1 Kitcher, Philip Abusing Science: The Case Against Creationism (Cambridge, MA: The MIT Press 1982)Google Scholar; Vaulting Ambition: Sociobiology and the Quest for Human Nature (Cambridge, MA: The MIT Press 1985)

2 See in particular Fraassen, Bas van The Scientific Image (Oxford: Clarendon 1980)CrossRefGoogle Scholar.

3 E.g., Churchland, Paul ‘The Ontological Status of Observables: In Praise of the Superempirical Virtues,’ in Churchland, Paul and Hooker, Clifford eds., Images of Science (Chicago: University of Chicago Press 1985) 35–47Google Scholar.

4 Actually the constructive empiricist could allow for the possibility of shrinkage or sensory alteration, while still maintaining that viruses were unobservable through the simple expedient of construing van Fraassen's characterization of observability as follows: X is observable (for me) as long as it is possible for a being with my biological structure (i.e., my structure at the actual world) to detect it with its unaided senses.

5 E.g., many papers in Churchland and Hooker, particularly Boyd, Richard ‘Lex Orandi est Lex Credendi,’ 41–82Google Scholar.

6 By ‘epistemological instrumentalism,’ I mean any position that holds that, although unobservables may exist, we cannot know or have justified beliefs about them. This position is in contrast with metaphysical instrumentalism, which denies the possibility of the existence of unobservables or at least of meaningful statements about them.

7 E.g., virtually every paper in Leplin, Jarrett ed., Scientific Realism (Berkeley: University of California Press 1984)CrossRefGoogle Scholar; and Laudan, Larry and Leplin, Jarrett ‘Empirical Equivalence and Underdetermination,’ The Journal of Philosophy 88 (1991) 449-72CrossRefGoogle Scholar.

8 I will discuss Kitcher's response to the Duhemians later in this review.

9 E.g., Laudan, Larry ‘A Confutation of Convergent Realism,’ Philosophy of Science 48 (1981) 19–49CrossRefGoogle Scholar; reprinted in Leplin, 218-49.

10 Rosenberg, Alexander and Hardin, Clyde L. ‘In Defence of Convergent Realism,’ Philosophy of Science 49 (1982) 604-15Google Scholar

11 See also Kitcher's, ‘Theories, Theorists, and Theoretical Change,’ The Philosophical Review 87 (1978) 519-47CrossRefGoogle Scholar.

12 The reference of a term is partially fixed by the speaker's mode of reference, which is largely determined by the speaker's intention, and of which there are three types: baptismal (that stuff there which was produced by heating the red calx of mercury is dephlogisticated air), descriptive (the stuff that results when phlogiston is removed from the air is dephlogisticated air), and conformist (whatever she meant by ‘dephlogisticated air’ is dephlogisticated air).

13 Buchwald, Jed Z. The Rise of the Wave Theory of Light: Optical Theory and Experiment in the Early Nineteenth Century (Chicago: University of Chicago Press 1989), 307Google Scholar

14 See Buchwald, Jed Z. From Maxwell to Microphysics: Aspects of Electromagnetic Theory in the Last Quarter of the Nineteenth Century (Chicago: University of Chicago Press 1985), 146-50Google Scholar.

15 Berkson, William Fields of Force: The Development of a World View from Faraday to Einstein (New York: Wiley 1974)Google Scholar

16 Here I am perhaps being unfair to Kitcher, who consistently shows a genuine respect for the referential intentions of the scientists he discusses. Nevertheless, referential intentions constitute only one among the many factors that determine the referential mapping for their utterances and explanations. The distinction between presuppositional and working posits is sufficiently fuzzy that future Kitcherians will enjoy much flexibility in applying to our practice.

17 Laudan, Larry Science and Values (Berkeley: University of California Press 1984)Google Scholar

18 However, that scientists pursue this goal does not entail that it is a fundamental goal. They could regard its attainment as instrumental in the pursuit of truth. Kitcher himself is ambivalent on which is more fundamental between pursuing significant truth and maximizing the unity of our belief-set.

19 Laudan, explicitly endorses this form of unificationism in Progress and Its Problems (Berkeley: University of California Press 1977)Google Scholar.

20 E.g., as Latour, and Woolgar, Laboratory Life (London: Sage 1979)Google Scholar assert of ‘TRF is Pyro-Glu-His-Pro-NH2.’

21 Note the counterfactual. I do not merely require that if Population N and Population M happen to perform the same experiment, they will get the same result. I also require that, even if they don't actually perform a given experiment, if they were to perform it they would get the same result. This is very rough and ready. For one, it obviously cannot apply to experiments in the social sciences concerning the practices of scientists. Also, I am assuming (as Kitcher does) that there is usually a theory-neutral way of describing the raw data generated by a theory or measurement.

22 Putnam, Hilary Reason, Truth, and History (Cambridge: Cambridge University Press 1981)CrossRefGoogle Scholar

23 In a footnote (173 n.) Kitcher adds that the existence of an ideal practice is not absolutely required, that the stability of the predicates and schema past a certain point in time is sufficient for their existence or correctness.

24 See Matheson, Carl ‘Is the Naturalist Really Naturally a Realist?’ Mind 93 (1989) 247-58CrossRefGoogle Scholar for a defence of the idea that naturalism is congenial to ideal-limit theories of truth.

25 For the purposes of simplicity, I am assuming that major theoretical clashes are between two practices. However, what I am saying would also work for many sided conflicts.

26 Feyerabend, Paul Against Method (London: Verso 1975)Google Scholar

27 One might argue that, instead of committing completely to method I or method II, a Pure 1 agent would divide her time between I and II according to the optimal distribution provided by the pi functions. Thus an optimal inter-personal distribution would be replaced by an optimal intra-personal temporal distribution.

28 Even if a given Pure were unable to communicate with others before choosing between method I and II, he could still do the following. If <j,k> is the optimal distribution (where j+k=N, the total population), then our Pure could generate a random number between 0 and 1, and pursue method I iff that number were less than or equal to j/N. If all the Pures followed this strategy, the resulting distribution would be nearly optimal for all but a few fairly degenerate pi functions.

29 The co-extensiveness of the distribution might not hold universally. Suppose that the community faces a very tough epistemic problem, one for which, according to the optimal distribution, a solution is not likely within a single lifetime. A community of Pure's will pursue a non-optimal strategy that offers a greater chance of a solution within their lifetime. A community of Pures would take the longer view. In other words, a community of Pure's is more susceptible to ‘Get Wise Quick’ schemes.

30 If Kitcher continues to insist that Pure 1 and Pure 2 are epistemic goals then we should bid him to consider: Pure 3 I want to maximize my chances of being the first to solve the problem. Pure 3s will of course generate the sullied-entrepreneur distribution. Since a Pure 3 is driven to solve the problem first for its own sake and not for the sake of a prize, Pure 3 is every bit as much an epistemic goal as Pure 1 and Pure 2, and it is every bit as plausible. Thus those who pursue epistemic goals can do just as well as those who are sullied. Even if Kitcher can rule out Pure’ as implausible, there is no principled distinction between epistemic and non-epistemic goals to support his contention that sullied communities outperform all plausible pure ones.

31 Many thanks to Philip Kitcher for some valuable clarifications and elaborations of his approach to social epistemology. I am also grateful to the Social Science and Humanities Research Council of Canada for providing some of the resources used in the writing of this notice.