DRAFT ONLY MAY 1994 PLEASE DO NOT RECOPY
penultimate text of: " 'World' is not a count noun", Nous 29 (1995), 139-157.
'World' Is Not a Count Noun
Bas C. van Fraassen
Princeton University
CONTENTS:
p.1 I. (Intuitive arguments about worlds)
p.3 II. (Whole, totality, composition)
p.6 III. (Scientific cosmology)
p.10 IV. (Ontology as 'of a piece with' science)
p.14 V. (Science and language develop in tandem)
p.16 VI. ("World" as schematic term; "isomorphism")
p.22 VII. (What if we meant what we said?)
'World' Is Not a Count Noun(1)
Bas C. van Fraassen
Princeton University
The word "world" has in fact many ordinary uses as a count noun; I shall discuss some of them below.(2) There is however also a distinctive philosophical use found in recent ontology (in the sense in which Quine reintroduced this term in analytic philosophy, for theories about what there is). As to this philosophical use, I shall argue that there is no reason to think that it refers to anything, if indeed it is intelligible at all.
The question addressed is not simply whether other possible worlds are real. If my argument is correct, then -- to put it in the material mode -- there is no reason to think that there is even such a thing as the (real, actual) world. My focus will be on that question: does the world exist? More important than its answer, however, are the crucial issues to which this question leads us: the conditions under which something is intelligible and the reasons we can have for thinking that it is so.
I
The word "world" has become common coin in philosophy, and occurs in arguments which we examine readily enough without doubting their credentials as arguments. I will list three, in increasing order of audacity. They are examples only, and perhaps more simplistic than any to be attributed to any actual philosopher, living or dead.
(1) We live in the world, as do also all animals and plants, and all people past, present, and to come. Similarly, all inanimate objects exist in the world and all events occur, all processes take place, in the world. Therefore the world is such that everything that is (except perhaps itself alone), is in it.
(2) We, you and I, are part of the terrestrial eco-system, which is in turn part of the solar system; and that in turn is part of the Milky Way galaxy. Indeed, it is possible to continue in this way, along the lines of scientific cosmology, so there is one thing, one system -- the thing we call the world -- of which everything (except perhaps itself alone) is part.
(3) You are acquainted with this world we live in, and so am I. Therefore we are acquainted with a world, that is, with at least one example of a thing properly called a "world." In consequence we can conceive of there being many things of that sort -- though of course, at most one of those is this (actual) world of our acquaintance and the others, if any, are other possible worlds.
The last argument concludes that there can be many (real) worlds of which this (actual) world is but one: I shall henceforth follow that terminological distinction between "actual" and "real" where appropriate.
Are these good arguments? Some parts have the grammatical form of logically valid arguments. Other parts do not -- so if they are valid, that must be because of the meanings of their non-logical terms. We can easily find examples of arguments which
(a) are definitely intelligible, (b) have the same grammatical structure as the above, (c) have true premises, and (d) have either a true or a false conclusion. For example, if someone thought that (2) is logically valid we could correct him or her by pointing to a similar argument about sets. Suppose we call one set a part of another set if it is included therein (all members of the former are members of the latter). Then set theory implies that any two sets are parts of some set, yet does not allow us to conclude that there is one set of which every other set is a part. On the other hand, in a heap of grain whose parts are reckoned to be just the grains plus the heap itself (and if you like the 'subheaps'as well) any two of its parts are part of some one thing, and there is one thing of which all parts are part. So argument (2) is not logically valid, but it may be valid nevertheless, at least as far as grammatical and logical analysis allows us to determine.
That is a modest conclusion: there are no purely logical or grammatical flaws in such sentences as "I shall keep this for you until you give it back to me" either. But it delimits our task. The same goes for all three: logic and grammar do not tell us that they are good arguments but they may be good (or bad) arguments as far as logic and grammar allow.
II
The philosophical uses of "world" evident in the above sample arguments are not new. Aristotle introduced the question of the plurality of worlds in De Caelo (276b-279a), and Locke speaks of the "great collective Idea of all Bodies whatsoever signified by the name World" (Human Understanding II, xxiv, 1). More sustained is Kant's treatment of this subject in his pre-Critical writings, beginning with Thoughts on the True Estimation of Living Forces. In his (Inaugural) Dissertation on the Form and Principles of the Sensible and Intelligible World, the first section deals with the notion of world in general, defined as "a whole which is not a part." Here Kant poses explicitly the question of the conditions requisite for the composition of an aggregate (multitudo) into a whole or totality (omnitudo). In both texts the idea of composition into a single world seems inspired by Newtonian universal gravitation: the criterion for composition is mutual interaction. Without the principle that mutual interaction is universal, bodies can combine into several real, separate worlds (Thoughts, sections 7, 8).(3)
This question was again raised in Peter van Inwagen's Material Beings(4) as the 'Special Composition Question.' Roughly phrased it is the question when one thing is [a] part of another thing (p. 21) and more precisely it can be phrased like Kant's question above (p. 22), but most precisely of all it has the schematic form (p. 30):
When it is true that there is something y such that the x's compose y?
The term "compose" is explicated in terms of "is [a] part of" as treated in mereology.(5)
Van Inwagen examines a number of answers to this question and proposes one of his own. The existence of the world, i.e. the entity which is composed of all (actual) things (bodies/events) whatsoever is by no means implied by most of these answers. David Lewis, who disagrees with van Inwagen's answer, holds to a view which explicitly implies that the world exists: the actual world is composed of all the entities spatio-temporally related to a given actual entity.(6)
One way to examine whether or not the world exists is there-fore to scrutinize those general philosophical answers to the Special Composition Question. Some such answers will actually entail the existence of the (or at least one) world. Others will at least allow for this, and leave it an open -- presumably factual -- question whether it is so. Still others entail that there is no such thing as the world. These different philosophical views may lay claim to quite disparate virtues.
Before we continue along this primrose path, let us by con-trast imagine a more impatient reaction, as follows. The question we confront, the one we have to face, is not whether some philoso-pher's theory has this or that virtue or implication. It is not the question of what is tenable, consistent, plausible, coherent, metaphysically or epistemologically satisfying. It is a question about the statement
The world exists
and is simply this question: is it true?
I mean this very seriously. We know how to think about technical issues in mereology, we can do the exercises. We can evaluate how and where philosophers run into trouble as they try to elaborate coherent positions on this, that or the other. Here, however, we face a putative statement of fact, and what we want to ask is: but is it true? Faced with this bare and unadorned question, I at least have no idea about how to go on.
This impatience, this insistence on facing a putative factual question, may strike us as naive or even philistine. Yet philosophers do tend to insist, when pressed, that pursuit of theoretical virtues is only instrumental, that they mean to find the truth. David Sanford (op. cit.) gave various reasons for thinking that mereology regiments language arbitrarily and even distorts our prior common understanding of "part." Accordingly he suggested that the Special Composition Question might be an "impossible question" (pp. 223-4). This suggestion can be understood in different ways; Sanford may well have meant that subtle, tacit shifts in idiolect have led philosophers into an idle word game. But van Inwagen's reaction reveals his commitment to a concern with the truth of the matter:
Well, perhaps so. In Material Beings (p. 68), I conceded that it might have no answer that anybody but God could know. I suppose that a question that had no answer that anybody but God could know could appropriately be described as an impossible question. But I don't see any reason to think that the [Special Composition Question] is in fact impossible.(7)
What God at least knows is under what conditions certain things compose another thing, and hence -- since he also knows whether the conditions obtain -- whether the world exists. This implies that these are intelligible questions of fact, that there is a fact of the matter. So all right, is it so? The trouble is that I have no idea of how to go on.
It may well seem that now, commenting on this imagined naive or even philistine impatient reaction, I too have slipped into objectionable naiveté. Once upon a time, analytic philosophers like to say, we were verificationists: we thought that for any factual statement of interest there is a procedure for determining whether it is true or false. But now we know that not even scientists are really in that enviable position. The best even they can do is to develop theories which are in accord with the evidence, and to choose the ones for which the evidence is most telling. The question whether anything exists has to be replaced by the question: does the status we accord to our best theories require its existence?. It is not that the former question is unintelligible or does not arise, it just does not arise for us directly: we can only sidle up to it via the latter question.
If this is so, we will be well advised to look to science for our answers. I will take this advice and move on from mereological armchair cosmology to scientific cosmology. Later on we will, however, have to consider whether the methodological contentions of the preceding paragraph do not leave the issue in still too naive a form.
III
There is indeed such a subject as scientific cosmology, there has been throughout the history of science. In a well-known intro-duction to the subject G. J. Whitrow wrote "Our direct knowledge of the universe is confined to a limited region of space and time. In order to obtain some idea of the universe as a whole we must extrapolate and construct a world-model which will reproduce satisfactorily the principal features of the observable region."(8) This captures, I think, the common understanding of what cosmology is about, and it presupposes of course that there is such a thing as the world or universe as a whole. Indeed, we might think of scientific cosmology as the theory of (physically) possible universes (worlds). But is this the only or even the best way to understand it?
To construct his System of the World, Newton added to the three laws of motion of his Principia the single law that there is just one force: attractive, between any two bodies, proportional directly to their mass and inversely to the square of the distance between them. We may therefore say, using Whitrow's term, that a Newtonian world-model is any system of bodies subject to exactly those four laws. Is the real world (exactly like) one of these Newtonian world-models? Even Newton would not have asserted this without qualification. It seemed plausible even then that, while the laws of motion may be correct, there are additional forces connecting the various bodies. The world-models themselves are mathematical structures. In constructing these world-models Newton was happy enough to work with simplifying assumptions, in order to save the phenomena of planetary motion. But of course, his text also points to a richer cosmology in which the world-models contain other forces as well, though all in accord with his dynamics.
Enter here, however, a fundamental divide in philosophy of science. I have taken the liberty of describing Newton's theory in the terms of the semantic approach: to present a theory is to present a family of models augmented with hypotheses about the extent to which the real things studied are adequately represented by these models. Philosophers willing to discuss science in these terms today differ nevertheless in what they see as the aim pursued in science, and consequently in what they think counts as scientific success. According to constructive empiricism, the aim is only to construct models in which the observable phenomena can be isomorphically embedded (empirical adequacy). Even given perfect success, then, not all elements of the models need have corresponding elements in reality. Rival positions going under the name of scientific realism often include the assertion that empirical adequacy is stronger than what counts as success in practice (similarity in some respects between real system and model being enough) but that if the model is to count as entirely successful then its elements purporting to represent unobservable structure must have corresponding elements in reality as well. The difference concerns in part how theory-acceptance in science is portrayed. In practice, there is of course qualified acceptance only, but this derives its meaning from what is understood as included in unqualified acceptance: belief in the empirical adequacy of the theory (constructive empiricism) or belief in its truth (scientific realism).
The different sorts of positions sketched here can be accompanied by varying understanding of physical cosmology. If a world-model need only save the phenomena then it need not as a whole correspond to anything at all. It's success is quite independent of whether all the phenomena are part of one thing, the world. For scientific realism the situation is a little more complex. There will always be aspects of models which everyone agrees are there for mathematical convenience only. Therefore when a scientific realist looks at cosmology, no reason of principle dictates that its models taken as whole structures must stand for the world as a whole. Hence, there is no reason of principle for the scientific realist to say that cosmology implies that there is such a thing as the (whole) world. Success could consist in every real physical system being adequately represented in the model, without there being a single physical system which is composed by all the others.
Recent philosophical debate has not shown scientific realists taking this option. To clarify that we should shift attention from Newton's System of the World to the space-time theories of relativ-istic cosmology. This subject is treated by Michael Friedman with explicit reference to the debate over scientific realism.(9) In relativistic cosmology, a world-model (to continue with Whitrow's terminology) is a space-time which is a four-dimensional differentiable manifold M with certain geometrical objects defined on M. These geometric objects are designated as absolute or dynamic objects, the latter depending on the distribution of matter-energy and the former independent thereof. They must satisfy the field equations of the theory while certain further equations, the equations of motion, specify the possible trajectories followed by material particles.
In such a cosmological model there is therefore one salient whole: the space-time itself. In Newton's case, the question whether the world exists is the question whether there is one Newtonian system of bodies of which all bodies are part (with major worries attaching only to the case of infinitely many bodies). In relativistic cosmology, mass and energy are quantities defined as space-time, and so the question seems to become whether space-time itself is real, a 'substance'. At least, that is what the question seems to become in the philosophical literature. In his recent World Enough and Space-Time, John Earman has examined various precise purported equivalents of this graphically phrased question.(10) The book ends on a distinctly uncomfortable note: while what he calls manifold substantivalism was in a much better position in classical physics than Leibniz or Mach appreciated, and perhaps even in a better position after the advent of General Relativity, it has recently been placed on the defensive, and Earman sees not much hope in any of its defenses. While Earman appears to broach these issues entirely from within a scientific realist perspective, Friedman (op. cit.) considers also constructive empiricism while arguing for realism with respect to space-time. His main argument is a version of what Putnam called the conjunction objection: if two theories are both true then so is their conjunction, but if they are both empirically adequate their conjunction may not even be consistent, let alone adequate. The logical point is correct of course, but its importance depends on how scientists go about combining theories which they accept. A typical pattern to be found in the history of science is that simultaneously accepted theories are used very gingerly in combina-tion, and that their successor-theory is inconsistent with both while accounting for their success -- by saving the phenomena which they saved.(11)
There is, as far as I can see, no indispensability argument, of the sort which Putnam gave for realism about mathematical entities, for example, for the reality of the world. Such an argument would have to show that the existence of the world as a whole, whether as system of bodies or as space-time, is required for the success of cosmology. It appears that this is the case at most relative to certain philosophical views about science, to wit, that the embeddability of all actual phenomena is such a model is not sufficient for success.
To conclude then: whether or not the world exists is not settled by the success or acceptance of physical cosmology, except relative to certain philosophical points of view.
IV
There is, however, a second tactic for bestowing the authority of science on philosophy. If the assumption that there is a world (at least one) is not indispensable to science, nor to our attempts to render science intelligible, can philosophers still reasonably argue for its existence? They can indeed if the philosophical enterprise is continuous with or 'of a piece with' science itself, and if it is the aim of this common enterprise to equip us with a coherent, comprehensive Theory of what there is and what it is like. For if that is so, the development of that Theory will surely involve audacious postulation, and might require postulation of the world's existence.
What exactly does this view of philosophy require? The view is not outré; it is clearly presented in Quine's well-known essay "Two dogmas of empiricism," to name only one example. The scientist asks whether our world-picture should include quarks, and arrives at an answer, through theory-choice in a scientifically acceptable manner. Then the philosopher asks whether our world-picture should include abstract entities such as sets, numbers, properties, or propensities, and also certain kinds of concrete entities: events, persistent objects, time-slices, worlds. If philosophers can show that postulating these entities is indispensable to science, they have simply brought to light an ontological commitment already implicitly present in science. However, in philosophy as in science such open and shut existence proofs are at best rare. Enter, therefore, theory-choice as practiced by the philosopher: it is legitimate and authoritative if it parallels scientific practice as regards theory choice. In this way ontology, as pursued by the analytic philosopher, takes on the legitimacy of science.
The point of this methodological digression is just this: many objections to philosophical theory construction can be rejected on the basis that, if cogent, they would equally apply to science. If the objector does not want to criticize science, this ends the matter. First, ontological claims are indeed not directly verifiable or falsifiable. We cannot simply go and check to see whether the world exists, let alone whether other possible worlds are real. But neither are many scientific statements so fortunate: we cannot verify the existence of quarks or the Big Bang directly. Our science faces the tribunal of experience as a whole. Since this is so for science, it should be allowed for ontology. As for the goose, so for the gander.
This argument is incomplete without a clear description of what theory-choice is like in scientific practice. Quine's essay added a certain methodological holism to this rejection of verificationism. Among the theories that admittedly fit the data, the one that is best overall is chosen -- and in the determination of "better" and "best" all theoretical virtues, including purely pragmatic advantages, are allowed to count.
Scientists infer to (or at least, they accept) the best explanation, it is said, or if you like, the best theory. This means that after weighing all the virtues and limitations of rival theories the scientific community -- or the rational epistemic subject/agent -- believes the best of these, believes it to be true. (More modestly, the conclusion is that the best of these is believed to be empirically adequate, still a very audacious belief since it implies accuracy of future predictions.) Accordingly, if in philosophy we do likewise and believe the best ontology to be true, we shall be scientific and rational and act as good citizens of the world of science. In philosophy, of course, "best" does not refer to more accurate empirical predictions or control of nature but to theoretical virtues alone: answering philosophical questions, solving conundrums, resolving paradoxes. The clear conviction here is that if theory-choice in ontology takes the same form as theory-choice in science, then it has the same epistemic status.
There are however two points which seem to me to defeat this defense of ontology. The first, which should at least raise a suspicion, is that an entire epistemology has to be marshalled in defence of even very minimal ontic commitments. To establish even the legitimacy (never mind the plausibility!) of "the world exists," one needs to appeal to the possible -- at least conceivable -- theoretical benefits that might accrue to a Grand Unified Theory which implies that statement. The appeal of those simple, plausible sounding, attractive arguments in section I above is explicitly disowned by this point. No, the premise "You are acquainted with this world, and so am I" has no independent cogency; it stands unmasked as a highly theoretical postulate, to be supported only by the promise of overwhelming benefits of some envisaged total Theory which says so.
The second critical point is the much weightier. Whatever form theory-choice actually takes in science, we must carefully consider its rationale. Doing so, we may well discover that this form of theory-choice is appropriate under some conditions and not under others. The form described above is of the 'inference to the best explanation' type. I will not here contest it general epistemological status.(12) It is clearly ampliative, therefore it can have the bad consequence of giving us false beliefs. Any rationale for this practice must therefore imply that this risk of disvalue is outweighed by the chances of sufficiently valuable good consequences.
Precisely at this point the relevant differences between science and philosophy spring to the eye. The legitimacy of audacious postulation in science derives exactly from what is at stake in science but not in philosophy. The gleam in the scientists' eyes is the prediction and control of nature: cure and eradication of disease, enhancement of communication and transport, population control, new sources of food and energy. If scientists come to believe a new postulate, then one of the bad consequences that may happen is that they contract a false belief. But the disvalue of that consequence pales into insignificance besides the values and disvalues of other possible consequences.
Is it so with philosophy? Adoption of an ampliative method of audacious postulation of ontological claims may indeed lead to good philosophical theories, whatever 'good' in this context means. It comes most certainly the risk -- a risk not easily assessed -- of giving us false beliefs about what there is. Are these risks outweighed somehow? Consider two examples to evolve our own judgement in this matte. It was discovered not long ago that the success of Newtonian science had indeed given us incredibly powerful, for reaching false beliefs, about nature for several hundred years. No one, I think, took that as reason to regret Newton's life and work or its scientific influence. Imagine instead that Descartes had achieved the philosophical domination once enjoyed by Aristotle or Aquinas, and that we had discovered the untenability of mind-body dualism not in the late 17th century, but in the 20th. Would any of the reasons for judging the acceptance of Newtonian science beneficial after all have their parallels in this case? I think not. As far as I can see, benefits we can hope to gain from metaphysics-played-well pale into insignificance beside the disvalue (small as it may be from a practical standpoint!) of falsely believing in the existence of unreal entities dreamt of only in our philosophy. It is a good thing that pragmatism re-emerged to defeat the excesses of logical positivism, and it is true that it thereby removed much of the positivist critique of metaphysics. But if we take the pragmatist critique one step further, recalling its insistence on the values genuinely implicit in human practice, the legitimacy of metaphysics disappears once more into thin air.
There are several obvious objections to this conclusion. First of all, it hinges on a value judgement while not all may share. The benefits of having a comprehensive ontology which fits well with science and resolves more philosophical puzzles than any known competitor may be thought to greatly outweigh the risked disvalue of false beliefs. Moreover, the conclusion hinges on probability judgements: what are the chances of this method of theory choice leading us to a true ontology? But if these value judgements and probability assessments are agreed to be merely subjective, then ontology has the marginal status of something that some people like to do -- and my critical conclusion is in effect fully granted. If they are claimed to be communal (in the sense, perhaps, that this country is dedicated to the ?? that all men are created equal) or objective, then it is remarkable that no one engaged in ontology has attempted to demonstrate that. It is certainly not the case that the value of philosophy in general hinges on the value or success of ontology in the sense of ??! In that case, then, we must conclude that very far from having provided a rationale for this enterprise, ?? left us with an open question which no one has yet tried to answer.
V
There is a third and final tactic for transferring authority from science to ontology. It is similar to the second in that it rests on the assertion that a certain form of theorizing found in ontology is exactly the same as is found in science. The form of theorizing in question is not audacious postulation but linguistic innovation; in both cases justification is said to come from the virtues of the newly crafted theory taken as a whole. Science and its language develop in tandem, and no sharp line can be drawn between new factual postulates and linguistic innovation. A new theory may be capable of being formulated only in a richer language than we had heretofore; new wine cannot be poured into old skins. Quine's "Two dogmas of empiricism" made the case well, mentioning that even so-called logical truths may be jettisoned if our old linguistic framework is too Procrustean to accommodate the new physics. At one point Putnam argued that we had indeed found it to be so, and claimed an essential similarity between the rejection of absolute simultaneity (the two-place simultaneity relation being replaced by the three- place simultaneity in a frame of reference) and a switch from classical to quantum logic.
Again it is claimed that what suits the goose of science goes for the gander of ontology. The formulation of an ontological theory may involve linguistic innovation. Perhaps higher-order quantification is introduced, or special terms are coined (such as "natural" in "natural kinds" or "natural classes"), or the transitivity of the part-whole relation is made into a point of logic by regimenting the language. A naive empiricist might regard such new-fangled discourse as unintelligible. But this linguistic innovation is accepted practice in science: recall the feat of turning "simultaneous" into a three-place predicate, the introduction of theoretical terms anchored in mathematical models rather than observable phenomena, the fact even that Newton's "mass" and "force" (let alone Einstein's) had no place in Descartes' language. If philosophers introduce new count nouns or other linguistic resources, they are entirely within their scientific rights.
I have no wish to quarrel with this contention, as far as it goes. It will be necessary a little later on to take a closer look at responsible forms of linguistic innovation. Meanwhile, however, the health and well-being of ontology itself requires very strict limits on what can be accomplished thereby. The distinction between factual postulate and linguistic innovation cannot be allowed to have wholly disappeared, if ontology (and perhaps even science) is to have any point at all.
The crucial juncture is exactly at claims of existence. Carnap proposed in "Empiricism, semantics, and ontology" that existence assertions (e.g. for abstract entities) be allowed as concomitants of the adoption of forms of language, and that such adoption be regarded as purely a practical rather than a theoretical decision. We cannot but suspect some irony here, or at least an attempt to extend Quine's pragmatism so far as to ruin the whole idea of analytic ontology. Linguistic innovation had better not bring into existence any more than new ways of speaking about what there is already, if ontology is to chart a course between fallacy and idle word play.
That, at least must be the reaction of any philosopher seriously engaged in ontology. It is quite explicit in van Inwagen's Material Beings, as point (9) of his Preface (pp. 6-12). We may fashion our language anyway we like, and perhaps some things exist which we can neither describe nor denote until we refashion our language. But whether or not there are things, of the sort whose existence is asserted after this refashioning, is a question of fact and is quite independent of the history of language.
Perhaps we can also put it the other way around. If philosophers were driven to argue for the assertion "the world exists" or "more than one world exists" or, to use van Inwagen's example
there is an object which has X, Y, and Z and no other objects as parts (p. 8-9)
on the basis that they are simply engaged in a kind of linguistic innovation that makes these statements (in their new idiolect) true, then they would have lost the game altogether. If such extensions of the language are innocuous and cost-less, then they are also benefitless. On the other hand, if such linguistic extension bring it about that the ontological claims are true, then they certainly cannot be the substantive assertions of existence, of what there is, of our common understanding which alone gives point to ontology, if it is to have a point at all.
VI
So far my argument has been predominantly critical. More constructively I want to describe a use of "world" which, I believe, will suffice for all legitimate common and philosophical needs. Moreover it seems that this sort of use accounts for all the uses familiarly detailed in dictionaries. (Literal use may of course be augmented with metaphorical and analogical uses: as long as those are not mistaken for putative new contributions to ontology, who could take them amiss?) The 'ordinary' sort of use I shall now describe does not rely on an inflated ontology; this alone may suggest a serious doubt as to the intelligibility of any use which does.
The use of "world" which I wish to explain I shall call, for lack of a better or a standard term, schematic. My explanation of this sort of use will have two parts. I shall first discuss how we can construe more 'ordinary' (less 'philosophical') uses of "world", and then I shall describe the very similar case of another, more technical term, "isomorphism".
Below I shall cite dictionary examples, but for now consider:
"All the tears in the world won't help you now"
"Not a single judge in the whole world would believe you."
In both of these we could take "world" to stand for the entire Universe, assuming that there is such a thing. But equally, we can in each case take "world" to stand for something smaller or more delimited, though not explicitly specified: the whole of human history for the tears, the whole of current humanity for the judges. In each case we could also have utilized a purely quantificational phrase:
All the tears there are or ever have been...
Not a single judge alive today...
If we now add the "world" phrase again, we do not get nonsense but only rhetorical redundancy. ("All the tears there are or ever have been in this world...."). The additional "world" phrase is empty: it has no relevant effect on the content (which is not to deny that the domain of discourse is a relevant parameter and that we could take "the world" to denote that set or alternatively some putative object composed by its members).
Let us see how this accords with the standard record of English usage, the Oxford English Dictionary. This source explains that the etymology of "world" is "wer" (man) plus "ald" (age, life) -- thus "age of man," "life of man." The main headings under which it lists examples of use are:
I. Human existence; a period of this.
II. The earth or a region of it; the universe or a region of it.
III. The inhabitants of the earth, or a section of them.
IV. Idiomatic uses and phrases.
Classification of the examples under these headings suggests that both context and explicit content are used to hypothesize denotations for the term when found. For the first heading a striking example commenting on Henry VIII's divorce mentions "[o]thers which foretold this dolorous doleful wretched world that followed." A related old but still used locution is "of the world" by which religious denote the secular sphere of life. "The Roman world" denotes a part or period of human existence.
Perhaps the most common uses of "world", however, fall under the first part of the second heading: to denote the earth or a part thereof. Around the World in Eighty Days is a story of an earthly circum-navigation. Paradise Lost says of Adam and Eve that "The World was all before them, where to chose Their place of rest" and means the earth, not the galaxy. When W.S. Gilbert wrote "It's Love that make the world go round," he meant no more. The New World, of course, is part of the earth, as is the Old and the Third. The request to "imagine a world free of hunger and poverty" may sound like I am asking you to think of a possible world -- a truly philosophical request. But it is quite easily construed as asking you to imaginatively entertain the supposition that the human race here on earth is free of hunger and poverty, and what things would be like then.(13)
To continue with the second part of heading II: "World" is also used to denote larger systems, and then occurs readily in the plural. Before this century, disquisitions on the plurality of worlds concerned other planets, stars, solar systems, galaxies. This yielded idioms concerning great quantities and vast distances: "they are worlds apart," "I would not stand in her way for worlds." Sometimes the demarcation is less natural, more abstruse, or more fanciful than astronomy gives us: Locke (Hum. Und. IV, iii, 27) claims that the intellectual World is greater and more beautiful than the material World. Somewhere in between are the world of physics or chemistry, the inorganic world, the Outdoor world, the underworld, a Boy's World.
With some of these we may have strayed into classification III: The inhabitants of the earth, or a section of them. The OED lists the world of Fashion and the Learned World, and quotes "Two noblemen, whose names are as eminent in the poultry world as in rank" from the Poultry Chronicle of 1854.
Finally, worlds can be something like mythical regions: the nether world, the world to come, a better world, the dream world, the world of gods and heroes. It seems to me that they deserved a heading of their own, though the OED sprinkles them throughout. To this sub-cluster belong Dante's "With Virgil ... visited the nether world of woe," Shelley's "gleams of a remoter world Visit the soul in sleep" (Mont Blanc), Tennyson's "second world" and "world-to-be," and Fizgerald's dismissive lines in the Rubayyat (xxv) about "All the Saints and Sages who discuss'd Of the Two Worlds so learnedly." It is among such purported regions of reality magically accessible to us that I think we should locate the worlds discussed in the following conversation famous among children:
"But do you really mean, Sir," said Peter, "that there could be other worlds -- all over the place, just round the corner -- like that?"
"Nothing is more probable," said the Professor, taking off his spectacles ....
found in C. S. Lewis' The Lion, the Witch and the Wardrobe.
The dictionary proceeds on the assumption that "world" always denotes something, some supposedly existing entity or system, but not that it always denotes the same thing. In each example found there it seems that we could take that something to be something which did not exhaust all there is, but rather to encompass some relevant class or region. When we interpret quantifiers, we specify a domain of discourse. But (a) we do not require that domain to be designated by any expression in the statement, (b) the domain we specify need not be a physical entity, it can easily be a set, (c) we take its identification to be contextual and (d) the identification is typically, even in context, largely underdetermined. The "world" phrase which seems to point to this domain of discourse may help to determine it, or may just signal the need to do so.
If this is correct then we can often enough regard "world" as either not standing for anything or else denoting a set. Moreover, when we do regard it as standing for something (even a set), its referent is contextual -- there is no single thing which it denotes in all such occurrences. As an instructive similar example, less in common use but equally hovering at the edge between technical precision and philosophical reification, I want to mention "isomorphism." There are many important general insights expressed in statements in which this noun occurs 'all by itself' so to speak. Mathematics describes its objects only up to isomorphism; that is an important point about mathematics. But this air of linguistic self-sufficiency is deceptive; the term is context-dependent. To say that structure A is isomorphic to structure B means that there is a one-to-one correspondence between them which preserves certain operations and relations. That is an explication of the vaguer phrase "A and B have the same structure." But explication, as usual, forks in many directions, and in this case remains itself vague unless those "certain" operations and relations are specified.
In mathematics, the way to precision is straightforward. If I ask whether two groups are isomorphic, in those words, using the word "group," then I am asking about correspondences which preserve the group operations. It may still be added that the two structures are very different in other respects. For example: the elements of the one are permutations and the elements of the other are operations on a Hilbert space. There is a very precise sense in which the two entities have the same structure and an equally precise sense in which they do not. But what if I ask the question about two physical entities, or one physical and one mathematical? Is this pendulum, a physical table-top model, isomorphic to the one in your study, is it isomorphic to the mathematical model described in your physics text?
It seems that we can ask of such things only whether they have the same structure under a certain description -- and we must specify the description before the question can be completely identified. There are however philosophical views according to which that is not so in principle. Such views impart a definite and complete sense to the question whether two physical entities have the same structure, whether they are, so to speak, totally isomorphic. For example, when van Inwagen lays out his assumptions at the beginning of Material Beings, he says that the world is particulate, i.e. every material being is composed of parts which themselves have no proper parts (mereological atoms). Add to this any of the varieties of anti-nominalism which entails that there is only a restricted set of real relations (as opposed to arbitrary sets of pairs, triples, or other sequences). Then it makes sense to ask of two structures whether they are isomorphic, tout court, without qualification. For that means then: is there a correspondence between their mereological atoms which preserves all the real relations?
Relative to some philosophical views, therefore, "isomorphic" is not, in the end, a merely schematic or context-dependent term. There is on such views a straightforward, non-contextual assertion of isomorphism (which may of course be restricted, with the restrictions imposed by contextual factors in discourse, but that is a general point about all terms). Lacking such a philosophical view, however, we can use the term "isomorphic" without qualification only in a schematic way. That is, lacking such a view, we must take a bare assertion of isomorphism to be not a full fledged statement but a statement schema. Statements are produced therefrom by 'filling in the blanks.'
The very same point may be made about uses of "world." The form "All... in the world are..." is schematic in several ways: the "in the world" phrase is itself a blank, to be filled in. That blank is filled in when the domain of discourse is specified for the quantifier. There are again philosophical views relative to which this is not so, in the end. These are views according to which there is a single thing, the world, which is the ultimate and complete candidate that can fill every such blank. Lacking such a view we must do what the editors of the OED do: for each occurrence of "world" we identify some plausible region, system, or set of entities which delimits the statement's domain of discourse. In other words, we treat the general and unqualified use of world as schematic only, relying on contextual factors to turn the schema into a statement.
To sum up then, here is my suggested alternative to the idea that world is a count noun. Instead it is a context-dependent term which indicates the domain of discourse of the sentence in which it occurs, on the occasion of utterance. It plays this role sometimes by denoting the domain (a set), and sometimes by purporting to denote an entity of which the members of the domain are parts. In the latter case we need not take that very seriously (it may be metaphor, colorful language, rhetorical extravagance); important is only the indicated domain of discourse.
VII
There is one major contention which I have left at least uncontested so far, to which I now turn. It is easy to imagine someone who accepts my criticism so far, but considers the meta- physical enterprises here criticized as perfectly legitimate, if idiosyncratic. This imagined reader says that, having noted the schematic use of "world" we can after all postulate that there exists a single entity of which all others are parts. That entity can then in principle always be taken to be the referent of "world" and to demarcate the range of the quantifier, although context will often require that range to be restricted to one of its parts. The postulate may, if my arguments here have been cogent, lack all support or rationale, its benefits being illusory and its practical consequences nil, but is otherwise unobjectionable.
The failure of verificationism, and perhaps even more its ill-advised bludgeoning use before that failure became apparent, have made us wary of calling any philosophical discourse unintelligible. But, warily, I do want to raise the question of intelligibility.
Analytic philosophy is a great critic, but does not always hold itself to its own standards. It is easy enough to criticize, on purely grammatical grounds, such sentences as "the nothing noths" and "the world worlds." The author of such texts invites grammatical criticism, deliberately giving his language an unfamiliar, suggestively defective form to draw attention to philosophical perplexities. It is not equally easy to use analytic tools on philosophical texts which strive mightily to preserve grammatical and logical form, to express putative solutions in plausible-sounding, familiar-looking language. By those means, form criticism (if I may adapt that term) is preempted, and defects are buried deep below the linguistic surface. But we should not relinquish those stringent demands upon intelligibility that analytic philosophy introduced -- we should refuse to weaken the criteria of intelligibility when we come to its own products, or to reduce those criteria to mere compliance with logic and grammar.
In an attempt to understand philosophical (or any) discourse, we rely crucially on Vorverständnis, 'fore-understanding' or 'pre-understanding' based on clues that align this discourse with what we understand already. Even if in utter frustration with the recalcitrance of certain philosophical problems the familiar language is distorted into "The world worlds" we are not completely at a loss. The sentence has the structure Noun Phrase + Verb, and the Noun Phrase is of the form 'The' + Common Noun; we know a great deal about the form of the truth conditions for sentences of this type, and hence about the logical inferences in which they can occur. The verb is unfamiliar, it is a noun turned into a verb. But we have seen that happening before, and know something about how that original noun is meant to be connected to the new verb. The unfamiliarity of the coined verb makes us aware of how little, after all, our pre-understanding gives us: it is not enough for full understanding.
Not all fractures of meaning draw attention to themselves by peculiarities of diction. In metaphors, the pre-understanding is not halted by any flaunting of grammatical categories. No new words are added there to our categories of verbs and nouns. Yet we are halted at a point where we have only pre-understanding: we know the form of the truth conditions for "the evening is spread out against the sky" but at first don't believe that those conditions obtain -- we realize suddenly we are on our own if we are to reach some fuller understanding.
It seems to me incumbent on analytic philosophers to be explicit in the construction of their own discourse. It is a rhetorical strategy, not to be respected among us, to kill these moments of unfamiliarity by linguistic care, for it is in such moments that genuine lack of understanding comes to light. Good grammar and dead metaphors do not, by themselves, intelligible discourse make.
So how shall I, if I am an analytic philosopher engaged in ontology, proceed responsibly when introducing a theory in which it is asserted that the world, or even more than one world, exists? I must carefully exhibit my compliance with grammar, logic, and my preferred form of semantics. I must indicate by linguistic markers where I might be departing from dictionary-recorded or other standard usage, to note where common pre-understanding does play a role and where it ends.
Suppose, for utter clarity, I keep the word "world" reserved for schematic use, and introduce "Sworld" to designate the supposed totality of everything. Suppose in addition I take Sanford's strictures to heart and introduce the term "Spart" as follows: (1) if A is [a] part of B then A is [a] Spart of B; (2) the Spart-of relation satisfies the axioms of mereology; (3) the Sworld is that of which everything is [a] Spart. This can be done carefully so that "Sworld" is a count-noun in its grammatical and logical behaviour, and "Spart of" a similarly qualified relational term. Moreover, we can give an abstract semantics -- i.e. a model theory, something that has the form of specifying truth conditions -- to fix their use in all logical respects. But this may not be enough: it seems to me that we do not yet understand "Spart" nor, therefore, "Sworld". There would seem to be very many relations that Spart, if could be, given (1) and (2), and we should not pretend that we have specified one of them.
What would happen if we relaxed that demand for rigor? Since we might regret that we had created a language system only distantly related to known human or even philosophical interests, it might seem fortunate, might it not, that there are already words in common use, (to wit, "world" and "part") which play almost all of the role of "Sworld" and "Spart"? If we suppress the initial sibilant, we will sound as if we are making perfect sense, at least logically and grammatically. We may then encounter some stares that we take to be incredulous, since our statements do not accord with the pre-understanding of "world" and "part." Perhaps the stares are not incredulous? Perhaps, they express a sense of utter lack of understanding?
It seems to me that however we continue this rational recon-struction of the philosopher's art, we reach a negative verdict on its works. One possible continuation is to say that in the philosopher's mouth, "the world exists" is only a funny way of saying "the Sworld exists." The transformation is only a bit of rhetoric designed to give us a mistaken sense of familiarity. But "the Sworld exists" has such a thin veneer of meaning that the question of credence, of credulity or incredulity, does not arise.
The other possible continuation is to say that on the contrary, through this careful construction, the philosopher has found a stronger guarantee of meaningfulness and literal significance. The criteria explicitly met are all the criteria there are; hence they are the sufficient marks of meaning -- ipso facto, this discourse makes sense. "Sworld" is intelligibly related to "world," taking over a carefully selected family of uses, regimenting them, and is then used to make new, logically contingent, fully intelligible assertions. If we are careful not to let other usage of "world" creep back into our professional discourse, then "the world exists" is a perfectly good way of saying "the Sworld exists." The unfortunate negative verdict forced on us by this second line of reasoning is that it is very easy -- too easy -- to make sense. We can sit in our closets and in a perfectly meaningful way, kneading and manipulating language, create new theories of everything, and thereby, important contributions to ontology. In other words, to put it a little more bluntly, this 'world play' we engage in here is but idle word play; though shown to be meaningful, just idle word play nevertheless. On both verdicts, ontology is not what it purports to be; but to me the second verdict is the more devastating one.
1. The author wishes to thank Jenann Ismael, Marc Lange, Béatrice Longuenesse, Alvin Plantinga, Gideon Rosen, Steven Savitt, Abraham Stone, and Peter van Inwagen for helpful discussion and correspondence.
2. Indeed, the word "world" has very few uses as a mass noun, the only familiar one perhaps the occurrence in Marvell's "Had we but world enough, and time."
3. For a careful study of the progress of Kant's pre-Critical idea of world see Michael Friedman, Kant and the Exact Sciences (Cambridge, MA: Harvard University Press, 1992), pp. 5, 7, 13-14, 25, 27, 32n.54, 33-34.
4. Ithaca: Cornell University Press, 1990. See further his Metaphysics (Westview and Oxford, 1993), 112-113.
5. Lesniewski's mereology, the theory of the part-whole relation, was given a more use-friendly form in Leonard and Goodman's calculus of individuals. I write "is [a] part of" agreeing with David Sanford ("The problem of the many, many composition questions, and naive mereology" Nous 27 (1993), 219-220) that this distinction is not observed in the formal theory. For the resources of mereology see further David Lewis, Parts of Classes (Oxford: Blackwell, 1991).
6. On the Plurality of Worlds (Oxford: Blackwell, 1986). He indicates that in general a world is the sum of all entities spatio-temporally related to a given entity, but also allows for the replacement of spatio-temporal relations in this notion by any similarly structured family of relations (page .....)
7. "Naive mereology, admissible valuations, and other matters," Nous 27 (1993), 229-34; quote from p. 229.
8. Structure and Evolution of the Universe (New York; Harper, 1959), page ... Much less bridled usage can be found in recent cosmology, of course; see for instance C.B. Colling and S.W. Hawking in the Astrophysical Journal (1973), 319 for a striking shift in discussion from models to universes. (I owe the reference to Alvin Plantinga).
9. Foundations of Space-Time Theories (Princeton: Princeton University Press, 1983); see also Steven Savitt, "Selective scientific realism, constructive empiricism, and the unification of theories," forthcoming in Midwest Studies in Philosophy, vol. 18.
10. Cambridge, MA: MIT Press, 1989. See also the review by Steven Savitt, Dialogue 31(1992), 701-06.
11. Cf. The Scientific Image (Oxford: Oxford University Press, 1980), p. 72. Savitt argues correctly ("Selective scientific realism...." cited above) that this is also the pattern in Freidman's own main example of the kinetic theory of heat in relation to atomic theory.
12. See my Laws and Symmetry (Oxford, 1990), Chapter 6 and Chapter 7, sect. 4.
13. I owe this example to Abraham Stone rather than to the OED. We can easily make up many examples of quite ordinary discourse which it is fun to construe as really being about other possible worlds, but the fun comes in part from the recognition that the imagined speakers had a very different concern.