Refining Realism: Laudan's Challenge as a Path to Stronger Foundations — Epoche B2
Refining Realism: Laudan's Challenge as a Path to Stronger Foundations Larry Laudan's 'A Confutation of Convergent Realism' (1981 [1] ) is usually filed as the paper that broke scientific realism. The claim of this essay is that it did something more useful: it forced realists to say which parts of a successful theory their optimism is about. Before 1981 they had mostly not been asked. The realism that emerged from answering — selective, and much more specific about what it predicts — is a better position than the one Laudan attacked. That is the sense in which the challenge was a resource. Some realists have been tempted to go further and hold that the selective strategies answer Laudan's cases and close the matter. They do not, and the reason they do not is worth stating precisely, because it marks where the argument now stands. What the realist claims, and why Scientific realism , in the version at issue, holds that our mature and successful theories are approximately true, and that their central theoretical terms refer to things that exist. Electrons are not a convenient bookkeeping device; there are electrons. The main argument for this is the no-miracles argument. Hilary Putnam gave it its slogan in Mathematics [2] , Matter and Method (1975): The positive argument for realism is that it is the only philosophy that doesn't make the success of science a miracle. The reasoning is an inference to the best explanation. Our theories predict things nobody expected, and the predictions come out right. If the theories were wholly off the mark, this run of luck would be inexplicable. Approximate truth explains it. So we should believe the theories are approximately true. Note the argument's form: it moves from success to truth by way of explanation, and it treats a theory as a single unit that is either on the right track or not. Laudan's list Laudan's attack targets exactly that treatment. He assembled a list of theories that were successful by the realist's own standards — they organised their domains, they yielded predictions, they were accepted by competent scientists for long periods — and whose central terms we now take to name nothing at all. The list includes the crystalline spheres of pre-Copernican astronomy, the humoral theory of medicine, the phlogiston theory of combustion, the caloric theory of heat, the optical and electromagnetic ethers, and the theory of circular inertia. From this he draws two conclusions. The first is a direct challenge to the no-miracles argument: if reference and approximate truth are not necessary for success, then success is not evidence for them. The second is the pessimistic induction proper: our current theories are in the same position relative to the future as phlogiston chemistry was relative to us, so we have no better grounds for confidence than Lavoisier's predecessors had. The argument is powerful because it is empirical. It does not dispute the realist's logic; it disputes the realist's premise that success tracks truth, and it does so with cases. The reply that does not work The first realist instinct is to raise the bar. Perhaps those theories were not mature enough, or their successes were not of the right sort, or they lacked the breadth of independent evidential support that our theories enjoy. Add enough conditions and Laudan's list is excluded while contemporary physics is not. This fails, and it is important to see why, because the failure is instructive. Any criterion assembled by inspecting the theories we now reject, and tuned until it excludes them, is fitted to the data it is meant to explain. Laudan can simply ask what the criterion predicted before the fact. If the answer is nothing, the realist has produced a description of the past rather than a reason for confidence about the future. The manoeuvre is not merely inelegant; it is unfalsifiable, and an unfalsifiable defence of realism is a poor advertisement for a philosophy that takes its cue from science. Selective realism: the answer that does work The productive response drops the assumption that a theory succeeds or fails as a whole. Suppose that only some parts of a theory are responsible for its predictive successes, and that the realist commitment is confined to those parts. Then a theory can be radically wrong overall and still have got something right — and if that something is retained by the successor, both the successes and the discontinuity are explained. Two versions of this have been developed. John Worrall's 'Structural Realism [3] : The Best of Both Worlds?' (1989) argues that what is retained across theory change is mathematical structure rather than ontology. Fresnel's account of light presupposed an elastic ether, which does not exist; but Fresnel's equations for the intensities of reflected and refracted light carry over into Maxwell's theory essentially unchanged, with the terms reinterpreted. The realist should be committed to the structure, and agnostic about the furniture. Stathis Psillos [4] , in Scientific Realism: How Science Tracks Truth (1999), argues for a different division, sometimes called divide and rule. Commitment attaches to those theoretical constituents that were essentially involved in generating the novel predictions — the working posits — and not to the idle ones. On his reading of the caloric case, the successful results of the caloric theory of heat, including Carnot's analysis of the efficiency of heat engines [5] , depended on assumptions about the conservation and flow of a quantity, not on the assumption that the quantity was a material fluid. The fluid was idle. What did the work survived. Either way, the realist now has a position with a testable consequence: the constituents identified as doing the predictive work should be the ones that survive. That is a claim about future science, not a redescription of past science, and it is what the blanket maturity criteria could not supply. A separate point about the induction's arit