Is goodness a homeostatic property cluster?

Ethics

Volumn 118, Issue 3, 2008, Pages 496-528

  Rubin, Michael ^a  

^a UNIVERSITY OF MASSACHUSETTS   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 48949107383     PISSN: 00141704     EISSN: None     Source Type: Journal    
DOI: 10.1086/533507     Document Type: Conference Paper

References (93)

1. I employ a convention of using small caps for terms referring to kinds and italics for terms referring to properties. I have found that treating properties as distinct from the kinds whose membership they define helps in the exposition of Boyd's view. However, nothing I say in this article depends upon kinds and properties actually being distinct sorts of entities.
2. John Stuart Mill, A System of Logic (New York: Harper & Brothers, 1867), 434.
3. See Richard Boyd, "Kinds, Complexity and Multiple Realization: Comments on Millikan's 'Historical Kinds and the Special Sciences,'" Philosophical Studies 95 (1999): 67-98, 81;
4. John Dupré, "Natural Kinds and Biological Taxa," Philosophical Review 90 (1981): 66-90, 68;
5. Philip Kitcher, "Species," Philosophy of Science 51 (1984): 308-33, 315n;
6. Joseph LaPorte, Natural Kinds and Conceptual Change (Cambridge: Cambridge University Press, 2004), 19;
7. W. V. Quine, "Natural Kinds," in Ontological Relativity and Other Essays (New York: Columbia University Press, 1969), 114-38, 126;
8. Bertrand Russell, Human Knowledge: Its Scope and Limits (New York: Simon & Schuster, 1948), 318.
9. Boyd, "Kinds, Complexity and Multiple Realization," 81.
10. Hilary Kornblith, Inductive Inference and Its Natural Ground: An Essay in Naturalistic Epistemology (Cambridge, MA: MIT Press, 1993), 35.
11. See Ruth Garrett Millikan, On Clear and Confused Ideas (Cambridge: Cambridge University Press, 2000), 15-32.
12. Richard Boyd, "How to Be a Moral Realist," in Essays on Moral Realism, ed. Geoffrey Sayre-McCord (Ithaca, NY: Cornell University Press, 1988), 181-228, 197.
13. Ibid.
14. See Richard Boyd, "Homeostasis, Species, and Higher Taxa," in Species: New Interdisciplinary Essays, ed. Robert A. Wilson (Cambridge, MA:MIT Press, 1999), 141-85, 153-54,
15. and "Kinds, Complexity and Multiple Realization," 79.
16. The real essences of kinds are contrasted with "nominal" essences. Roughly, the nominal essence of a kind, K, is something like an analytic definition that speakers conventionally associate with the predicate or kind term that corresponds to K. If a kind has only a nominal definition, then its membership conditions depend solely upon linguistic conventions and are discoverable by a priori conceptual analysis. By contrast, the real essence of a kind determines that kind's membership conditions independently of linguistic conventions and thus cannot be discovered by mere conceptual analysis (see Boyd, "How to Be a Moral Realist," 194-95,
17. and "Homeostasis, Species, and Higher Taxa," 142, 146;
18. and Brian Ellis, Scientific Essentialism [New York: Cambridge University Press, 2001], 32. Note that Boyd uses 'real essence' interchangeably with 'natural definition' and 'a posteriori definition').
19. An editor for Ethics recommends a different reading of Boyd with respect to the relationship between HPCs and the essences of natural kinds. On this alternative reading, HPCs are not themselves to be identified with the essences of natural kinds. Instead, the HPC correlated with a natural kind constitutes something like its "operational definition" that can be used to pick out some other property that is the kind's genuine a posteriori essence (or "natural definition"). While there are some passages in "How to Be a Moral Realist" that prima facie permit this alternative interpretation, other passages - especially in Boyd's later writings - strongly favor my preferred interpretation, according to which the HPC just is the essence or natural definition of the relevant natural kind. For instance, Boyd writes: "I conclude that individual species have (homeostatic property cluster) essences, so that a form of 'essentialism' is true for species" (Boyd, "Homeostasis, Species, and Higher Taxa," 142); "there are a number of scientifically important kinds . . . , biological species among them, whose natural definitions are very much like the property-cluster definitions postulated by ordinary-language philosophers except that the unity of the properties in the defining cluster is mainly causal rather than conceptual" (Richard Boyd, "Kinds as the 'Workmanship of Men': Realism, Constructivism, and Natural Kinds," in Rationalität, Realismus, Revision: Proceedings of the Third International Congress, Gesellshaft für Analytische Philosophie, ed. Julian Nida-Rümelin [Berlin: de Gruyter, 1999], 52-89, 67;
20. cf. Boyd, "How to Be a Moral Realist," 196); "Species are defined, according to the HPC conception, by those shared [phenotypic] properties and by the mechanisms . . . which sustain their homeostasis" (Boyd, "Kinds, Complexity and Multiple Realization," 81, emphasis in the original). For additional passages that favor my interpretation, see n. 22 below.
21. The notion that natural kinds are defined by clusters of properties can be found in Mill's A System of Logic and Russell's Human Knowledge. Russell's own account strikingly anticipates Boyd's. Russell writes: "The essence of a natural kind is that it is a class of objects all of which possess a number of properties that are not known to be logically interconnected" (Human Knowledge, 317). His claim that the defining properties are not known to be logically interconnected suggests that he recognizes that their belonging to the kind's essence is not a matter of our linguistic conventions but rather is a matter of a nomological connection. Furthermore, Russell backs away from the claim that every member of a kind needs to share all of the kind-defining properties. Like Boyd, he accepts indeterminacy in the extensions of natural kind terms: "Assuming evolution, there must have been outlying members so aberrant that we should hardly know whether to regard them as part of the [intension] or not" (ibid., 443).
22. Boyd, "Kinds, Complexity and Multiple Realization," 82;
23. Kornblith, Inductive Inference, 92ff;
24. Russell, Human Knowledge, 318.
25. Boyd's HPC conception of natural kind real essences also departs from more traditional views insofar as it denies that the essences of natural kinds must be (i) "unchanging" and (ii) composed only of properties that are both "ahistorical" and (iii) intrinsic to the kind's members (Boyd, "Homeostasis, Species, and Higher Taxa," 146-47, 153-57;
26. contrast this with Ellis, Scientific Essentialism, 19-23).
27. Boyd, "How to Be a Moral Realist," 197,
28. and "Homeostasis, Species, and Higher Taxa," 143.
29. Boyd, "How to Be a Moral Realist," 197.
30. Not only is the possibility of imperfect homeostasis and extensional indeterminacy important for the plausibility of HPC definitions of biological species but also it plays a role in Boyd's defense of moral realism. Some have thought that the existence of actions whose moral status is irresolvable is best explained by the hypothesis that there are no moral facts (see, e.g., J. L. Mackie, Ethics: Inventing Right and Wrong [New York: Penguin, 1977], 37).
31. Boyd's HPC account of moral properties makes an alternative explanation possible. If moral terms designate HPC phenomena, then, as with species, we should expect instances where it is indeterminate whether or not an individual action or state of affairs falls within the extension of a given moral term. Thus, not only are such indeterminate cases not an embarrassment to an HPC conception of moral properties but they are predicted by it (Boyd, "How to Be a Moral Realist," 213).
32. Boyd, "Kinds, Complexity and Multiple Realization," 83-84.
33. Here and elsewhere I treat the relevant homeostatic mechanisms as properties. I do so in this case because the mechanism associated with the kind METAL is evidently something that has multiple instances. Although Boyd is not explicit about what the ontological status of the homeostatic mechanisms is supposed to be, his own examples are also of things that admit of multiple instantiations. In any case, nothing much here turns on whether we understand homeostatic mechanisms as properties rather than individuals.
34. Boyd offers these and other examples of homeostatic mechanisms unifying the properties of biological species in his "Homeostasis, Species and Multiple Realization," 165 (cf. Ernst Mayr, "What Is a Species, and What Is Not?" Philosophy of Science 63 [1996]: 262-77). I have added italics in keeping with my convention of italicizing property-referring terms.
35. Of course, given the evolution of biological species, the homeostatic unity of certain property clusters is bound to be disturbed over a long enough period of time. This implies that the constituents of a biological kind's HPC definition change over time. Boyd recognizes and accepts this consequence of his view. He writes, "the properties which determine the explanatory definition of a species (and, thus, the conditions for membership in it) may vary over time (or space), while it continues to have numerically the same definition" (Boyd, "Kinds as the 'Workmanship of Men,'" 68).
36. Boyd, "Homeostasis, Species, and Higher Taxa," 142, and "Kinds, Complexity and Multiple Realization," 81.
37. From here on, I will typically drop the adjective 'moral' from 'moral goodness' and 'the moral good'. Unless otherwise indicated, 'goodness' and 'the good' should be taken to refer to moral goodness and THE MORAL GOOD.
38. As evidence that Boyd means to identify moral goodness with an HPC of the sort that I am about to introduce, note that Boyd explicitly writes "the term 'good' in its moral uses refers to the homeostatic cluster property" (Boyd, "How to Be a Moral Realist," 205).
39. Note also that my reading of Boyd as identifying goodness with an HPC also accords with the way Nicholas Sturgeon - himself a supporter of the HPC conception of goodness - understands Boyd. He writes that Boyd thinks of "moral properties such as intrinsic goodness as homeostatic clusters of various natural features" (Nicholas L. Sturgeon, "Moore on Ethical Naturalism," Ethics 113 [2003]: 528-56, 550).
40. Boyd, "How to Be a Moral Realist," 203.
41. Ibid.
42. It should be clear from this list that I understand the properties that putatively compose goodness to be properties whose instances satisfy the various human needs (where, e.g., instances of being in love satisfy the need for love). One might be tempted to read Boyd as claiming instead that the properties that compose goodness are instances of a broader property, namely, having a need satisfied. I do not think such a reading could be correct. If it were, then there would be a single property that composes goodness: i.e., the property having a need satisfied. In that case, goodness could not be thought of as a cluster of properties.
43. Boyd, "How to Be a Moral Realist," 203.
44. Ibid., 204n.
45. It should be acknowledged that Boyd presents his particular account of the human goods and the homeostatic mechanisms that unify them as speculation. He is careful to note that the question of exactly which properties and mechanisms belong to the cluster that defines THE GOOD is amatter for empirical inquiry. The success of the HPC conception of goodness does not depend upon the correctness of precisely this list of goods and mechanisms (though Boyd believes that his characterization of the HPC is "close to the truth" [ibid., 202]). With the exception of the argument I offer in Sec. V.B, my arguments against Boyd's view can be directed against other HPC proposals of goodness with little or no modification.
46. In the original text, Boyd uses 'moral goodness' where I use 'the moral good'. I have modified this passage in order to preserve the symmetry between moral HPC kinds and biological and chemical HPC kinds. Thus, although he writes that moral goodness is defined by the cluster of goods and their homeostatic mechanisms, I take him to mean that the property moral goodness is constituted by this cluster and its mechanisms. In turn, the HPC moral goodness defines the kind THE MORAL GOOD.
47. Boyd, "How to Be a Moral Realist," 203.
48. Boyd evidently thinks of homeostatic consequentialism proper as a broader moral theory that includes the present HPC account of goodness as just one component. This larger view would presumably include a consequentialist account of moral obligation alongside the HPC conception of value. (Boyd discusses his consequentialist view of right action in greater depth in his "Finite Beings, Finite Goods: The Semantics, Metaphysics, and Ethics of Naturalist Consequentialism, Part II," Philosophy and Phenomenological Research 67 [2003], 24-47.) However, since he does not offer a distinct name for the theory of value he proposes, it will be convenient for our purposes to use 'homeostatic consequentialism' to denote only the HPC theory of value.
49. Just below, I explain why making a distinction between noninstrumental goodness and instrumental goodness is desirable for the homeostatic consequentialist. Although Boyd does not himself acknowledge the distinction in "How to Be a Moral Realist," fellow homeostatic consequentialist Nicholas Sturgeon attributes to Boyd the view that homeostatic consequentialism is an account of intrinsic goodness (Sturgeon, "Moore on Ethical Naturalism," 550). However, Sturgeon suggests that the kind of intrinsic goodness he has in mind is not "a property that depends only on the intrinsic, nonrelational properties of the things that have it" (ibid.). Because intrinsic goodness is sometimes thought to be just the sort of goodness that a thing has in virtue of its intrinsic, nonrelational properties, I prefer to label the sort of goodness that Sturgeon describes as "noninstrumental." I cannot here attempt a precise account of the distinction between instrumental and noninstrumental goodness. Perhaps a slogan will suffice: "a thing is noninstrumentally good if and only if it is good as an end, rather than good merely as a means to some other good thing."
50. For a useful discussion, see Shelly Kagan, "Rethinking Intrinsic Value," Journal of Ethics 2 (1998): 277-97. (Note, however, that Kagan takes the label 'intrinsic goodness' to apply both to noninstrumental goodness and to the kind of goodness a thing has in virtue of its intrinsic, nonrelational properties.)
51. Here are two additional considerations in support of this exegetical decision: First, as we saw earlier, Boyd represents himself as claiming that "the term 'good' in its moral uses refers to the homeostatic cluster property just described" ("How to Be a Moral Realist," 205). This way of representing homeostatic consequentialism is hard to square with the view that an act is good just in case it "tends to foster" the HPC of human goods. For if that were Boyd's view, he should have said that 'good' (or better, 'goodness') refers to the property of tending to foster the HPC just described. A second consideration concerns the sorts of items that Boyd cites as bearers of moral goodness in the passage cited above. The three kinds of items he cites are actions, policies, and character traits. However, on a standard consequentialist conception of morality, only states of affairs are taken to be the fundamental bearers of noninstrumental (or intrinsic) value; actions, policies, and character traits are typically understood to have only instrumental (or extrinsic) value. This gives us yet more reason to treat the passage cited above as describing instrumental moral goodness, a property that is distinct from the more fundamental noninstrumental moral goodness.
52. The most influential argument against "analytic" ethical naturalism is G. E. Moore's open question argument (see G. E. Moore, Principia Ethica [Cambridge: Cambridge University Press, 1903/1993], 66-69).
53. Another argument that receives less fanfare, though it is to my mind more persuasive, involves the claim that analytic ethical naturalism cannot make sense of the possibility of substantive moral disagreement between two speakers who subscribe to different moral standards (see Boyd, "How to Be a Moral Realist," 186-87;
54. R. M. Hare, The Language of Morals [New York: Oxford University Press, 1952], 49, 148-49;
55. and Moore, Principia Ethica, 62-64).
56. In addition, the abandonment of analytic ethical naturalism is sometimes motivated by skepticism about analyticity in general. (Boyd expresses such a skepticism in "How to Be a Moral Realist," 196.)
57. Boyd, "How to Be a Moral Realist," 199-212;
58. David O. Brink, Moral Realism and the Foundation of Ethics (Cambridge: Cambridge University Press, 1989), 166, 175;
59. Sturgeon, "Moore on Ethical Naturalism," 533-35.
60. Terence Horgan and Mark Timmons, "New Wave Moral Realism Meets Moral Twin Earth," Journal of Philosophical Research 16 (1990-91): 447-65.
61. For his own part, Boyd seems to agree that independent justification is needed. He writes that, if the naturalistic moral realist is to legitimately make use of the epistemological and semantic claims characteristic of synthetic ethical naturalism, then there need to be "good reasons to think that moral terms must possess natural [i.e., nonanalytic, a posteriori] rather than stipulative [i.e., analytic or nominal] definitions" (Boyd, "How to Be a Moral Realist," 210, cf. 201).
62. Ibid., 210.
63. There is probably some correlation between experiences of pleasure and a person's physical health. So we may have to grant that this particular episode of pleasure makes a causal contribution to the hermit's health, however slight.
64. In light of the discussion of Sec. II.B, it should be observed that an HPC "as a whole" may be instantiated by an individual even when some of the cluster's constituent properties are not instantiated by that individual. Even so, it should also be clear that an HPC itself is not instantiated in an individual that instantiates only one of its constituent properties; e.g., an individual's being striped is not sufficient for it to be properly classified as a TIGER.
65. There is some indication that a homeostatic consequentialist would be willing to bite the bullet here. Sturgeon allows that "nothing would have a property such as intrinsic goodness at all, given a radical enough breakdown" in a certain part of the HPC that constitutes goodness (Sturgeon, "Moore on Ethical Naturalism," 550). Since the Hobbesian world exhibits a breakdown of the HPC that constitutes goodness, it would appear that Sturgeon is willing to accept that the hermit's being pleased is not an instance of goodness. (However, Sturgeon is here speaking of a breakdown only in a specific part of the HPC. In particular, he is considering the breakdown in the part of the cluster that involves "the existence of purposive, valuing creatures somewhat like us" [ibid.]. The claim that a world without purposive, valuing creatures contains no goodness strikes me as far less controversial than the claim that a world that included such creatures would nevertheless fail to contain intrinsic or noninstrumental value, if there were a lack of homeostasis between the human goods.)
66. I do not claim that this is the best or most useful conception of rigid application. It is, however, the conception that the defender of HC2 needs in order to avoid the revised problem of isolated goods. As will be seen below, I think this conception of rigid application is defective. An arguably better conception can be found in Michael Devitt, "Rigid Application," Philosophical Studies 125 (2005): 139-65.
67. Unfortunately, Devitt's conception is of no help to HC2. Furthermore, I doubt that the more traditional notion of rigid designation could be deployed in the service of HC2 without making questionable assumptions. On the traditional view, a term, t, rigidly designates an entity, e, if and only if t designates e in every possible world in which e exists and t designates no other entities (Saul Kripke, "Identity and Necessity," reprinted in Naming, Necessity, and Natural Kinds, ed. Stephen P. Schwartz [Ithaca, NY: Cornell University Press, 1977], 67-101, 78, 79). As an initial difficulty, it isn't clear that predicates are the sort of items that can be rigid designators: if predicates (be they natural kind predicates or nominal kind predicates) designate their extensions, then none are rigid, since their extensions are different at different possible worlds. If they designate properties, then every meaningful predicate is a rigid designator. In that case, rigid designation marks no interesting distinction among predicates. This last consequence might be avoided if we suppose that only those predicates that designate sparse properties (or universals) rigidly designate. Assuming this restriction were defensible, it would still require some maneuvering to get this conception of rigid designation to do the work HC2 needs of it. For one thing, given the pluralistic assumptions of HC2, there is no one property that 'good' designates; there is a plurality. Thus, it is not true of 'good' that it designates an entity in every possible world in which that entity exists and designates no other entities. One solution is to suppose that 'good' designates the conjunctive property made up of all the different good-making properties. But now HC2 has no reply to the problem of isolated goods. As we saw, the hermit's episode of pleasure does not instantiate a conjunctive property that includes all the other putative good-making properties as constituents. What is needed instead is an account where 'good' designates a disjunctive property. Here we face more trouble. We have had to assume that rigidly designating predicates designate only sparse properties.
68. However, on familiar conceptions of sparse properties, disjunctive properties do not qualify as sparse (see, e.g., D. M. Armstrong, Universals and Scientific Realism, vol. 2, A Theory of Universals [Cambridge: Cambridge University Press, 1978], 19-23). Perhaps there is more that can be said that would make it plausible that rigid designation can do the work that HC2 needs it to do. I hope to have said enough to make it clear that rigid designation does not provide a quick or easy solution to the problem of isolated goods.
69. Of course an instance of a property like being quadrupedal might be a part of a larger state of affairs that constitutes some individual's being a tiger. But this does not mean that the state of affairs consisting in a given individual being quadrupedal itself has the property of being a tiger. In such a case, we should say instead that one and the same individual has the property of being quadrupedal and has the property of being a tiger.
70. In Sec. V.D I introduce putative examples of HPC social kinds. It should be noted here that even those kinds behave like the paradigm HPC kinds and not like the good. For instance, suppose that the property of keeping kosher is part of the cluster of properties that defines the social kind HASIDIC JEW. Some particular man's keeping kosher does not have the property of being a Hasidic Jew.
71. Boyd suggests that the predicates 'healthy' and 'is healthier than' express HPC phenomena (Boyd, "How to Be a Moral Realist," 198). It may be that my challenge (to find a paradigmatic example of an HPC that shares the structural features of goodness as understood by HC2) could be answered by developing a plausible HPC account of HEALTH. Unfortunately, Boyd says very little about what sorts of properties might compose the HPC that defines HEALTH. In the absence of a more detailed account, it is difficult to tell whether or not the example of HEALTH will help the homeostatic consequentialist meet the present challenge.
72. Of course, individual persons that are constituents of these states of affairs might have these properties. However, because states of affairs are the primary basic bearers of goodness, this is of little help. The current difficulty might be avoided by noting that persons themselves may reasonably be taken to be bearers of moral goodness. This reply helps if we are advancing some sort of HPC account of virtue. Still, I presume that Boyd and other homeostatic consequentialists want to say that things other than persons may be noninstrumentally morally good. If so, they are faced with the present difficulty.
73. See, e.g., Alfred Jules Ayer, Language, Truth and Logic (New York: Dover, 1952), 105;
74. Simon Blackburn, Spreading the Word (New York: Oxford University Press, 1984), 187ff.;
75. Hare, The Language of Morals, 91;
76. Mackie, Ethics, 38ff.;
77. and Charles L. Stevenson, Ethics and Language (New Haven, CT: Yale University Press, 1944), 336. I should add that, in selecting these particular properties as potentially definitive of THE GOOD, I was inspired by various nonnaturalist and constructivist analyses of 'good'.
78. A. C. Ewing expresses roughly the same thought in his Ethics (New York: Macmillan, 1953), 85. He suggests that the claim that what is good or right is what an impartial spectator would approve of "is equivalent to saying that something is good or right when it is approved by somebody who only approves what is really good or right."
79. Of the properties I have recommended for this alternative HPC proposal, the one exception seems to be the property of being pursued by rational beings. If one takes a Humean or instrumentalist view of rationality, then it will be at best a metaphysically contingent fact that rational beings pursue, e.g., fitting states of affairs. Thus, being pursued by rational beings may well be nomologically linked to the other properties mentioned. I doubt that this one exception can give the homeostatic consequentialist what he needs to get past the present worry. However if more properties of this sort could be found, and if there were a compelling case to be made that these other properties are indeed contingently clustered, then this worry could be put to rest. But new difficulties are likely to arise. My suspicion is that the sorts of properties that are needed here would consist primarily in various sorts of characteristic human responses to good states of affairs. If I am right about this, then an alternative HPC definition of THE GOOD that incorporated these properties would raise its own problems for Boyd's larger project of defending naturalistic moral realism. Briefly, a cluster definition involving such properties threatens to make moral goodness a response-dependent property. Response-dependent accounts of moral properties, however, are at odds with the robust sort of moral realism that Boyd means to defend.
80. Of course, the reliability of this inference requires the support of some contingent features of the observer's environment as well. For instance, there must not be too many things that look like rabbit ears but are neither attached to rabbits nor creatures that share many (but not too many) properties that are characteristic of rabbits. (Keep in mind, however, that in my example what you infer is not that this thing is a rabbit but rather that it has such and such morphology, anatomy, behavior, etc. If the creature should turn out to be a hare, these conclusions are every bit as correct as if it turns out to be a rabbit.)
81. To make matters worse, even if we could inductively infer the presence of some of these goods from others, it is not clear that it is reference to THE MORAL GOOD that facilitates these inferences. Our inductions may well turn out to be grounded by the properties that cluster around sporting activities qua sporting activities.
82. The first five examples are culled from Richard Miller, "Half-Naturalized Social Kinds," Philosophy of Science 67 (2000): S640-S652.
83. Boyd offers CAPITALISM as an example of a HPC social kind in "Kinds, Complexity and Multiple Realization," 83.
84. I am here offering these sample inferences after having done only a minimum of research. They are based on my own limited casual observations of Hasidic Jews (along with bits of testimony from others). No doubt, some of these observations need refinement or correction (e.g., I have not said - because I do not know - on which occasions tallis is worn). In any case, there can be little doubt that it would require only a modest amount of sociological research to extend both the number and the reliability of inductive inferences that can be made about Jacob in his capacity as a Hasidic Jew.
85. It should be recognized that we should expect inferences to be reliable only when they concern properties that are homeostatically clustered. Suppose that my sample of Hasidic men was small, consisting of only five men. Suppose further that I observed that all five men have gray beards. I might be tempted to infer that Jacob's beard will be gray as well. It should be clear, however, that, even if this conclusion were to turn out to be true, this inference is not epistemically warranted. What this observation suggests is that, if our practice of making inductive inferences from a small sample is to be practical, then we had better have some skill at detecting which properties of an individual are essential to its kind. For a defense of the claim that human beings really do possess such a skill, see Kornblith, Inductive Inference, 83-107.
86. See Boyd, "Kinds, Complexity and Multiple Realization," 68. It is worth adding here that Ron Mallon defends an HPC conception of certain social kinds where the homeostatic mechanisms include the members' own recognition of themselves as members of a kind.
87. See Ron Mallon, "Social Construction, Social Roles, and Stability," in Socializing Metaphysics: The Nature of Social Reality, ed. Frederick Schmitt (New York: Rowman & Littlefield, 2003), 327-53.
88. Some might object that the inferences about Bill are not inductive at all. It may be that 'hippie' has an analytic cluster definition where the properties I have attributed to Bill are just those that are analytically associated with 'hippie'. I am not inclined to protest very loudly against this objection. But note that this should be of no comfort to the homeostatic consequentialist. After all, the same objection may be raised against any (supposedly) inductive inference involving the good (i.e., it might be objected that such inferences are not examples of a posteriori inductions at all, but are, instead, examples of the analytic a priori). In any case, unless a clear case of a HPC social kind that weakly grounds inductive inference can be found, the homeostatic consequentialist cannot appeal to a comparison with social kinds in order to answer the arguments of Secs. V.B and V.C.
89. Boyd, "Finite Beings, Finite Goods," 34-38.
90. Ibid.
91. Boyd uses 'the good' here.
92. Ibid., 36.
93. Ibid.