).
Color: Some Philosophical Problems from Wittgenstein. (Oxford: Blackwell, 1991).
For example, in Fact, Fiction, and Forecast (Cambridge, Mass.: Harvard University Press, 1983), Goodman shows that “grue” (green and examined prior to t or blue and not examined prior to t) is disjunctive if we take “green” and “blue” as primitive, but “green” is disjunctive in exactly the same way if we take “grue” and “bleen” as primitive.
Charles W. Misner, Kip S. Thorne, and John Archibald Wheeler, Gravitation (San Fransisco: W. H. Freeman, 1973), pp. 23–29.
See his Experience and Nature (LaSalle, Ind.: Open Court, 1926).
One camp—the “left-wing Sellarsians”—would abandon the notion of “picturing” altogether. Richard Rorty belongs to this camp. Right-wing Sellarsians would retain the notion.
Ethics, p. 140
Descartes, p. 299.
This is so even granting that the truth conditions of a sentence are not completely fixed by the conditions for its warranted assertability; for if translation is determinate at least up to conditions of warranted assertability, then ascriptions of content are very strongly constrained. My own view, argued in Reason, Truth, and History and in the last chapter of Representation and Reality is that truth conditions are fixed by conditions of warranted assertability together with conditions of the form “epistemic conditions A are better than epistemic conditions B for making judgments of warranted assertability about S”. Note that these latter conditions are also normative. Of course, a metaphysical realist might hold that (what we take to be) conditions of warranted assertability do not constrain truth conditions at all; but I do not believe that this can be a position Williams would find congenial.
See, for example, Contingency, Irony, and Solidarity (Cambridge: Cambridge University Press, 1989), pp. 3–22, and “Pragmatism, Davidson and Truth” in E. Lepore, ed., Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson (Oxford: Blackwell, 1986), esp. pp. 341–342.
See Ethics, p. 148, and the discussion leading up to it.
Ibid., p. 143.
In particular, Williams cannot mean that it “lapses” in only a pragmatic sense, since the contrast he draws with the scientific case would then make no sense. “When relativism is rejected in a given area, this does not mean that there are no notional confrontations. The confrontation between phlogiston theory and any contemporary theory of combustion is without doubt notional, and phlogiston theory is not now a real option; but on the nonrelativist view of such theories there is something to be said in appraisal of phlogiston theory, that it is false” (Ethics, pp. 161–162). This immediately follows his explanation of what he means by “taking a relativist view” toward ethical conflicts: that it is only in the case of real conflicts that the language of appraisal can be used. Since “false” (and, by implication, “true”) is “something to be said in appraisal”, Williams clearly continues to hold the view that the question of truth lapses in ethics (but not in science) when the conflict is notional.
One of the anonymous referees of this book for the Harvard University Press suggested that this is how Williams could respond to the “blatant contradiction”.
This is prohibited by the Talmud.
Ethics, p. 148.
Williams is quite deliberate about this. “In introducing this kind of relativism, I have mentioned ethical outlooks rather than particular practices, and it is to fairly large-scale systems or bodies of beliefs and attitudes that it has to be applied” (Ethics, p. 162).