Van Fraassen: Arguments concerning scientific realism 1. Scientific realism and constructive empiricism a) Minimal scientific realism 1) The aim of scientific theories is to provide literally true stories about the world; 2) Acceptance of a scientific theory involves the belief that it is true. The anti-realist denies the conjunction of 1) and 2). b) Positivism/instrumentalism: denies 1) that theories are to be construed literally (but accepts that they aim to be true). Rejected by van Fraassen: - linguistic distinction between theoretical/non-theoretical terms is hard (impossible) to draw - translation is impossible to carry out c) Constructive empiricism: denies that aim is truth. 1) Scientific statements should be construed literally. 2) Science aims to give us theories that are empirically adequate. 3) Acceptance of a theory involves only the belief that it is empirically adequate. Re 2): A theory is empirically adequate if its implications about observable things are true. Note: 1. All observable things, not just those observed in the past. 2. Not the same as the positivist T e, the part of T couched entirely in non-theoretical language. Q: Difference between truth and empirical adequacy? Re 3): Acceptance involves belief. It implies commitment and confidence that the theory will be vindicated, and these go beyond belief. [This aspect of acceptance, however, is nearly the same for realists and antirealists, and gets little attention from van Fraassen.]
2. Theory/Observation distinction Clarification: - linguistic distinction: theoretical vs. non-theoretical terms. - metaphysical distinction: observable vs. unobservable (putative) objects. The linguistic distinction is rejected: all of our language is theoryinfected (1071). The metaphysical distinction (the relevant one) is maintained. [If not, then we can t isolate the consequences about observable things, so constructive empiricism collapses.] Counter-arguments: 1) Maxwell s continuity argument: one cannot draw the distinction. Van Fraassen offers many criticisms: Maxwell s continuous series pertains to acts of observation, not observability. [Don t make too much of this.] Series shows only that observable is a vague predicate. A vague predicate is usable provided it has clear cases and counter-cases. There are clear cases of non-observation: particles in a cloud chamber. [But is this a clear case of unobservability?] Further: we should define observability relative to human biological limitations. [What is the argument?] 2) Maxwell s argument that observable in principle makes observability theory-dependent and human-dependent, hence irrelevant to existence, and hence not an important or relevant distinction. Irrelevance of observability to existence is a red herring The human-dependent distinction is relevant it figures in the notions of empirical adequacy (p. 1074) and theory acceptance The distinction is relative to an epistemic community, but so is theory acceptance Question: Replace observable with observable or manipulable?
3. Inference to the Best Explanation IBE: Given evidence E, we should infer H [is true] rather than H if H is a better explanation for E than is H. Realist argument: 1. We follow IBE in ordinary, everyday situations (e.g., the mouse behind the wall) 2. We use it in scientific contexts 3. So it can be used by philosophers as well. 4. And H (realism about entities) explains the evidence E better than any anti-realist H : so we should infer H. Van Fraassen s critique: Premise 1) is a psychological hypothesis (not a normative principle). As such, it s debatable. We might be following the following principle instead: VFP: Given evidence E, infer that H is empirically adequate if H is the best explanation of E. VFP is indistinguishable from IBE in everyday contexts (the mouse), because the mouse is observable. It is question-begging to suppose that we use IBE rather than VFP in scientific contexts. That undercuts the philosophical use of IBE in the argument for realism.
4. Realism and Explanation The miracle argument: only realism (entity and theory realism) can explain the usefulness of scientific theories. Smart s version (regress argument): 1) The success of instrumentally useful theories can (in principle) be explained. 2) This success can only be explained as following either from the instrumental usefulness or the truth of another theory. [e.g., usefulness of Ptolemaic model explained by truth of Copernican model] 3) But each backward branch of the explanation must terminate (begin?) with a true theory, or there would be no explanation at all. Van Fraassen s counter-argument: Reject premise 2): success could be just as well explained by an empirically adequate theory If asked what explains the empirical adequacy: the observable regularities are just a brute fact Theological analogy The point: Explanation has to stop somewhere. The realist accepts this too so there is no reason why explanation can t stop with empirical adequacy and basic regularities among observable phenomena.
6. Limits to explanation Sellars: - descriptive generalizations about observables are unsatisfying - offers a thought experiment for demanding belief in the truth of statements about unobservable entities Aqua regia: mixture of nitric and hydrochloric acid Evidence: different (observationally indistinguishable) samples of gold dissolve at different rates in aqua regia. Theory: two distinct micro-structures for gold; different rates are explained by mixture. Argument based on this thought experiment: 1. Science aims to explain (which motivates the hypothesis of differential rates). 2. Explanation is possible only at the unobservable level (and in this case has no observable consequences). *3. Explanations must be true. So science requires belief in truths about unobservables. Main objections: 1. Premise 1 is false: the point of introducing the differential rates might be to come up with a more empirically adequate theory. 2. Premise 2 is (partly) false: there are observable consequences. 3. Quantum mechanics shows that science does not *always* demand explanations for observable regularities. (Indeed, the demand for explanatory hidden variables is inconsistent with quantum mechanics.) 4. An anti-realist could also be motivated to provide this sort of explanation: it could lead to refinements in other parts of chemistry that result in a more empirically adequate science.
7. The ultimate argument [Omit discussion of Dummett] Putnam: Realism is the only philosophy that does not make the success of science a miracle. In more detail: 1. The regularity of interest is that scientific predictions are regularly successful. 2. This regularity requires explanation. 3. The explanation offered is that the [theoretical] terms of the theory refer to real entities, and the [theoretical] statements are approximately true. Van Fraassen s response: 1. Demand for explanation need not be pushed this far. 2. But if it is pushed this far: the explanation is weak. 3. There is a better (selectionist) explanation: only successful (= empirically adequate) theories survive. This does not require realism in the sense defined by Putnam.