Midwest Studies in Philosophy 12 (1988): 145-55.
There may be small differences between
this text and the published version.



Curtis Brown


Idealism is an ontological view, a view about what sorts of things there are in the universe. Idealism holds that what there is depends on our own mental structure and activity. Berkeley of course held that everything was mental; Kant held the more complex view that there was an important distinction between the mental and the physical, but that the structure of the empirical world depended on the activities of minds. Despite radical differences, idealists like Berkeley and Kant share what Ralph Barton Perry called "the cardinal principle of idealism," namely, the principle that "being is dependent on the knowing of it."1 I believe that Hilary Putnam intends his "internal realism" to be a version of idealism in this broad sense; although many of his arguments concern semantic notions like truth and reference, he takes these semantic arguments to have ontological consequences. This is strongly suggested, for instance, by his claim that "'objects' themselves are as much made as discovered, as much products of our conceptual invention as of the 'objective' factor in experience."2 Or again there is this rather Kantian metaphor: "the mind and the world jointly make up the mind and the world."3 But just what is Putnam's ontology?

Putnam has been defending a version of anti-realism in his writings of the last several years. He describes his view as "internal realism," where the term "internal" is intended to indicate his rejection of the view he calls "metaphysical realism." Putnam combines internal realism and metaphysical anti-realism in something like the way Kant combined empirical realism and transcendental idealism. I would like to investigate here one of the numerous lines of thought by means of which Putnam has defended internal realism. The argument I shall be focussing on is what might be called the "argument from reference." Putnam also has arguments from vagueness, from the nature of truth, and others; I will not discuss these.

It would be interesting to undertake a full-scale investigation of the relations between Kant's transcendental idealism and Putnam's internal realism. Putnam himself invites this comparison by his frequent use of Kantian terminology and his allusions to Kant's work, and indeed the parallels are numerous. Both consider, somewhat uncomfortably, the view that there is an unknowable noumenal world behind the phenomena. Both are motivated in part by the threat of scepticism: Kant by scepticism about our ability to know the external world, Putnam by scepticism about our ability to refer to it. Both Kant and Putnam hold that the world we know and talk about is empirically real, but both hold also that it is mind-dependent. Putnam at least flirts with the Kantian idea that there are a priori truths about the world.4 Putnam, like Kant, stresses the pervasive importance of causation, and argues that causation is partly our own imposition on the world.5 Putnam argues that science presupposes values, in a way reminiscent of Kant's claim that science requires the regulative use of the Ideas of Reason. Like Kant, Putnam argues that the relation between morality and rationality is much closer than empiricists (from Hume to Bentham to the positivists) typically can allow.6

I shall not attempt such a full-scale investigation here. My purpose is to discuss one particular issue: Putnam's Kantian response to a kind of scepticism, namely scepticism about the possibility of reference. The paper divides into three parts. In section I, I discuss the view that in classifying or organizing the world into objects, we create objects. I explain what I take to be a plausible realist alternative to this view. The remainder of the paper discusses Putnam's reason for rejecting this realist account, namely, his argument that reference to mind-independent objects is impossible. I propose to accept the argument and see what follows from it. If determinate reference to mind-independent objects is impossible, this seems to leave open two possible responses. One might hold that we refer indeterminately to mind-independent objects, or that we refer determinately to mind-dependent objects. The first view retains a realist ontology at the expense of determinate reference; the second retains determinate reference by adopting an idealist ontology. In section II I discuss the former possibility, a response Putnam rejects. Putnam claims that this view introduces the problems of Kant's notion of the thing-in-itself; I question this. In section III I discuss the second response, that we refer determinately to mind-dependent objects. I suggest that this view is actually at least as Kantian in spirit as the one Putnam rejects. I also suggest that it is not clear that this response really succeeds in meeting Putnam's sceptical problem about reference.





Let us begin, then, with what we might call "the argument from organization." According to this argument, because our language organizes, selects, categorizes, and classifies things, the things we talk about are not part of the world as it is in itself, but rather are created by us. To be sure, we do not create the raw material out of which we produce objects, but the objects themselves, as opposed to neutral stuff of the world in itself, are our own creations. If this is correct, then the world we talk about is mind-dependent, and idealism in the broad sense I am concerned with is true.

I do not mean to suggest that Putnam is committed to this argument, at least in the unsophisticated form in which I have just phrased it. But he does say things that suggest it. He writes, for instance, that: "'Objects' do not exist independently of conceptual schemes. We cut up the world into objects when we introduce one or another scheme of description. Since the objects and the signs alike are internal to the scheme of description, it is possible to say what matches what."7 This suggests that if we did not "cut up the world into objects" there would be no objects, so that we bring objects into being by organizing the world. Again, Putnam criticizes the view that there are "Self-Identifying Objects," explaining that "this is just what it means to say that the world, and not thinkers, sorts things into kinds," and he opposes to this idea the view that "'objects' are as much made as discovered,"8 suggesting again that the fact that we organize the world shows that we create objects. To take a final example, Putnam speaks approvingly of the idea "that objects and reference arise out of discourse rather than being prior to discourse."9

As Putnam is well aware, similar arguments are common in contemporary thought; one particularly explicit and unapologetic defense of the view may be found in Leszek Kolakowski's "Karl Marx and the Classical Definition of Truth."10 Kolakowski, as the title suggests, attributes the view to Marx. Just as Putnam criticizes the view that objects are "self-identifying," Kolakowski criticizes "Aristotelian realism, which posits that the species and genera into which the sciences divide reality are merely copies of the genera and species of this reality reflected more or less exactly, but ever more exactly, in the mirror of consciousness."11 As Putnam says that objects are "as much made as discovered," Kolakowski says things like this: "Nature as composed of parts and species is an 'artificial' creation;"12 or again: "In this world the sun and stars exist because man is able to make them his objects;"13 or again: "If, for Marx, man replaces God-the-Creator, still He is not the God of Augustine or Thomas Aquinas, a God who gives birth to the world out of nothingness; rather, He reminds us of the God of the Averroists, who organizes the world out of previously existing material."14

In Kolakowski especially, one gets the idea that there are only two possible views. On the one hand there is "Aristotelean realism," which holds that the universe divides itself up into a single set of kinds, and that the proper role of language is to copy the world's own single set of divisions, to classify the world in its own way. This seems to be the same view Putnam describes as the view that the world contains Self-Identifying Objects. On the other hand, there is Kolakowski's preferred view, which seems also to be Putnam's: that the world "in itself" is undifferentiated (he speaks of a "pre-existing 'chaos'"15), and that we create objects by inventing classifications. Kolakowski's argumentative strategy seems to be to argue that since Aristotelian realism is false, his own version of anti-realism must be correct.

Now if this were the anti-realist's only argument that the world is mind-dependent (as in fact it seems to be Kolakowski's only argument), then the realist could quite plausibly assert that it simply is not good enough. One can grant that it is false that the world has one true classification or division into objects and kinds of objects without granting that it has no organization of its own. A modest realist might well say that the world has infinitely many classifications, that it contains infinitely many different kinds of objects. Such a moderate realist could say that when we develop a language we are not imposing an organization on the world, but selecting one of the world's organizations for our own use. On this view the world "in itself" has more objects than we usually talk about, not fewer.

Modest realism shares with Putnam's and Kolakowski's views an emphasis on the importance of human activity in determining what sorts of things we will be most concerned with, but, if accepted, makes unnecessary the idealistic view that in dividing up the world we are somehow populating our world with objects that were not there before.

But this is not the end of the matter. Kolakowski, as far as I can tell, simply does not consider modest realism. Or it may even be that he accepts modest realism, and that his talk of "creating" objects carries no really anti-realist implications but is only a metaphorical way of expressing a modestly realistic view. But Putnam carries the matter further than this. He argues that determinate reference to mind-independent objects is impossible; we may take this as an argument that if the goal of language is to select from among the world's many organizations, it can never succeed in picking out just one. To this argument I now turn.




Putnam places a good deal of weight on an argument that determinate reference to mind-independent objects is impossible. He has discussed the argument in a number of places, most notably in "Models and Reality"16 and in Chapter 2 of Reason, Truth and History. The general idea of the argument is that no amount of holding-true of sentences, including sentences that use the term 'reference', will suffice to uniquely determine referents for our terms. No matter how elaborate our theories, including our theories of reference, if they are consistent they can be made true by endless different assignments of referents to their singular terms. Indeed, Putnam argues in Reason, Truth and History that endless assignments of referents can even leave the truth conditions of all the sentences of any consistent theory the same. So no matter how we try to constrain reference from the "inside," so to speak, it will remain indeterminate.

The conclusion is strikingly similar to Quine's thesis of the inscrutability of reference,17 and indeed Putnam speaks of himself as extending Quine's result.18 The discussions of Putnam's recent work that I am aware of assume that the argument is successful; disagreement comes over what to conclude from it. Most of Putnam's critics have urged that, since the argument shows that "internal" constraints on reference do not suffice to pick out a single correspondence relation between words and the world, there must be some sort of "external" constraint which does this. Two sorts of external constraints have been suggested. David Lewis has suggested that what is needed is an "inegalitarianism" of classifications of the world, the view that the world really does in a sense come pre-divided into objects. What is needed, Lewis says, is "the traditional realism that recognises objective sameness and difference, joints in the world, discriminatory classifications not of our own making."19 It then becomes a constraint on interpretations that they maximize the assignment of "elite classes" as referents. Not that this constraint operates because we intend to refer to elite classes: perhaps we do, but that intention, like all our other intentions, is susceptible of countless interpretations, including interpretations in which 'elite class' refers to classes that are not objectively elite. Rather, it is just in the nature of things that some classes are more gerrymandered than others, and the correct assignment of referents will maximize the assignment of the least gerrymandered classes, so far as this is compatible with meeting the other constraints on reference.

The second sort of constraint that has been proposed is a causal constraint. The idea here is that the referent of a term might be the object suitably causally connected to peoples' use of the term. There is a kind of standoff about this constraint. Putnam points out that we cannot from the inside determine the referent of 'cause' any more than we can other referents, so adding some such sentence as "Cats cause the use of the term 'cat'" to our total theory will not help to fix the interpretation of that theory. The total theory could be interpreted by taking 'cat' to refer to some bizarre class of objects and 'cause' to refer to some bizarre relation that relates these bizarre objects to our use of the term 'cat'.20 Does it follow that we cannot single out a relation, causation, that narrows down the acceptable interpretations? No, as causal theorists reply. Our use of the term 'cause' is caused by the relation of causation, so that if the causal constraint on reference is correct it determines the interpretation of 'cause' "from the outside."21

Thus the standoff. If Putnam is correct that the only constraints on reference must be internal, then the causal theory of reference does not help narrow down interpretations, since it cannot succeed in picking out a single relation, causation. On the other hand, if there can be external constraints on reference, then the interpretation of the causal theory itself may be determined by the external relation of causation, so that it can succeed in narrowing down the possible interpretations of the rest of our theory. Nevertheless it seems to me that in one sense Putnam is in the stronger position here (and not just for the interesting dialectical reason noted by Lewis22). There are countless relations that the causal theorist could, for all internal constraints can determine, be referring to when he or she uses the term 'cause'. Granted, causality is causally related to 'cause', but then the weird relation shmausality is shmausally related to the word 'cause'. What makes causality rather than shmausality the important relation here?23 It cannot be anything about us, so it must be something about the world. Putnam suggests that the best the causal theorist can do is "claim that a relation can at one and the same time be a physical relation and have the dignity (the built-in intentionality, in other words) of choosing its own name."24 This seems unjust to me. But what the causal theorist must do is to advocate the inegalitarianism Lewis discusses. The causal theorist must hold that some relations are less gerrymandered and bizarre than others, and that what singles out causation as the referent of 'cause' is that, of the relations which are otherwise possible referents, causation is the most natural. If this is correct then it is misleading to think of there being a choice between causality and objective similarity as the right external constraint. Objective similarity must be appealed to whether there is a further appeal to causation or not.

For my purposes the interest of this discussion of external constraints is that it looks as though any external constraint will require what Kolakowski calls "Aristotelean realism." In section I, I suggested that there were three possible metaphysical views: Aristotelean realism, which holds that the world has an organization of its own; modest realism, which holds that the world has numberless organizations of its own; and anti-realism, which holds that the world has no organization of its own. Kolakowski defends a kind of anti-realism by attacking Aristotelean realism, thus ignoring the attractions of modest realism. But now it appears as though the modest realist will have to hold that reference is indeterminate, that we never manage to select a single organization of the world for our own purposes.

Let us now leave Aristotelean realism behind. I do not see that Putnam has managed to show that it is untenable, but I also do not see how to advance its discussion further. Let us consider modest realism and Putnam's internal realism (i.e., anti-realism, or "idealism" in the broad sense I have been employing). Putnam's argument appears to be perfectly compatible with modest realism; the straightforward reading of the moral of the argument would be only that there is a surprising and disturbing indeterminacy in our reference to the real, mind-independent world. It is not the argument about reference alone, but that argument together with his rejection of indeterminacy that leads Putnam to his idealistic ontology. But why exactly does Putnam reject indeterminacy?

In Reason, Truth and History, the earliest of the publications I am considering here, Putnam explains that he will be exploring an alternative to Quine's view that "it is just an illusion . . . that the terms in our language have determinate well-defined counterparts."25 But in this book he does not, so far as I can tell, offer any reason for rejecting indeterminacy. He does note that indeterminacy is disturbing because "no such 'indeterminacy' rises in the 'notional world' of the speaker,"26 but he does not take this to be a reason to reject indeterminacy, for as he nicely observes, "the fact that in our belief system or 'notional world' no cat is a cherry means that in each admissible interpretation of that belief system . . . the referents of 'cat' and the referents of 'cherry' must be disjoint sets. But the disjointness of these sets is compatible with the (remarkable) fact that what is the set of 'cats' in one admissible interpretation may be the set of 'cherries' in a different (but equally admissible) interpretation."27

In Realism and Reason, Putnam does offer a criticism of indeterminacy. He writes of the view that reference is indeterminate that: "I cannot accept it for my own language, because to do so would turn the notion of an object into a totally metaphysical notion. I know what tables are and what cats are and what black holes are. But what am I to make of the notion of an X which is a table or a cat or a black hole (or the number three or . . .)?"28 Let us pause here for a moment before picking up the quotation again. Isn't this description of the view unfair? First of all, the advocate of indeterminacy does not hold that there is some object which is not determinately a cat or a table or whatever. It is not the objects themselves that are indeterminate, but rather which objects we are referring to. So perhaps Putnam should ask rather what to make of the notion of a word which refers to tables or cats or black holes, etc. But this also would be unfair. For no interpretation of English makes it true that there is a word which refers to tables or cats or black holes. The most we can say is that some interpretations assign to 'cat' the same referents that other interpretations assign to 'table'. And even in saying this we need to keep in mind that this is only possible as part of wide-ranging differences in the interpretation of other singular terms and predicates.

Let us now return to Putnam's criticism of indeterminacy. Immediately after the last-quoted sentence, Putnam writes: "An object which has no properties at all in itself and any property you like 'in a model' is an inconceivable Ding an sich. The doctrine of ontological relativity avoides the problems of medieval philosophy (the problems of classical realism) but it takes on the problems of Kantian metaphysics in their place." But again, the interpretation of the indeterminist position seems incorrect. The indeterminist does not deny that objects have properties in themselves. The indeterminacy is in our reference to objects and properties, not in the objects or properties themselves. We can, to be sure, say that the referent of 'table' on one interpretation is hard and wooden while on another interpretation it is soft and furry; but that is not to say that there is an object which in itself is neither hard and wooden nor soft and furry, only that it is indeterminate which object 'table' refers to. Moreover, even the point that the referent of 'table' has different properties on different interpretations misleadingly suggests more looseness than is actually present, since on any interpretation the referent of 'table' has the property expressed by 'hard and wooden', and on no interpretation does the referent of 'table' have the property expressed by 'soft and furry'.

The problems with Kant's metaphysics come from the problematic relation between the two worlds he talks about: the phenomenal, empirical world and the noumenal world of things in themselves. The view that reference is indeterminate does incorporate the Kantian view that there are things in themselves, that there are mind-independent objects in the world. But this is just to say that it incorporates realism. It also holds that reference is to things in themselves and that knowledge is of things in themselves, though it admits that reference is less deterministic than we would like. It thus avoids Kant's interposition of a made-up world between our experiences and the really real world. It thus also avoids, I would suggest, the problems of Kantian metaphysics. In the next section, I will suggest that Putnam's own view is actually more Kantian than the one he here rejects.

The paragraph of Putnam's I have just quoted and criticized is, so far as I know, the only place where he justifies his rejection of the view that reference is simply indeterminate--with the exception of one sentence in "Is the Causal Structure of the Physical Itself Something Physical?" In that paper he describes the view as "incoherent," explaining that it "requires us to believe in a world of things in themselves that have no determinate relations to our language."29 This certainly describes the view fairly. But why is it incoherent? It seems to me to be perfectly coherent, though admittedly rather unsettling.

I have now discussed two realist responses to Putnam's argument about reference. Aristotelean realism agrees that nothing internal can fix reference, but holds that the world itself is organized into kinds and thus that there can be external constraints on reference. It thereby defends the possibility of determinate reference to mind-independent objects. Modest realism denies that any of the possible classifications of the world are inherently preferable to others, and so holds that we refer indeterminately to mind-independent objects. I turn now to Putnam's own preferred view.



Putnam's own view about reference, I take it, is that we refer determinately to mind-dependent objects. In Reason, Truth and History, after mentioning the view that reference is simply indeterminate, Putnam says that he "will explore the alternative . . . of giving up the idea . . . that words stand in some sort of one-one relation to (discourse-independent) things and sets of things."30 This is puzzling at first, since this is precisely what the view that reference is indeterminate does as well. But while indeterminism gives up the one-one relation, Putnam's internal realism gives up the discourse-independence: "a sign that is actually employed in a particular way by a particular community of users can correspond to particular objects within the conceptual scheme of those users. 'Objects' do not exist independently of conceptual schemes. . . . Since the objects and the signs are alike internal to the scheme of description, it is possible to say what matches what."31

I am not at all sure I understand this. I am not sure what it is for an object, as opposed to a description, to be within a conceptual scheme. But I take it internal realism is not supposed to do damage to the truths of common sense: it does not deny that there are objects that are separate from us and that we interact with causally. It is just that to some extent the character of the objects and relations is mind-dependent. But now it is tempting to ask how mind-dependence can help with the puzzle about reference. It is not at all clear that the argument that reference is indeterminate rests on any assumptions about the mind-independence of objects. Suppose that tables and cats and the rest are somehow mind-dependent. Isn't it still the case that talk that uses the word 'table' can be interpreted consistently either as talk about tables or as talk about cats? How does the mind-dependence of tables and cats help fix which one 'table' refers to?

On Putnam's view we ordinarily use terms to refer to objects that are mind-dependent. But Putnam's argument about reference does not appear to violate ordinary usage; it seems to use words in the same old familiar way, though it manages in doing so to arrive at startling conclusions. But if we ordinarily use words to refer to mind-dependent objects, and Putnam's argument uses words in the ordinary way, then it seems that the argument shows that reference to mind-dependent objects is indeterminate.

Perhaps it misrepresents Putnam's position to say that he holds that our terms correspond determinately to mind-dependent objects. He suggests that the whole notion of correspondence needs to be overthrown. When he offers a positive account of reference, he advocates a kind of disquotational theory of reference according to which "we understand 'refers to' not by associating the phrase 'refers to' with a 'correspondence', but by learning such assertability conditions as the following: . . . 'Cat' refers to an object X if and only if X is a cat."32 He goes on to explain that "reference . . . is not something prior to truth; rather, knowing the conditions under which sentences about, say, tables, are true is knowing what 'table' refers to."33

It is not altogether straightforward to locate the difference between Putnam and the defender of indeterminacy here. The defender of indeterminacy will also say that knowing the truth conditions of sentences containing 'table' is all there is to know about reference. For the indeterminist, there are no facts about reference beyond those fixed by our use of the language; this is what distinguishes the indeterminist from the Aristotelean realist. But, the indeterminist will point out, precisely what the argument about reference shows is that truth conditions of sentences containing 'table' are not enough to fix a unique referent for 'table'. Many different referents for 'table' will make the same sentences come out true, including the sentence "'table' refers to tables." It is precisely because knowing truth conditions is knowing all there is to know about reference that reference is indeterminate.

I have two suggestions as to what distinguishes Putnam from the indeterminist realist, neither of them very satisfactory. The first is that Putnam really wants to deny that there is any such relation as reference between words and things. 'Refers' is merely a device of semantic descent, so that ultimately it only relates words to other words: words with quotation marks around them to words without quotation marks. The relation between 'table' and the world cannot be indeterminate because there is no such relation. This view would have the consequence that, at least at the level of individual terms, language is not about the world at all. But if this were Putnam's view then he ought not to talk about the mind-dependence of objects; he ought instead to insist that talk about objects simply makes no sense.

This brings us to the second suggestion about what distinguishes Putnam from the indeterminist. Perhaps Putnam means to simply make it true by definition that reference is determinate. Perhaps he simply intends to use the word 'object' in such a way that knowing the truth conditions of sentences does suffice to determine what objects those sentences are about. (But wouldn't this establish the determinacy of reference in the same way that defining 'God' as 'omnipotent omniscient omnibenevolent existing person' establishes the existence of God?) An "object" in this sense would perhaps be whatever is common to the our uses of 'table' on any interpretation that meets the available constraints: the word 'table' itself and perhaps associated descriptions and experiences. Talk about objects would then really be talk about words and descriptions and experiences. Or rather, since we surely need to preserve some sort of distinction between purely linguistic and mental phenomena, on the one hand, and objects, on the other, we would perhaps make up a kind of imaginary world, a notional or phenomenal world, and treat reference as reference to this phenomenal world. On this account objects would be mind-dependent in a very strong sense. But also on this account the determinacy of reference would be a kind of pretense, either pretended determinate reference to real things or really determinate reference to pretend things. There would still be real, mind-independent objects in the world, and the fact that we could not determinately refer to them would still be troubling.

The unsatisfactoriness of both of these suggestions is evidence that neither of them is what Putnam has in mind. I wish I had a better suggestion to offer. My guess is that the second suggestion comes closer than the first. But if anything resembling the second suggestion is correct, than Putnam's internal realism is very similar indeed to Kant's transcendental idealism. For it introduces precisely the Kantian contrast between the phenomenal world which we can know (refer to) precisely because we largely make it up ourselves, and the mind-independent noumenal world to which we have no (determinate) access.

Making up a phenomenal world can hardly make the things themselves go away. And Putnam himself occasionally verges on acknowledging that there must be mind-independent objects behind the phenomenal ones: "I am not inclined to scoff at the idea of a noumenal ground behind the dualities of experience, even if all attempts to talk about it lead to antinomies,"34 and again: "Today the notion of a noumenal world is perceived to be an unnecessary metaphysical element in Kant's thought. (But perhaps Kant is right: perhaps we can't help thinking that there is somehow a mind-independent 'ground' for our experience even if attempts to talk about it lead at once to nonsense.)"35 So Putnam himself is enmeshed in the Kantian distinction between appearances and things in themselves. We have seen Putnam object to indeterminism that "it takes on the problems of Kantian metaphysics." But it now appears that this criticism may be applied with at least equal justice to internal realism itself.35



1. Ralph Barton Perry, Present Philosophical Tendencies (New York: Longmans, Green and Co., 1912; Third Impression, Revised, 1916), p. 114. Perry meant this characterization to cover Berkeley, Kant, and Hegel, among others; I take it that his attempt to state what they have in common would meet with wide agreement. More recently, Thomas Nagel has suggested that the "common element" in a variety of versions of idealism, including both historical versions and the views of such contemporaries as Davidson, Strawson, and Wittgenstein, is "a broadly epistemological test of reality." Nagel, The View From Nowhere (New York: Oxford University Press, 1986), p. 91.

2. Hilary Putnam, Reason, Truth and History (Cambridge: Cambridge University Press, 1981), p. 54.

3. Reason, Truth and History, p. xi.

4. Hilary Putnam, Realism and Reason, Philosophical Papers Volume 3 (Cambridge: Cambridge University Press, 1983), papers 6 and 7.

5. Realism and Reason, papers 12 and 16; "Is the Causal Structure of the Physical Itself Something Physical?" Midwest Studies in Philosophy volume 9 (1984), 3-16.

6. Reason, Truth and History, especially chapters 6, 7, and 9.

7. Reason, Truth and History, p. 52.

8. Reason, Truth and History, pp. 53-4.

9. Realism and Reason, p. xvi.

10. Leszek Kolakowski, "Karl Marx and the Classical Definition of Truth," in Kolakowski, Toward a Marxist Humanism, translated from the Polish by Jane Zielonko Peel (New York: Grove Press, 1968).

11. Kolakowski, p. 46.

12. Kolakowski, p. 52. Here I quote the slightly different version of this paper published in Kolakowski's Marxism and Beyond (p. 73). The version in Toward a Marxist Humanism attributes to Marx "the idea that nature is composed of separate parts and that species are an 'artificial creation'," which changes the meaning in a way that seems contrary to Kolakowski's intention.

13. Kolakowski, p. 48.

14. Kolakowski, p. 57.

15. Kolakowski, p. 54.

16. Putnam, "Models and Reality," in Realism and Reason.

17. See especially Quine's "Ontological Relativity," in Quine, Ontological Relativity and Other Essays (New York: Columbia University Press, 1969), pp. 26-68.

18. Reason, Truth and History, p. 33.

19. David Lewis, "Putnam's Paradox," Australasian Journal of Philosophy vol. 62 (1984), 221-36, at 228. Lewis credits the suggestion to G. H. Merrill, "The Model-Theoretic Argument Against Realism," Philosophy of Science vol. 47 (1980), 69-81.

20. Compare Realism and Reason, p. xi.

21. See e.g. Michael Devitt, "Realism and the Renegade Putnam: A Critical Study of Meaning and the Moral Sciences," Nous vol. 17 (1983), 291-301, at 298-9; Devitt, Realism and Truth (Oxford: Basil Blackwell, 1984), pp. 188-91.

22. Lewis, 225-6.

23. Compare Putnam, Realism and Reason, p. 296; and "Is the Causal Structure of the Physical Itself Something Physical?" 6-7.

24. "Is the Causal Structure of the Physical Itself Something Physical?" 7.

25. Reason, Truth and History, p. 41.

26. Reason, Truth and History, p. 44.

27. Reason, Truth and History, p. 44.

28. Realism and Reason, p. xiii.

29. "Is the Causal Structure of the Physical Itself Something Physical?" 6.

30. Reason, Truth and History, p. 41.

31. Reason, Truth and History, p. 52.

32. Realism and Reason, p. xv.

33. Realism and Reason, p. xvi; cf. Reason, Truth and History, p. 52.

34. Realism and Reason, p. 226.

34. Reason, Truth and History, pp. 61-2.

35. Work on this paper was supported in part by a National Endowment for the Humanities grant to the Trinity University Philosophy Department for a curriculum development project in the history of philosophy.