Real realism: The galilean strategy

Philosophical Review

Volumn 110, Issue 2, 2001, Pages 151-197

  Kitcher, Philip ^a  

^a Howard Hughes Medical Institute   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 1842830148     PISSN: 00318108     EISSN: 15581470     Source Type: Journal    
DOI: 10.1215/00318108-110-2-151     Document Type: Review

References (111)

1. So, for example, a realist might adopt a different attitude towards the ψ function in the Schrödinger formulation of quantum mechanics and towards the molecules discussed in molecular genetics. Both Michael Gardner's "Realism and Instrumentalism in Nineteenth Century Atomism," Philosophy of Science 46 (1979): 1-34
2. and Richard Miller's Fact and Method (Princeton: Princeton University Press, 1987) are admirably clear that realists campaign for a license, but that their uses of that license are sensitive to details of the scientific situation.
3. This terminology is due to Richard Boyd (see, for example, "Realism, Approximate Truth, and Philosophical Method," in Scientific Theories, ed. C. Wade Savage (Minneapolis: University of Minnesota Press, 1990), 355-91).
4. The term is well suited to some who articulate the second antirealist theme (for example, Hilary Putnam, Nelson Goodman, Thomas Kuhn) but not to others (Richard Rorty, Arthur Fine) who strenuously reject what they see as a bad metaphysics of "constructivism" (see, for example, Fine, "Science Made Up," in The Disunity of Science, ed. Peter Galison and David Stump (Stanford: Stanford University Press, 1996), 231-54). But Boyd's term does mark an important distinction, and I'll employ it in this essay, subject to the caveat just noted.
5. I don't mean to suggest that this common family of arguments exhausts the arsenal of antirealists. As I'll concede explicitly in section 9, further challenges remain. But I hope to address some lines of attack that have been influential, and to prepare the way for taking up those further challenges.
6. In fact, as many commentators have noted, Fine's own position is not entirely clear. At times he writes as though he were contributing to the empiricist/realist debate, for example, in many of the discussions that run through the first two sections of 'The Natural Ontological Attitude" (chapter 5 of The Shaky Game (Chicago: University of Chicago Press), 1986).
7. The dominant trend of NOA, however, seems to be a democratic one. Matters are further complicated by occasional remarks in which Fine seems to concede that he thinks of truth in the familiar sense that a statement "is true just in case the entities referred to stand in the referred to relations" (ibid., 130).
8. These inspire interpreters to welcome Fine as a realist (see, for example, Alan Musgrave, "NOA's Ark: Fine for Realism," Philosophical Quarterly 39 (1989): 383-98). Despite the insights and wit of Musgrave's essay, it seems to me that Fine isn't so easily assimilated, in part because of his reluctance to embrace the characteristically realist idea of reference as a relation between language and mind-independent objects. I'll explore this issue in detail below.
9. I owe to the editors of the Philosophical Review the suggestion that this is unnecessarily cautious and that the NEA can be traced to the ancients. That suggestion may well be correct. My hesitation is based on the possibility that attributing to ancient or medieval thinkers a concept of action-guiding representational states may be an anachronism.
10. Following Barbara von Eckardt's ecumenical survey, I take contemporary cognitive science to suppose that people have psychological states that both express content and are governed by computational rules (see her What is Cognitive Science? (Cambridge: MIT Press, 1993), especially 191).
11. In some studies, notably those of Susan Carey (Conceptual Change in Childhood (Cambridge: MIT Press, 1985))
12. and Frank Keil (Concepts, Kinds, and Conceptual Development (Cambridge: MIT Press, 1989)), the scientific discussion seems continuous with our everyday attributions of states and contents and our everyday understandings of semantic relations.
13. Broader ventures in cognitive science-for example, those of John Anderson (The Architecture of Cognition (Cambridge: Harvard University Press, 1983))-sometimes posit a wider variety of states, but retain connections between some of these states and the notions of folk psychology.
14. There are, perhaps, some programs in cognitive science that appear at odds with folk concepts: see, for example, Ray Jackendoff's Languages of the Mind (Cambridge: MIT Press, 1992)
15. or Ronald Langacker's Foundations of Cognitive Grammar (Stanford: Stanford University Press, 1987), but even in these instances, it seems to me that we can discern the kinds of states and semantic relations that my discussions presuppose. Nonetheless, it should be conceded that the degree to which cognitive science will preserve everyday idioms is currently controversial, and there are possible developments of the subject that could undermine both the claim of this sentence and the broader strategy of the essay.
16. It's worth being explicit about the relation between real realism and NOA. As pointed out in note 4, one difference is that the realist thinks we can make sense of a notion of mind-independence and use it to develop a more substantive notion of truth than Fine would allow. Real realists also believe that Fine's "homely" argument for acquiescing in science is weak (on an uncharitable reading, the NOAer appears at least uncritical if not credulous) and that, in consequence, the brief for democracy fails. As we'll see, real realists agree that their views should be supported by "homely" considerations ('The Natural Ontological Attitude," 126) but believe that Fine has mislocated them.
17. See J. L. Austin, Sense and Sensibilia (Oxford: Oxford University Press, 1962)
18. George Pitcher, A Theory of Perception (Princeton: Princeton University Press, 1971)
19. and Jonathan Bennett, Locke, Berkeley, Hume: Central Themes (Oxford: Oxford University Press, 1971), chap. 3.
20. Tor the latter case, see Nelson Goodman, Ways of Worldmaking (Indianopolis: Hackett, 1978), where the experiments of Paul Kolers are used in defense of constructivism.
21. See, for example, Thomas Kuhn, The Structure of Scientific Revolutions (Chicago: University of Chicago Press, 1962/1970), 206;
22. Hilary Putnam, Reason, Truth and History (Cambridge: Cambridge University Press, 1981), chaps. 1-3, esp. 49;
23. Richard Rorty, Objectivity, Relativism, and Truth (Cambridge: Cambridge University Press, 1991), esp. 6-7;
24. Arthur Fine, The Shaky Game, 116, 129.
25. Here I am again indebted to the editors.
26. Galileo Galilei, The Assayer. See Stillman Drake, ed., Discoveries and Opinions of Galileo (New York: Doubleday, 1957), 274.
27. It's possible to argue that contemporary understanding of the neurophysiology of color perception reinforces the antirealist cause by undermining the reductivist project. But, as Peter Bradley and Michael Tye have shown, in currently unpublished work, the details of the neuroscience can be exploited to identify more subtie reflectance properties of surfaces, and thus to generate a more sophisticated reduction.
28. Perhaps another would be to concede the point for secondary qualities and argue that only primary qualities matter to science, a view popular with some historical figures. This reply seems to me weaker, both because it raises the possibility that analogues of the skeptical arguments might be generated against some of the primary qualities and because of the apparent role that secondary qualities play in some areas of science (for example, in evolutionary biology).
29. Of course, an immense amount of ink has been spilt on questions about the status of secondary qualities, from the early modern period to the present, and there are many versions of both strategies I mention here, as well as arguments for and against them. I believe, however, that both strategies remain promising.
30. For a similar assessment, see Hilary Putnam, Renewing Philosophy (Cambridge: Harvard University Press, 1992), 91-97.
31. In The Quest for Reality (Oxford: Oxford University Press, 2000), Barry Stroud shows with great subtlety how difficult it is to formulate and to resolve the questions in this area. I am much in sympathy with the treatment he offers, particularly with his clear dissection of the idea that early modern science "unmasked" everyday views of the reality of colors.
32. Our ability to represent continuity is notoriously tricky. See, for example, Charles Parsons's classic essay "Infinity and Kant's Conception of the 'Possibility of Experience'," Philosophical Review 73 (1964): 182-97.
33. In the setting of contemporary science, quantum mechanics has been the main source of challenges to realism derived from the possibility of scientific corrections of our everyday picture of reality. Here too, I would claim, piecemeal realism need not fear the limited revision that may be required, but, in accordance with the general aims of this essay, I won't elaborate the point here.
34. As we shall see, this form of argument is very intricate. Versions of it appear in Goodman (Ways of Worldmaking) and Putnam (Reason, Truth and History).
35. Here I've given two variants on the same argument, one which has deep roots in the history of philosophy (proceeding via (4a)), and one that is more recent (proceeding via (4b)), surfacing in the writings of Kuhn and Laudan. I suspect that the argument via (4b) may articulate more fully the appeal to variation in belief that leads historians and historically-minded sociologists of science to oppose realism.
36. I cover some familiar territory here because it prepares the way for the general realist strategy that I'll advocate.
37. They were particularly concerned to show that the principal theoretical developments of twentieth-century physics were intelligible (and, to a lesser extent, to defend the constructs of psychological theorizing). For a thorough charting of the history, see Frederick Suppe's introduction to The Structure of Scientific Theories (Urbana: University of Illinois Press, 1977).
38. Thus, a set E is an admissible extension for a theoretical predicate T just in case there's a model of the first-order formalization of the theory in which T occurs in which all observational terms receive their standard extensions and T is assigned E. A theoretical predicate is fully interpreted if it has a unique admissible extension, partially interpreted if it has more than one admissible extension. Two points in this formulation deserve note. First, it should be clear that this is a reconstruction of a notion of partial interpretation that evolved over two decades (from the late 1930s to the early 1960s). Second, there may be tacit acceptance of further restrictions on acceptable partial interpretations. Thus, it seems likely that the domain is restricted to mathematical entities and physical objects, and that there are conditions on the extension of theoretical predicates (for example, that an n-place predicate be assigned a set of n-tuples whose kth members are physical objects);
39. it's already demanded explicitly, of course, that the extensions of the observational predicates be their usual ones.
40. So the normal form of a definition would specify a new predicate as equivalent to some predicate formed by conjoining and disjoining observational predicates;
41. unions and intersections of sets containing only observables will always contain only observables, so we seem to be able to support (SE3);
42. obviously, maintaining that all introduction of meaningful vocabulary must take this form is extremely restrictive.
43. The mistake was first pointed out by Putnam in a seminal essay, "What Theories Are Not" (reprinted in Putnam, Philosophical Papers, vol. 1 (Cambridge: Cambridge University Press, 1975), 215-27).
44. Of course, there is a famous Wittgensteinian puzzle about how this is done: see Philosophical Investigations (Oxford: Blackwell, 1958), §§143-42
45. and Saul Kripke, Wittgenstein on Rules and Private Language (Cambridge: Harvard University Press, 1982).
46. Without attempting to say how we do it, I shall assume that we are able to project from a Finite range of instances. Other Wittgensteinian concerns about the limits of our power to project have been raised by Sklar in "Saving the Noumena" and "Semantic Analogy" (both reprinted in Sklar, Philosophy and Space-Time Physics (Berkeley: University of California Press, 1985).
47. The example is Putnam's, in "What Theories Are Not."
48. Both arguments have a long pedigree and both are prominent in contemporary discussions of realism. Version A, revived by van Fraassen, is akin to considerations that Mach urged against realist interpretations of atomic chemistry (even though, as noted, van Fraassen repudiates Mach's views about perception);
49. version B, which seems to underlie some of Laudan's writings, recapitulates themes of Poincaré. It should be noted, however, that Laudan might not view his argument as making any contribution to empiricism. He describes his position as "pragmatist" see Science and Relativism (Chicago: University of Chicago Press, 1990), and there are reasons to think he would question the distinction between claims about observables and claims about unobservables that empiricists deploy.
50. This idea-the "ultimate argument" for realism-became popular in the 1970s, largely because of the formulations of J. J. C. Smart, Putnam, and Boyd. It has deep historical roots. Descartes, for example, defended parts of his physics on the grounds that his successes would be incomprehensible unless his theoretical hypotheses were right.
51. The latter challenge is formulated by Laudan ("A Confutation of Convergent Realism," Philosophy of Science 48 (1981): 19-48). It seems to me that the challenge can be deepened by placing it in the context of the first question, since a liberal criterion of "success" would expand the range of relevant historical examples. The discussion that follows extends the response to Laudan begun in The Advancement of Science.
52. In both instances these formulations are not entirely precise. The first is intended to allow for "predictions about the past," for example, paleontological statements that identify the likely whereabouts of as yet undiscovered fossils. The second is vague about how complete the instructions have to be (it would be folly to insist that they specify a mechanical procedure, for example) . Attending to the niceties of precise formulations would divert us from the main points of the present discussion.
53. Laudan provides a long list in "A Confutation of Convergent Realism." His list includes such things as the humoral theory of medicine, catastrophist geology, and theories of spontaneous generation. In none of these examples does it seem right to hail the theory as successful or to judge that the proponents of the theory were justified in holding it to be successful. Other doctrines on his list are more intractable.
54. Fresnel submitted a memoir presenting his theory in a prize competition sponsored by the French Academy of Sciences. One of the jurors, Poisson, a proponent of the corpuscular theory of light, recognized that Fresnel's principles implied that there would be a bright spot at the center of the shadow of a small, rotating disk, thinking that this would provide a reductio of the theory. Another juror, Arago, performed an experiment to check, and discovered the bright spot. It is a nice irony that it is named for Poisson, who had been so confident that it did not exist.
55. This proposal is lucidly elaborated by John Worrall ("Structural Realism: The Best of Both Worlds?" Dialectica 43 (1989): 99-124). As Worrall notes, the position develops insights of Poincaré.
56. This approach depends on developing ideas about reference originally formulated by Putnam, Saul Kripke, and Keith Donnellan. For details, see my 'Theories, Theorists, and Theoretical Change," Philosophical Review 87 (1978): 519-47.
57. I defend this interpretation in chapter 5 of The Advancement of Science, see, in particular, 145-48.
58. I suspect that scientific efforts to draw the distinction are relatively rare, primarily because scientists are more interested in extending theory and devising experiments than they are concerned with its "foundations."
59. I fell into this trap in The Advancement of Science. I am grateful to several reviewers (especially Jarrett Leplin, Carl Matheson, and Richard Miller) who pointed out the mistake.
60. This strategy has been pursued most vigorously by Richard Boyd in a series of articles, beginning with "Realism, Underdetermination, and the Causal Theory of Evidence," Noûs 7 (1973): 1-12). Boyd has seen very clearly that realists must avoid the trap of taking realism to be the default position and simply clearing away objections of the kind leveled in variation B.
61. The most prominent example for developing the transcendental strategy starts from Goodman's new riddle of induction, claiming that justified views about what predicates (or classes) are projectible must advert to hypotheses about underlying mechanisms. So, for example, geologists might legitimately project a claim about certain minerals because they believe that the minerals in question share a common microstructure that gives rise to their observable properties. Without this background belief, they would not be warranted in taking the predicates they employ to be projectible, and thus would have no justification for the commonplace types of induction that empiricists accept. For development of points like this, see Boyd "Realism, Approximate Truth, and Philosophical Method."
62. One obvious worry is that antirealists will be able to advance some other account of the justifiability of such things as projectibility judgments-as for example in Goodman's own suggestion that projectibility is measured by entrenchment (see the final chapter of Fact, Fiction, and Forecast).
63. Something like the Galilean strategy figures in Richard Miller's defense of realism (Fact and Method). It's also prefigured in Jerrold Aronson's arguments for realism (see 'Testing for Convergent Realism," British Journal for the Philosophy of Science 40 (1989): 255-59
64. and Realism Rescued (London: Duckworth, 1994), 194-96.
65. I don't pretend that all realists are either transcendentalists or Galileans. Jarrett Leplin's recent brief for realism (A Novel Defense of Scientific Realism (Oxford: Oxford University Press, 1998)) urges that a realist view of science is supported because of the ability of theories to generate novel evidence. Unfortunately, Leplin's notion of novelty appears flawed. He proposes that the novelty of evidence isn't a psychological matter, depending on what the proposer of the theory thought of in coming to believe it;
66. rather E*is novel evidence for T only if T already had an episte mological basis E that doesn't include E*. Yet if there's an epistemological basis for the theory of any kind it appears either that realists have exactly what they want-to wit, evidence for believing that the theory is true, or that the epistemological basis only allows for acceptance of the theory as empirically adequate. If the former, then the question has been begged against the anti-realist;
67. if the latter, then Leplin needs to explain why an extra scrap of evidence beyond that required for acceptance as empirically adequate does more than simply increase the degree of rational confidence that the theory is empirically adequate.
68. See Albert van Helden, The Invention of the Telescope (Chicago: University of Chicago Press, 1985)
69. and van Helden, ed., Galileo's Siderius Nuncius (Chicago: University of Chicago Press, 1989). As a number of historians have made very clear, Horky's fervent opposition to Galileo was viewed by his contemporaries as dishonest.
70. One can appreciate the importance of this line of reasoning in making Galileo's case by noting that, twenty years later, in his summary of the evidence for Copernicanism, he devotes much of the third day of the Dialogue Concerning the Two Great World Systems (Berkeley: University of California Press, 1967) to showing why the novelties are genuinely "beyond the moon," in the process crafting a sophisticated incipient theory of error.
71. The telescope was accepted remarkably quickly. Virtually all scholars admitted its reliability by 1615. Some of the early opposition was based on the difficulties of using telescopes, which yielded inconsistent results that could be exploited by Horky and others. Galileo worked both to improve the instruments and to offer guidance for their use, so that, within a year of the initial development, the major inconsistencies were resolved. (For valuable studies of the telescope and its validation, see the writings of van Helden.)
72. Putnam's important insight recalls M. Jourdain: we had been talking about unobservables all along. In an important respect, the Galilean strategy is thoroughly Wittgensteinian, recommending that one should "bring words back from their metaphysical to their everyday use" (Philosophical Investigations, §116).
73. As we shall see, this is the basis on which to respond to the Fine-Rorty charge that realist theorizing about the sciences involves tacking on dubious metaphysical assumptions, and it is also the source of pertinent concerns about realism, to wit, that realists try to extend the everyday usage into areas in which the enabling conditions of the "everyday uses" do not apply; see section 9.
74. It's possible that an epistemological empiricist might object that I've mis-described the situation. Galileo didn't show that conclusions about unobservables could be justified through use of telescopes. Instead he changed the boundaries of the observable. But to make this response would undermine the empiricist strategy. For the notion of the observable has now been detached from the connection with unaided observation, and must now come to something like "detectable with a reliable instrument." Realists can sympathize with that notion of observability, formulating their views as claims about the reliability of particular detection devices. If those claims are to be debated by the epistemological empiricist by anything like the line of reasoning we're considering, then the empiricist will have to rely on a more basic level of observability to serve as the ultimate standard of validation. The Galilean achievement will then be seen in terms of justifying conclusions about entities that are unobservable according to this more basic conception.
75. The latter version is suggested by van Fraassen's discussion of related forms of inference, especially in his example of the mice in the wainscoting. See The Scientific Image, 19-21.
76. Epistemological modesty is eloquently defended by van Fraassen (The Scientific Image, chap. 2). Thomas Nagel makes a criticism similar to mine in The View From Nowhere (Oxford: Oxford University Press, 1986), 92.
77. The point was beautifully made by Grover Maxwell in ('The Ontological Status of Theoretical Entities," Minnesota Studies in the Phibsophy of Science, vol. 3 (Minneapolis: University of Minnesota Press, 1962), 3-27).
78. Van Fraassen responds to one version of Maxwell's challenge (The Scientific Image, 13-19);
79. here I'm advocating another.
80. My gloss on systematicity here is hardly precise. For an attempt to do somewhat better, see my "Explanatory Unification," Philosophy of Science 48 (1981): 507-31.
81. There's irony here in that the reply to Semantic Empiricism built on insights of Putnam, who is one of the primary sources of the Semantic Constructivist argument.
82. It's worth noting at this point that I am relying on a common way of replying to a notorious argument of Berkeley's to the effect that we cannot conceive of an object existing unperceived. As many commentators have pointed out, Berkeley's mistake is to suppose that when we think about something, we must include ourselves "in the picture" (see, for example, Principles of Human Knowledge, §23). Similarly, I suppose that we can contemplate the possibility of physical objects existing without supposing that we have to be observers, parts of the imagined scene.
83. Forceful presentations of the point have been offered by Alvin Goldman (Epistemology and Cognition (Cambridge: Harvard University Press, 1986), chap. 7)
84. and Alan Musgrave, "NOA's Ark-Fine for Realism";
85. see, in particular, 53-54 of Musgrave's essay.
86. As we saw in section 3, realists ought to avoid the notorious "veil of perception" doctrine, according to which we perceive objects indirecly by perceiving sensa directly; the right thing to say is that we perceive the objects by being in (or "having") certain kinds of sensory states. The constructivist advocacy of (SC4) to accuse realists of engaging in metaphysics thus needs the kind of Austinian therapy I try to offer here.
87. This is how Putnam's celebrated arguments get started (see "Models and Reality," and the early chapters of Reason, Truth and History). Putnam proves a much stronger result than the standard conclusion that one can satisfy a set of sentences by permuting elements in the domain of the model; his construction reveals how the permutation can preserve the ways in which the truth values of sentences vary as the state of the world changes.
88. In responding to those, like Michael Devitt (Realism and Truth (Princeton: Princeton University Press, 1984)), who would insist on causal constraints on the assignment of referents, Putnam argues that the alleged causal constraints are just more sentences to be held true and thus fall under his permutation argument (Reason, Truth and History, chap. 2).
89. It should be noted that emphasizing the causal relations between the objects and the subject makes it look as though her language could be meaningful even in the absence of others, thus flouting the Private Language Argument (Wittgenstein, Philosophical Investigations, §§258ff.). But there's no need for realists to quarrel with this argument. The causal relations that fix meaning and reference can be taken to include relations among speakers. For the realist's purposes, the important point is not whether the causal constraints are so inclusive or whether they fix the referents for a group of speakers, but rather that they hold independently of the imaginary observer.
90. This part of the argument appeals to a point about causal explanation that goes back to Collingwood, a point that Putnam revives in his discussions (Reason, Truth and History, 51). It can be presented as the thought that attributions of causation always make selections from the complete set of causal antecedents and that the selection depends on our interests.
91. Dealing completely with all five objections would require a book or a series of essays. But I shall hope to say enough to demonstrate that realists have resources for answering the criticisms.
92. See Word and Object (Cambridge: MIT Press, 1960), chapter 2.
93. I am grateful to an anonymous referee for raising this worry.
94. See Jonathan Lear "Sets and Semantics," Journal of Philosophy 74 (1977): 86-102;
95. Lear applies Paul Benacerraf s famous dilemma about mathematical truth and knowledge to the case of the reference of mathematical terms (see Benacerraf, "Mathematical Truth, "Journal of Philosophy 70 (1973): 661-79).
96. Of course, those who hope for a fully naturalistic account of reference must either assume that mathematical terms are in principle dispensable or else come to terms with a more refined version of the semantic problem. My tentative preference would be to account for the terms of elementary arithmetic and geometry along the lines proposed in The Nature of Mathematical Knowledge (New York: Oxford University Press, 1983) although I would now adopt a fictionalist (Hilbertian) strategy for the higher reaches of mathematics.
97. For a forceful presentation of such worries, see Sklar, "Semantic Analogy," where the point is directed at the proposals made by Putnam ("What Theories Are Not").
98. See Paul Horwich, Truth, 2d ed. (Oxford: Oxford University Press, 1999), 22-23 and chapter 3.
99. Other important articulations of the position are provided by Michael Williams, "Do We (Epistemologists) Need a Theory of Truth?" Philosophical Topics 14 (1986): 223-42
100. and by Michael Levin, "What Kind of Explanation is Truth?" in Scientific Realism, ed. J. Leplin (Berkeley: University of California Press, 1984), 124-39.
101. Here I should note explicidy that the type of correspondence theory I envisage is one like that originally sketched by Hartry Field in "Tarski's Theory of Truth, "Journal of Philosophy 69 (1972): 347-75. One important difference, however, is that I don't see a naturalistic account of reference as necessarily involving a reduction to a physicalist vocabulary. Naturalists who are antireductionists with respect to some sciences (for example, parts of biology) should allow that the basis for a reduction of the notion of reference may outstrip the resources of physics.
102. The line of response I have offered is akin to that proposed by Anil Gupta, "A Critique of Deflationism," Philosophical Topics 21 (1993): 57-81, in particular note 30. I elaborate it in greater detail in "On The Explanatory Power of Correspondence Truth," forthcoming in Philosophy and Phenomenological Research.
103. This is doubly mistaken, for there is an important difference between Quinean underdetermination (roughly, cases in which rival theories are supposed to receive exactly the same support from the same body of evidence) and Kuhnian underdetermination (roughly, cases in which rival theories are successful in rather different ways). In the text, I suggest that Quinean underdetermination has varying force in different areas of science. I have tried to address issues of Kuhnian underdetermination in chapter 7 of The Advancement of Science.
104. In part, the bogey of underdetermination may stem from the view that scientific theories are readily formalized and that the formalization will automatically lead to ways of constructing serious alternatives. My quick appeal to a central doctrine within molecular biology is intended to reduce confidence in this view. I have attempted to address issues about underdetermination more thoroughly (but probably still not thoroughly enough) in Science, Truth, and Democracy (New York: Oxford University Press, forthcoming).
105. Rorty (Consequences of Pragmatism, xxvi) suggests that realists are committed to strong theses about the objective divisions of the natural world. Both Goodman (Ways of World-making) and Putnam (Reason, Truth, and History, 49) seem to agree.
106. So it seems to me that real realism can simply dodge Putnam's challenge to specify just how many objects there are in a particular (toy) world containing three individuals (see his "Truth and Convention," in Realism with a Human Face (Cambridge: Harvard University Press, 1994);
107. the challenge turns on whether we are to count mereological sums.
108. Here real realism diverges from stronger doctrines held by some realists (for example Richard Boyd and Michael Devitt) and from my own earlier self (as in chapter 4 of The Advancement of Science). An advantage of the Boyd- Devitt commitment to objective natural kinds may lie in facilitating the development of a detailed account of reference (see the earlier discussion in this section).
109. For an approach that seeks a causal account without any such commitment, see P. Kyle Stanford and Philip Kitcher, "Refining the Causal Theory of Reference for Natural Kind Terms," Philosophical Studies 97 (2000): 99-129.
110. Real realism is, I take it, strictly between NOA and Putnam's metaphysical realism. It is also quite distinct from Putnam's internal realism. Ernesto Garcia has, however, suggested to me that it may be consonant with views expressed in Putnam's recent work, particularly the return to Austin begun in The Threefold Cord: Mind, Body, and World (Columbia University Press, 1999).
111. That seems to me a possibility worth exploring, but the exploration must await another occasion.