1 Thus James Robert Brown argues that thought experiments show empiricism to be false, since their epistemic potency can only be explained by appeal to a Platonist epistemology, while Nancy Nersessian, Nenad Miscevic, and Michael Bishop echo Mach in holding that their epistemic relevance comes from the fact that they are simulations run on mental models of the world. See Brown, J. R. The Laboratory of the Mind (London: Routledge 1991);Google Scholar Brown, J. R. ‘Thought Experiments: A Platonic Account,’ in Thought Experiments in Science and Philosophy, Horowitz, T. and Massey, G. eds. (Savage, MD: Rowman & Littlefield 1991);Google Scholar Brown, J. R. ‘Why Thought Experiments Transcend Empiricism,’ in Contemporary Debates in Philosophy of Science, Hitchcock, C. ed. (Oxford: Blackwell 2004);Google Scholar Brown, J. R. ‘Peeking Into Plato's Heaven,’ Philosophy of Science 71 (2004) 1126–38;CrossRefGoogle Scholar Brown, J .R. ‘Thought Experiments in Science, Philosophy, and Mathematics,’ Croatian Journal of Philosophy 19 (2007) 3–27;Google Scholar Nersessian, N. ‘How Do Scientists Think? Capturing the Dynamics of Conceptual Change in Science,’ in Cognitive Models of Science, Giere, R. ed. (Minneapolis: University of Minnesota Press 1992);Google Scholar Nersessian, N. ‘Thought Experimenting as Mental Modeling,’ PSA 1992, vol.2 291–301 (East Lansing, MI: Philosophy of Science Association 1993);Google Scholar Miscevic, N. ‘Mental models and thought experiments,’ International Studies in the Philosophy of Science 6 (1992): 215–26;CrossRefGoogle Scholar Bishop, M. ‘An Epistemological Role for Thought Experiments,’ in Idealization IX: Idealization in Contemporary Physics, Poznan Studies in the Philosophy of the Sciences and the Humanities, vol. 63, Shanks, Niall ed. (Amsterdam: Rodopi 1998).Google Scholar Tamar Szabó Gendler, who develops Kuhn's idea that thought experiments work by exposing latent conflicts in the application criteria for concepts, and Kuhn himself, also belong in this broad camp, as do the more eclectic positions of Roy Sorensen and Tim De Mey. See Gendler, T. S. ‘Galileo and the Indispensibility of Scientific Thought Experiment,’ British Journal for the Philosophy of Science 49 (1998) 397–424;CrossRefGoogle Scholar Gendler, T. ‘Thought Experiments Rethought — and Reperceived,’ Philosophy of Science 71 (2004) 1152–63;CrossRefGoogle Scholar Sorensen, R. Thought Experiments (Oxford: OUP 1992);Google Scholar Kuhn, T. S. ‘A Function for Thought Experiments,’ in The Essential Tension (Chicago: Chicago University Press 1977);CrossRefGoogle Scholar T. De Mey, ‘Thinking Through Thought Experiments’ (PhD disseration, University of Ghent 2003). A position according to which thought experiments may be fruitful, but only under tightly circumscribed historical conditions, is advocated by James McAllister. See McAllister, J. ‘The Evidential Significance of Thought Experiment in Science’ Studies in the History and Philosophy of Science 27 (1996) 233–50.CrossRefGoogle Scholar

2 Thus Kathleen Wilkes argues that thought experiments in philosophy of mind typically fail to provide good evidence. See Wilkes, K. Real People. Personal Identity without Thought Experiments (Oxford: OUP 1988).Google Scholar And John Norton claims that thought experiments are simply arguments, albeit dressed up in exotic narrative garb. See Norton, J. ‘Thought Experiments in Einstein's Work,’ in Thought Experiments in Science and Philosophy, Horowitz, T. and Massey, G. eds. (Savage, MD: Rowman & Littlefield 1991);Google Scholar and ‘Are Thought Experiments Just What You Thought?’ Canadian Journal of Philosophy 26 (1996) 333-66; Norton, J. ‘Why Thought Experiments do not Transcend Empiricism,’ in Contemporary Debates in Philosophy of Science, Hitchcock, C. ed. (Oxford: Blackwell 2004);Google Scholar Norton, J. ‘On Thought Experiments: Is There More to the Argument?’ Philosophy of Science 71 (2004) 1139–51.CrossRefGoogle Scholar Adherents of the argument view also include Nicholas Rescher, Andrew Irwine and John Forge. See N. Rescher, ‘Thought Experimentation in Presocratic Philosophy’; A. Irwine, ‘On the Nature of Thought Experiments in Scientific Reasoning’; and Forge, J. ‘Thought Experiments in the Philosophy of Physical Science,’ all in Thought Experiments in Science and Philosophy, Horowitz, T. and Massey, G. eds. (Savage, MD: Rowman & Littlefield 1991).Google Scholar

3 Some arguments in this paper parts draw partly on my Thought Experiments in Philosophy (Stockholm: Almqvist & Wiksell International 1996).

4 Not everything called ‘experimentation’ meets this condition, of course; the term may refer simply to someone's doing something to see what happens.

5 One of my referees in effect articulated this complaint.

6 See N. Rescher, ‘Thought Experimentation in Presocratic Thinking,’ in Horowitz and Massey.

7 See J. N. Mohaney, ‘Method of Imaginative Variation in Phenomenology,’ in Horowitz and Massey. The point that the term encompasses very diverse things was forcefully made by Tamar Gendler in a review of Horowitz and Massey; see T. Gendler, ‘Tools of the Trade: Thought Experiments Examined,’ The Harvard Review of Philosophy VOL?? (1994) 81-5.

8 See De Mey, ‘Thinking Through Thought Experiments.’

9 ‘Why Thought Experiments Transcend Empiricism,’ 25

10 Ibid., 26

11 I shall argue below that thought experiments are typically devised in order to refute a theory one already takes to be false (just like many ordinary experiments), and to induce others to share this belief.

12 An anonymous referee did just that.

13 See Häggqvist, Thought Experiments in Philosophy. In the case of Galileo, the thought experiment has to be buttressed by a thought experiment refuting the reverse of Aristotle's actual theory, i.e. the claim that light bodies fall faster than heavy ones. Then the theory suggested by the thought experiment — that bodies fall at the same speed regardless of weight — emerges as the negation of the disjunction of the two refuted theories. For more on Brown's treatment of Galilei, as well as an extended discussion of his views on thought experiments, see Häggqvist, S. ‘The Priori Thesis: A Critical Assessment,’ Croatian Journal of Philosophy 19 (2007) 47–61.Google Scholar For instructions on how to run the reverse experiment, see De Mey, ‘Thinking Through Thought Experiments.’

14 ‘Thought Experiments Rethought — and Reperceived,’ 1154. Although they may not entirely endorse this conception, none of her three co-symposiasts — James McAllister, James Robert Brown and John Norton — actively opposed it.

15 See Williamson, T. ‘Armchair Philosophy, Metaphysical Modality and Counterfactual Thinking,’ Proceedings of the Aristotelian Society 105 (2005) 1–23.CrossRefGoogle Scholar

16 Chapter 6 of Williamson, T. The Philosophy of Philosophy (Oxford: Blackwell 2007).CrossRefGoogle Scholar

17 ‘Thought Experiments in Einstein's Work,’ 129.

18 Ibid.

19 Bishop, M. ‘Why Thought Experiments Are Not Arguments,’ Philosophy of Science 66 (1999) 534–41.CrossRefGoogle Scholar

20 Sorensen, R. Thought Experiments (Oxford: OUP Press 1992), 214.Google Scholar

21 Of course, Norton would disagree with this (as a referee remarked), so in a sense this claim begs the question against Norton. But I submit that most theorists not already wedded to the argument view will find this observation plausible. And Sorensen's and Bishop's points speak independently against the argument view.

22 After all, some thinking arguably takes place outside our bodies in such parts of what, following Dawkins, might be considered our extended phenotype. See Dawkins, R. The Extended Phenotype (San Francisco: Freeman 1982).Google Scholar A corollary to viewing thought experiments as psychological processes is that they are not properly contrasted with real experiments, since such processes are themselves patently real.

23 Again, this is not the only use of everything reasonably called ‘thought experiments.’

24 As a matter of etiology, this schema emerged from reflections on Sorensen's suggestions for regimentation in Thought Experiments, ch. 6. But as a referee pointed out, it seems a pretty straightforward proposal once one thinks of thought experiments as aiming at testing theories. For discussion of what I take to be the shortcomings of Sorensen's proposal, see my Thought Experiments in Philosophy, ch. 5.

25 Sorensen, Thought Experiments, 135Google Scholar

26 Thanks to an anonymous referee who prompted this paragraph.

27 ‘Why Thought Experiments Are Not Arguments’

28 The formal availability of these ‘ways out’ is quite parallel to the holism of hypothesis testing in general. For explicit discussion of the Duhemian holism in connection with thought experiments, see Häggqvist, Thought Experiments in Philosophy (ch. 6), and Bokulich, A. ‘Rethinking Thought Experiments,’ Perspectives on Science 9 (2001) 285–307.CrossRefGoogle Scholar

29 Hardline utilitarians can sometimes be observed making this move.

30 E.g. Davidson: ‘I have a general distrust of thought experiments that pretend to reveal what we would say under conditions that in fact never arise.’ (‘Epistemology Externalized,’ reprinted in D. Davidson, Subjective, Intersubjective, Objective [Oxford: OUP Press 2001]). Davidson writes this a propos Burge's and Putnam’s thought experiments in favour of externalism; for the record, let's note that the comment inaccurately characterizes the intent of these cases: their authors pronounce on what we in the actual world should say about the counterfactual circumstances they depict (and hence what would be the case in them), not what we would say in them.

31 For a discussion of such hypophobia, see Sorensen, Thought Experiments, 275.

32 In section VI below, I argue that this response is harder to assess than the corresponding response in the case of an ordinary experiment, and that this is detrimental to the epistemic value of thought experiments.

33 This seems to be the gist of Daniel Dennett's reply to various zombie, blockhead and swampman cases. See e.g. Dennett, D. ‘Get Real,’ Philosophical Topics 22 (1994) 505–68.CrossRefGoogle Scholar

34 Insisting on soundness of the regimentation would be asking too much. The issue here is with justification, not truth, since we are asking when a thought experiment aiming at justified belief revision achieves or fails to achieve this aim.

35 This thesis is an amalgam of two theses: (i) that speaker's psychology determines the intension of her words, and (ii) that intension determines extension/reference. So giving up the target thesis may be done by giving up either of these claims. Putnam opts for relinquishing (i). Also note that ‘psychology’ here means ‘narrow psychology’ (as it used to before the distinction between broad and narrow psychology was invented).

36 Fodor, J. Psychosemantics (Cambridge, MA: MIT Press 1987)CrossRefGoogle Scholar

37 Bohr, N. ‘Discussions with Einstein on Epistemological Problems in Atomic Physics,’ in Albert Einstein: Philosopher-Scientist, Schilpp, P. A. ed. (La Salle, IL: Open Court 1949).Google Scholar

38 Bohr, ‘Discussions with Einstein,’ 228.Google Scholar A referee asks whether this doesn't concede Norton's point in reply to Bishop (‘Why Thought Experiments do not Transcend Empiricism,’ 63-4): that Einstein and Bohr in effect contemplate different thought experiments. Though granting this would not jeopardize my general claims, I do not think this is the right way to think about the disagreement between Einstein and Bohr, or generally between people disagreeing over a thought experiment. Such disagreement will always involve some difference in opinion, which will be reflected in the premises they accept in arguments concerning the case at hand. It seems wanton to infer that they thereby contemplate different cases. The same point obviously applies to controversies surrounding ordinary experiments.

39 ‘Why Thought Experiments Are Not Arguments.’

40 Or rather, they did so initially, since Einstein quickly accepted Bohr's argument and abandoned his own.

41 Several other examples are given in Häggqvist, Thought Experiments in Philosophy.

42 Nersessian, ‘How Do Scientists Think?’

43 Brown, The Laboratory of the Mind, ch. 4Google Scholar

44 See, e.g. Chalmers, D. The Conscious Mind (Oxford: OUP 1996),Google Scholar Jackson, F. From Metaphysics to Ethics (Oxford: OUP 1998),Google Scholar and Yablo, S. ‘Is Conceivability a Guide to Possibility?’ Philosophy and Phenomenological Research 53 (1993) 1–42.Google Scholar Yablo has since criticized the use of conceivability as evidence for modal claims; see Yablo, S. ‘Textbook Kripkeanism and the Open Texture of Concepts,’ Pacific Philosophical Quarterly 81 (2000) 98–122.CrossRefGoogle Scholar

45 This was suggested by a referee.

46 Cf. the discussion below.

47 The obvious exception being necessity claims. After this paper was submitted, Tim Williamson has suggested a formalization of philosophical thought experiments according to which they offer counterfactual counterexamples to metaphysical necessity claims, with Gettier's objections to the analysis of knowledge as justified, true belief as a prime example. See e.g. Williamson, ‘Armchair Philosophy, Metaphysical Modality and Counterfactual Thinking’ and The Philosophy of Philosophy. The quick rejoinder is that most theses targeted by thought experiments are not metaphysical (or logical) necessity claims. For discussion of Williamson's views in general, and his formal model in particular, see S. Häggqvist, ‘Modal Knowledge and the Form of Thought Experiments,’ forthcoming in The A Priori and Its Role in Philosophy, N. Kompa, C. Nimtz and C. Suhm, eds.

48 Cf. Norton's characterisation of thought experiments as ‘arguments which … invoke particulars irrelevant to the generality of the conclusion’ (‘Thought Experiments in Einstein's Work,’ 129). On the view I am advocating, any experiment aspires to deliver premises for an argument invoking particulars — those of the experimental set-up — that are relevant to the generality of the conclusion, by standing in an appropriate relation of instantiation or counterinstance to the general statement mentioned in the conclusion. But with thought experiments, even what counts as relevance is moot.

49 This was suggested by an anonymous referee.

50 This is for instance the gist of John Perry's reply when he argues that physicalism isn't committed to the claim that Mary learns nothing new when she leaves her black-and-white room. See Perry, J. Knowledge, Possibility, and Consciousness (Cambridge, Massachusetts: MIT Press 2001).CrossRefGoogle Scholar