Microsoft word - jb proposal.doc

A PRIORI REVISABILITY AND THOUGHT EXPERIMENTS IN SCIENCE In the present paper I will discuss the revisability of scientific principles and in particular the role thought experiments play for the revision. For this purpose I will assume that there exist constitutive a priori principles in science. The most elaborated recent account in this spirit is proposed by Michael Friedman. In a series of recent books and articles, he supports and develops the concept of relativized yet constitutive a priori. His model of scientific theories, such as Newtonian mechanics and Einsteinian relativity theory, consist of two distinct parts: empirical part that contains empirical laws (like the law of universal gravitation, Maxwell’s equations of electromagnetism or Einstein’s gravitational field equations) and constituvely a priori part that contains both the relevant mathematical principles used in the formulation of the theory (Euclidean geometry, Minkowski space-time, etc.) and some fundamental physical principles (Newtonian laws of motion, light principle, equivalence principle). He claims that these principles change along the development of the natural science and also admits of the possibility that the change is to certain extent a response to empirical considerations. Still, he argues that they could be held constitutively a priori as in a Kantian sense. In what follows I will discuss some important consequences of this thesis. The challenge before Friedman is to defend this view against opposition of a Quinean type by explaining what could mean for principles that change and develop in response to empirical findings to be a priori. Could they turn out to be empirically false? Friedman follows two general lines of defense. The first one is to distinguish between mere status of empirical fact and status of defining principle that constitutes new framework within which only the formulation of empirical laws and their testing is possible. The second line is, to use Poincaré’s formulation, the “elevation” of an empirical law to the status of convention, or what Friedman substitutes for it, constitutive principle. The non-empirical component is the “decision” that provides radically new mathematical spatio-temporal framework with empirical meaning. Without it, the empirical laws formulated and tested within the framework are not and cannot be empirical true or false but only could be empirically undefined. Since I believe that such type of defense may not be definitive against the natural objections about the empirical origin and (or) justification of the principles in question I would like to propose additional, third line of defense for their a priori character. I will argue that these principles’ revisability which is beyond question, even that in some cases could be and is actually done because of empirical considerations in a small number of cases it is not conducted so but is conducted through a priori revision. This, I believe, would allow defending the a priori status of the constitutive principles more efficiently since the danger of their turning out to be empirically justified through argument claiming empirical grounds for their truth and meaning would be eliminated. This danger is presupposed by the empirical component that is present in the principles, as Friedman himself admits, a fact that makes it in principle difficult to hold them true on purely a priori grounds. I do not think that complete isolation of the empirical component could be achieved entirely satisfactory when the matter comes upon the justificatory status of the In his famous essay “A Function for Thought Experiments” Kuhn unambiguously proclaimed that thought experiments are “one of the essential analytical tools which are deployed during crises and which then help to promote basic conceptual reform”1. My claim will be that in the case with scientific principles that underlie paradigms of Kuhnian type the aprioristic revision could be done through thought experiments. I will not claim that the thought experiments alone are reasons for paradigm shifts, or that they are the only engine behind such revolutionary changes. I will pursue the more modest claim that in some cases, which nevertheless parallel paradigmatic changes in the scientific framework, some of the fundamental principles of the revised framework have been actually revised a priori. My historical illustration for that will be based on an influential thought experiment from modern physics, Einstein’s train thought experiment which revises the fundamental principle of the For that purpose I will make use of the most exhaustive account of thought experiments in science available recently, namely, the one proposed by James Brown in The Laboratory of the Mind and related articles. He explicitly argues for the a priori character of selected set of thought experiments. I will support his claim and I will use it with respect to the model of Friedman by arguing that in a very few instances we have cases of a priori revision of constitutive principles where the considerations for the revision were purely a priori and not empirical. I would not deny that in most cases, empirical dissatisfaction and tension played certain role but I will insist that when it comes to the justification of the revision, the actual considerations remain non-empirical. Even if though it may seem paradoxical at first sight to combine platonic account of thought experiments and the laws of nature proposed by Brown with the neo-Kantian account of constitutive scientific principles defended by Friedman I will argue that at a certain point they may turn out to be complementary and not contradictory. The common base for both accounts is the a priori character of the principles (Friedman) and thought experiments (Brown) that revise some of them and establish other through a priori reasoning. This base could be interpreted more into direction of neo-Kantian understanding by saying that the a priori reasoning of the thought experiments is due to some defining a priori features of our mental structure or in direction that is more platonic by saying that the constitutive principles are reached by something like “seeing” the relations in a kind of platonic realm. At the same time, I believe that reasons could be proposed for general compatibility between Friedman’s model of scientific theories and Brown’s platonic basis for the thought experiments. My main considerations are the following: on the one hand platonic view of mathematical knowledge is still accepted as most compelling if we believe that standard Tarskian semantics holds for it, and Benaceraff convincingly showed that we have more than good reasons for that2. On the other hand, the constitutive role that mathematical principles play in Friedman’s model to provide the framework within which the empirical laws are to be formulated is indispensable. Serious defense of theoretical model with mathematical principles at the core needs the model at least to be shown to be epistemologically compatible with the most compelling account of mathematical knowledge that we have. However, this line is beyond the In general, we could distinguish between two types of a priori principles – principles that are conclusions of sound arguments and separate principles, that mere figure by themselves in axiomatic system or scientific theory – they have their truth not by virtue of sound logical deduction but either for reasons of intuitive certainty (the 5th postulate) or good fitting in scientific theory, the analogy here could be with inference to the best explanation. The first type could be considered a priori exclusively if the premises of the argument which conclusion they are are a priori, we may call it inferential a priori. The second type principles, the ones that figure separately and are not conclusions of an argument could be considered a priori if we are justified to hold them true without appeal to experience, we may call justificatory a priori. The question whether we could consider conclusion result of aprioristic revision of an argument with empirical premises as a priori is controversial. One way to argue that we could is to distinguish between justificatory a priori and derivational a priori. In the case where the empirical premises are only derivationally a posteriori but not justificatory we still have grounds to hold that this is sufficient for the conclusion to be a priori. The other possibility where at least one of the premises is justificatory empirical looks without prospects for a priori conclusion. Further question is whether we can revise a priori argument that has empirical premises, this would be possible only if we allow the aprioristic revision operation to extend beyond arguments with a priori premises and revise arguments with even empirical ones. With proper work definition of the epistemic status of the revision and with the help of historical examples we could support the claim for a priori revision of arguments with empirical premises. I propose the following requirement for revision to be considered a priori: The belief or set of beliefs that causes the revision have to be a priori Probably the empiricist opponent would not object as much to the a priori character of the procedure per se as much as he would object to the a priori character of the knowledge provided by its conclusion. If the conclusion delivering the a priori principle subject to revision is reached on the basis of empirical induction or good theoretical fitting, we could show through aprioristic revision that it is wrong or with insufficient explanatory power (within a theory). Here cases from history of science will fit where empirical principle has been revised but not in the light of experiential considerations but by means of a priori reasoning alone. If we accept that the main justification behind the Aristotelian-Scholastic view for the motion of falling bodies is ultimately empirical we can count Galileo’s falling bodies experiment as refutation of this kind. Important remark has to be made that aprioristic revision does not automatically mean that empirical revision is impossible for that case. It is natural to suppose that every revisable scientific principle one way or another could turn out to be subject of direct or indirect empirical revision. However, as Friedman points in connection to this, in several cases the principles are so fundamental that they provide the very possibility for a statement about empirical phenomena to receive definite value such as true or false. In this sense we could argue that in some cases the empirical revision is not possible unless some fundamental principles are present. A priori principles in science: Friedman’s Model Michael Friedman proposes complex three-layered model of dynamical system for scientific knowledge. The structure of the model is presented by the following three levels: 1. Surface – concepts and principles of natural sciences: empirical laws of nature, like Newtonian Law of gravitation or Einstein’s equations for gravitational field. Faces tribunal of experience by means of 2. Second level – constitutive a priori principles. Defines the fundamental spatio-temporal framework, within which only it is possible formulation of empirical laws and their testing. These principles constitute Kuhnian paradigms – relatively stable set of rules of the game that allow for the problem solving of the sciences and the formulation and testing of the empirical law candidates. In conditions of conceptual revolution, these are the principles that change under empirical pressure and findings. In periods of revolutions, no empirical testing of them is possible. 3. Third level – philosophical meta-paradigms, meta-frameworks. Guiding, motivating and sustaining the transition between the paradigms (conceptual frameworks). Friedman argues that the relativised a priori principles accommodate the conceptual revolutions. In fact the revolutions themselves revealed that our scientific knowledge has foundation layers of such type. The revision of the frameworks requires expansion of our space of intellectual possibilities to such extent, that mere direct appeal to empirical evidence is not relevant during the revolutions. The philosophical and constitutive layer guides the articulation of such new space of possibilities. Therefore, the various levels of our total (scientific) beliefs are not distinguished by mere degree of epistemic security or Quinean degrees of centrality but by their different and still complementary contributions to the total development of scientific knowledge. In this model, each scientific theory (Newtonian, Special Relativity, General Relativity) has three asymmetrically functioning parts: a. Mathematical part – contains the basic mathematical theories, representations or structures, intended to describe the spatio-temporal framework in question [infinite Euclidean space, 4d Minkowskian space-time, b. Mechanical part – in order (c) to succeed using (a) it needs principles of coordination [Newtonian laws of motion, light principle, equivalence principle], which set general correspondence between the math part and the concrete empirical phenomena in such a way that empirical laws could have empirical meaning. c. Physical (empirical) part – attempts to use the math part in order to formulate precisely empirical laws which describe concrete phenomena [law of universal gravitation, Maxwell’s equations for EM field, Einstein’s equations for gravitational field] The laws of nature comprising the physical part could be empirically tested. Below is given reconstruction of Friedman’s argument in Dynamics of Reason3: 1. There are three types of principles comprising scientific theory: mathematical, constitutive and physical 2. The constitutive framework defines space of empirical possibilities (set of statements that could empirically be True or False) by setting general correspondence between the math part and the concrete empirical phenomena in such a way that empirical laws could have empirical meaning. 3. The constitutive framework, together with the mathematical part (which is non-empirical) provides the very conditions for a scientific statement to be empirically True or False, therefore 4. The constitutive framework could not be empirically justified, 5. The constitutive framework is nevertheless justified, hence 6. The constitutive framework is justified a priori The constitutive framework is built up of a priori revisable principles. What are the possible types of revision available? I will concentrate on the role of the a priori revision for reasons that Since the infallibilism proved to be heavy burden for most doctrines of the a priori and especially in the case with principles in the natural sciences, candidates for a priori, I will argue in favor of the possibility for the a priori pieces of knowledge and those that fulfill the role of principles for the science to be revised. The revision in this sense should not be taken solely as complete revision, i.e. rejection of the principle in the light of some empirical or rational evidence, but should be taken also as the additional case where we have change of the meaning of the principle while at the same time some part of it (substantial or not) remains the same. We could distinguish between two general types of revisions – empirical and aprioristic. It is important to introduce the sub-distinction between revision in conditions of “empirical pressure”, by the words of Friedman, and revision made for purely empirical considerations. While every revision made because of empirical considerations is clearly an empirical one, the other case allows revision to be done by a priori means, though in the conditions of empirical pressure. The pressure here could be interpreted as empirical lack of scientific progress that induces revision or change on the paradigm shift. Historical example of a case where “revision” means not refutation of the meaning of the principle in question in the traditional sense, but change in its meaning, is the 5th postulate of the Euclidean geometry – after the formulation of non-Euclidean geometries it is not refuted as false but its scope of application and validity is limited after the revision. Illustration of a case with complete refutation could be Galileo’s refutation of Aristotelian-Scholastic view on motion of bodies or Einstein’s refutation of absolute simultaneity. Probably most people would agree that if a principle is a priori it could be revised a priori, so this would not seem very surprising. I will also make the additional claim though, that the a priori revision is more important that the empirical one. We have cases where supposedly a priori principles and even the most secure ones have been empirically revised, in harmony with Quine’s famous dictum that no statement whatsoever is immune to empirical revision. For example, we could take the measurements of the light paths in the space as empirical revision of the supposed a priori principle that “The Euclidean geometry is the only true one of our world”. The experiential revision demonstrates that the claim is at least partly false, in the sense the even if it is true of some regions of the world it is not true of all of them. However, our reasons to hold the axioms of the Euclidean geometry true could not be experiential – we do have only intuitive reasons to hold them true, we do not have and we could not have experience of, for example, infinite extension of a line, we could draw this only in our imaginative space. Upon careful investigation it becomes clear that we have had presupposed some more primitive properties of the space in which we draw the lines (the space of our imagination), namely (for ex.) curvature of the space (with value zero in the case of Euclidean geometry). We have two lines – to claim that these are properties with a priori or with a posteriori origin. Reichenbach’s general line is that at least partly they could be a posteriori4. Convincing response to that has been given by Lukas5 and Cassirer6 that we do not have much of a choice and actually could not change the primitive properties7 of our phenomenal space since we will lose our ability to imagine and therefore make use of geometrical invariants upon the most simple and primitive group transformations (like reflection, displacement and rotation) which, as a result, will not allow us to form the concept of a material object the way we have it now. In this sense the a posteriori line of the origin of Euclidean geometry does not look We have empirical revision of a statement (principle) P when its presumed truth value changes when reviewed in the context of other statements, say Q and R and they are of empirical origin. For example, the conjecture “Water starts to boil at 99 degrees Celsius” is revised when we conduct exact experiments and establish that in all cases, under standard experimental conditions the water starts to boil at 100 degrees Celsius. Let us call the resulting affirmative statement “In every measured case water starts to boil at 100 degrees” Q-type. If subsequent experiences do not show deviation of this measurement – we have an additional statement that excludes for counterexamples of the affirmation, something like “There are no known empirical exceptions of the measurement that water starts to boil at 100 degrees”. Further, since any measurement presupposes theoretical context, within which every result is interpreted (as basic set of rules that concern both the interpretation of the results and the physical behavior during the experiment) let us call the set of context statements, illuminating the experiment “set C”. If we generalize and subsume under the Q-type all statements (of affirmative and limiting character) that are responsible for the revision of P (conjecture, principle) we can accept that there could be as many Q-type statements as necessary for correct and exhaustive description of the experiment. In addition, every Q statement should be consistent with both state of affairs exhibited during the experiment and C. In all cases when statements Q are empirically justified statements, i.e. we know them to be true solely on some experience I will take them as empirical statements. When all Q statements are not justified empirically but are held on a priori grounds they are respectively a priori statements. Example of such statements could be any statement involved in the revision of the 5th postulate. The question about the character of the set C justification is treated in length by Friedman. His proposal is described in his three layered model of scientific theories and if we take C to be the underlying structure for any revision that not only illuminates but also allows for the empirical formulation and the empirical testing of the empirical phenomena (laws) we have reasons to accept that in this general sense the C set is a priori. Important distinction that has to be made here is the one between origin and justification of the statement – since every statement presupposes some, no matter what experience, in order to exist at all I will exclude this trivial sense of empirical and I will not consider such statements as empirical ones for merely this reason. The only sense in which the empirical statements are empirical is the justificatory sense. Justificatory here would mean that the employment of the statements (principle) couldn’t be justified from experience but only a priori. Further distinction is the one that divides statements with respect to the origin of their justification: Type C – those are justified solely from experience and cannot be justified without it. This has the consequence that the only reason for us to hold these statements true could be experiential justification. These are experiential statements. Type B – those could be justified (in equal or different degree) both or separately by experience and by non-experiential consideration. These are joint justificatory statements. Type A – those could not be justified by experience and that are still held true on some (non-experiential) grounds – these statements will owe their truth solely on rational, i.e. a priori grounds. These statements will be meant as a priori statements. When the revision of P takes part because of Q statements, in the context of C and when at least one of the Q statements is experiential we have case of experiential revision. On the basis of the above classification we can define two kinds of a priori revision: Weak a priori revision – in the case where at least some of the Q and set C’s statements are justified a priori (being type A) and P is revised in their light we have a case with weak a priori revision. Strong a priori revision – in the case where all Q and set C’s statements causing the revision are justified a priori (being type A) we have case with strong a priori revision. It is important to point out that only statements that concern P’s meaning and cause the revision of P can qualify for Q status. Statements that make possible Q statements in the case of the a priori revision could be derivationally empirical but not justificatory empirical. I will argue that small number of cases satisfy the requirements for strong a priori revision. Why the a priori revision is any better than the empirical one? Why is it necessary to have a priori revisions instead of empirical ones? Isn’t it true that a priori principle in mathematics or natural sciences could be empirically verified or refuted? What is the status of the a priori revisions among the various ways to revise principles, most of which look straight empirical? At last, why is the a priori revision any better than the empirical one in the case with Friedman’s theoretical model? All these are legitimate questions when it comes to the relation between the a priori principles in science and the aprioristic revision. I will propose two The first line is to admit that as far as these principles do have empirical component, as Friedman admits, they must be in some degree revisable empirically. This revision could be more like directly revising the claim of the principle by ad hoc experiment or observation or indirect one where an empirically testable effect of the principle is examined. On the same line however, must be stressed that since the justification of the principles is a priori our reasons to hold them true are not empirical. The supposition that if a principle was held true for rational reasons when revised the very rational reasons must have been changed or enriched. Thus the axioms of geometry like the 5th postulate or fundamental principles of physics like the light principle (which is rationally derived from the Maxwell’s electromagnetic field equations) are believed not because they are some highly supported inductive generalizations but because they are either intuitively true or derivatively true or they are conventionally chosen to be true. In the first two cases we have nets of beliefs which are connected between each other by logical relations the core part of which is justified a priori. Since of the accepted thesis of fallibilism the core beliefs are subject to revision and there are two general candidates for revision considerations, the empirical ones and the a priori ones. As we have seen historically it is very difficult for core principles like the 5th postulate to be revised and still, when overturned they have been revised by a priori means. In the case with mathematics, this is more or less uncontroversial. In the case of the natural sciences and physics in particular the history of paradigm shifts or Kuhnian scientific revolutions shows that we always have at the core of the reasoning one or more central set of revisionary considerations, which are dressed up in the form of thought experiments. Examples are Galileo refutation of the Aristotelian –Scholastic view about motion, Galileo’s ship thought experiment for the Galilean relativity, Newton’s bucket thought experiment in support of the absolute space, multiple thought experiments by Einstein – the famous train thought experiment, that establishes the relativity of simultaneity, the elevator thought experiment that establishes the Principle of Equivalence, the famous EPR that was meant to undermine the anti-realist Copenhagen Interpretation as well as some others (chasing a light beam, rotating disk thought experiment), further Shrodinger’s Cat thought experiment (also was directed against the Copenhagen Interpretation)8. In all of the above instances, we have shifts in the paradigms, i.e. revolutionary revisions of our previous scientific frameworks or at least fundamental questioning the ruling principles of the reigning theoretical framework. In all these as well as in other cases we have revisions of scientific principles by means of a priori reasoning. I believe that main support for this view could come from Brown’s taxonomy of thought experiments. He holds that the main feature of some thought experiments is their a priori character. In this sense, I believe that approach that views the a priori revision of scientific principles as more important than or at least as important as the empirical revision for the paradigmatic shifts has good prospects. The second line is more of epistemological character. If an a priori principle turns out to be empirically false because of successful empirical revision, it would be difficult to continue to be considered as justified at the first place. Even in cases where we have grounds to keep the a priori justification of empirically false principle, as Casullo points, the strength of such principle and especially in science would be negligible. In this sense, it would be better for the a priori strength of the principle if we show that its revisability could be done and actually historically has been done via a priori reasoning. This would lessen the danger of this principle turning from a priori justified into a posteriori false. In addition, as Casullo argues convincingly, the matters of a priori justification and empirical truth of a priori justified belief are possible to be viewed as being independent.9 In this sense it is possible for a principle to be a priori justified and its empirical interpretation to be with undefined value. The other option is empirical definition of certain principles to be impossible for reasons, which taken from Friedman’s model could be of the following sort: since the a priori constitutive principles provide the framework within which the empirical laws are not only tested but even formulated they by themselves could not be subject to such test. In summary, strong role of the a priori revision will provide better support for both revisability and a priority of the principles in question. For the sake of clarification we should in addition distinguish between two roles of the aprioristic revision – competitor role where it is at odds with empirical revision and exclusive role where no empirical revision could be proposed in principle. In the latter case, the aprioristic revision is the only option. In this sense it could be powerful tool for the scientist where experience for one or another reason does not allow observations or physical experiments to be conducted. This is another scientifically important reason for the superiority of the a priori versus Einstein’s Train Thought Experiment In this section I will reconstruct Einstein’s famous train thought experiment10 and I will argue that it could serve as an example for scientific thought experiment that leads to new a priori knowledge of nature. For that purpose I will state the assumed principles in the thought experiment and I will try to defend the view that in this particular case they have been held a priori. In order to secure the claim for the a priori results of the thought experiment I will argue that there was no new empirical data involved nor the conclusion was somehow derived from an old empirical data. At the end I will give brief defense against recent criticism proposed by Norton, namely that thought experiments lead to nothing else but a kind of logical truth. The two major principles assumed in the thought experiment are the principle of relativity (in restricted sense, PR) and the principle of the constancy of speed of light (light Principle of Relativity (in restricted sense): If, relative to K, K’ is a uniformly moving co-ordinate system devoid of rotation, then the natural phenomena run their course with respect to K’ according to exactly the same general laws as with respect to K for all inertial Principle of constancy of the speed of light: The constant speed of light (c) in vacuo does not depend on the (magnitude and direction of ) velocity of the light The Experiment and the conflict between the two principles: We consider an embankment, a carriage that moves with velocity v with respect to the embankment and a ray of light with speed c emitted along the embankment. Einstein’s question is: “What is the velocity of the light ray relative to the carriage?” Following the classic rule for addition of velocities it should be w = c - v. In this case the velocity of the light ray relative to the carriage comes our smaller than c. This, however, is in conflict with the principle of relativity since, like every other general law of nature, when reviewed with accordance of the principle of relativity, the law of the transmission of light in vacuo should be the same for the carriage as reference system as well as for the embankment as a reference system. The dilemma: either PR or LP is to be abandoned. But LP is necessary consequence of Maxwell’s equations and Lorentz’s theory of electromagnetic phenomena and cannot be abandoned. At the same time Einstein holds that PR also cannot be abandoned. After an analysis of the concepts of time and space Einstein claims that it becomes evident that in reality there is not the least incompatibility between PR and LP. By maintaining both of them we can arrive at logically rigid theory (Special Relativity, STR). Lightning strikes the rails at two distant places A and B simultaneously. Einstein asks whether there is a sense in this statement? The common sense answer is “Yes”. For the physicist the concept should be void unless he has the possibility to discover (not simply measure or observe, p.22) whether it is fulfilled in the actual case. We thus arrive at the need for a definition of the simultaneity such that it supplies us with method by means of which we can decide by experiment whether or not both strikes occurred simultaneously or not. In every case the definition of simultaneity should provide us with an empirical decision as to whether or not the conception that has to be fulfilled has been actually fulfilled.
M’

Speed v ===== I =========== I =========== I =========== train /
--------- I ------------------- I ------------------- I ------------------- embankment
People in the train will regard all events in reference to the train, they will regard the train as rigid coordinate system (reference body). The question is: Are two events (A and B lightning strokes) simultaneous with reference to the embankment and with reference to the A and B are simultaneous with reference to the embankment when the light rays from A and B meet at the point M on the embankment. But A and B also correspond to A and B events on the train. M’ is the middle point on the train and when the flashes occur it coincides with M, but still it moves with speed v to the right. If observer at M’ did not posses this velocity it will remain permanently at M and the flashes from A and B will reach him simultaneously. But, in reality, when considered with respect to the embankment, he will ride ahead on the beam from A and he will haste towards the light beam coming from B. Therefore, he will see the light bean from B earlier than the light beam from A. Observers who take the train as their reference body will regard B flash as happening earlier than A flash. Thus, they will disagree with observers on the embankment about the simultaneity of the flashes. Hence Einstein concludes that simultaneity is relative (to the coordinate system or reference body) and unless we have chosen a reference system the statement of time of an event is meaningless. Before the theory of relativity it has been assumed that the statement of time has absolute significance, i.e. it is independent of the state of motion of the reference body (system). If we discard this assumption the conflict between PR and LP disappears. We can summarize the 4. contradiction between PR and LP in the light of AS 5. AS investigated by the train thought experiment 6. different observers found to disagree about simultaneity of events (M’ sees B light before A light) 8. Relative simultaneity established (RS) 9. in the light of the operational definition of RS, PR and LP are not contradictory any more I believe it to be more or less uncontroversial that the very nature of the experiment, namely being thought experiment and not real physical one, should be sufficient to grant the empirical free status of the transitions between the assumptions. If this is so, then there is one general thing left that has to be responded to the opponent here. Namely, that the principles assumed in the experiment have been held as justified a priori. In order this line to be supported I will consider the traditional objections against the a priori character of both the assumptions of the thought experiment and the justification for the transitions between them and I will propose answers to them. These objections can be summarized in the following three points: i. Motivation for the Train thought Experiment came from empirical realm. TE is meant to explain the result Empirical considerations tell Einstein that LP Train thought experiment – assumed principles come from empirical science The first point concerns the actual historical situation that took place with respect to the role of the Mickelson-Morley experiment. The view that major motivation behind Einstein’s work on STR and on the revision of absolute simultaneity in particular was actually an attempt to account for the negative result of the Mickelson-Morley experiment, was in fact the dominant one in both popular expositions and textbooks on the subject. 11 An intense debate among historians of science and philosophers of science discussed that particular issue. The most prominent figures in the debate from the recent past are the historians of science Gerald Holton and Michael Polanyi, the psychologist Max Wertheimer and the eminent philosopher of science Adolf Grünbaum. In brief, Polanyi argued that Einstein held the principle of the constancy of the speed of light on intuitive grounds12 and gives this as a historical example for scientific discovery as an insight into the physical reality. Wertheimer, basing on his own conversations Therefore, when Einstein read about these crucial experiments …, their results were no surprise to him, although very important and decisive. They seem to confirm rather than to undermine his ideas13. From this line it is clear that whatever the role of the Mickelson-Morley experiment in the discovery of STR it did not provide the motivation for STR. Grünbaum is the major opponent to this interpretation, and yet he restricted himself to claiming that the evidences available on Einstein’s use or lack of use of the results of Mickelson-Morley experiment are not unambiguous and are inconclusive.14 He argued that in absence of decisive historical evidence we must accept that Einstein must have relied on the Mickelson-Morley experiment. However, the end of the debate on the historical aspect has been provided by, as Cutting puts it, the “definite” study of Gerald Holton for the actual historical influence of the Mickelson-Morley experiment in Einstein’s discovery of STR15. This exhaustive work showed unambiguously that … the role of the Mickelson experiment in the genesis of Einstein’s theory appears to have been so small and indirect that one may speculate that it would have made no difference to Einstein’s work if the experiment had never All this, I believe, should be sufficient to show that the actual historical situation regarding the influence of the Mickelson-Morley experiment was that it had negligible significance and no motivation whatsoever for the developing STR and the revision of the absolute simultaneity in particular. As a final remark a quotation of Einstein’s reply to a direct question, given by Polanyi, is unambiguous: “the … experiment had a negligible effect on the discovery of The second point is a powerful one since it claims that LP is either an empirically originated or empirically justified principle. I will briefly comment on two aspects of this claim. First, de facto, as Polanyi claims, Einstein held LP on “purely intuitive grounds”18. In the same spirit is Wertheimer’s position, who traces down the origin’s of Einstein’s conviction in LP as far as to another thought experiment, namely the one where Einstein was puzzled how would beam of light look like if he chased it with velocity c (the puzzle is that in the context of the classical laws of addition of velocities the beam had to look as electromagnetic field at rest and this would violate LP)19. The second aspect is strictly an epistemological one. It has to meet the claim that since LP was derived from Maxwell’s equations and they express a law of nature it must have been an empirical principle. The objection receives even more strength since Maxwell’s equations figure in Friedman’s model under the empirical part. Much weaker variation is the one that LP has been actually established by the Mickelson- Morley experiment. The latter view is historically mistaken since the experiment has been conducted after the formulations of the equations, in the early 1880 while Maxwell’s paper has been published in 186420. Regarding the epistemological objection one can retreat and say that LP is a subspecies of the inferential a priori. Suitable formulation in this sense is the ultimate a priori proposed by Robert Audi. The ultimate a priori is not self-evident or self evidently entailed by self-evident proposition but it is ultimately traceable to an a priori proposition21. We can trace down LP to the constitutive principles that allows for its’ receiving defined empirical value and thus claim that LP is only weakly a priori in the sense of ultimately a priori. This would allow for LP to be a priori even in the light of its derivability from the Maxwell’s equations. However, I do not think that we have to retreat to this weaker claim when we can argue for stronger a priori for LP. If we take LP as formulated the following way: The speed of light (c) does not depend on the speed of the light source and does not depend on the direction of space for every inertial system, it is clear that this is a physical principle with universal validity. In case of empirical justification we have two possibilities. Either this principle has been inductively justified by some (supposedly great number of confirmation instances) or has been deductively justified following deductive rules of inference from empirical premises. The first option is obviously implausible, since there are not but just a very few candidates for empirical confirmation of LP, most of which available after the formulation of Maxwell’s equations, they could not possibly grant the truth of a statement with universal pretensions. The second option appeals to a possible empirical justification of Maxwell’s equations. However, there is a very important point with respect to the relation between LP and the equations that has to be made. In the present paper I am concerned not simply with LP but with LP as an assumption in the train thought experiment. This additional seemingly insignificant condition makes nevertheless significant difference. Because, as Cutting Einstein began with convictions that (a) Maxwell’s equations are valid and that (b) Maxwell’s equation – and all other laws of nature must have the same form in all inertial systems.22 This means unambiguously that it were not the Maxwell’s equations per se but only the Maxwell’s equations as universally valid for all possible inertial systems that entail the LP as an assumption in the thought experiment. Universal physical validity cannot come from inductive empirical justification. In this sense the justification of LP, whatever it is, it could not be experiential. In the thought experiment however, the true source of the universal validity is a universal principle that Einstein held a priori, the principle of relativity. This opens the way for a priori in a stronger sense. Hence in order to verify the epistemic status of LP as utilized in the train thought experiment we ultimately rely on the principle of relativity, which is the substantial part of the answer to the third objection. At the third point I will again propose two-fold response. This first is the historical level response. Holton’s claim is that the principle of relativity (in restricted sense) has been held by Einstein primarily on non-experimental grounds. Einstein’s own words for that are “intuitively clear”.23 I believe this to be sufficient to enlighten the historical state of affairs regarding the actual epistemic grounds on which Einstein held PR and in particular in the thought experiment. The second line of response is the following: could one possibly hold the proposition If, relative to K, K’ is a uniformly moving co-ordinate system devoid of rotation, then the natural phenomena run their course with respect to K’ according to exactly the same general laws as with respect to K for all inertial on grounds of experiential justification? The claim for universal validity, namely, “all inertial systems” could not be exhausted by any finite number of empirical instances of confirmation. Even few of them would not be sufficient. More, at the time the PR was accepted by Einstein there weren’t any “few” confirmations for this at all. Yet, Einstein held it as true. Yet, it proved historically as a principle of immense importance for the physical science. If we accept that there is a bit of truth in PR this bit has been achieved without relying on any empirical justification for The last principle assumed (though not explicitly in the initial exposition of the experiment) is the classical principle of the absolute simultaneity (AS). We may reformulate this principle using Einstein’s relativistic conceptual framework as follows: If we consider two frames of reference (train and embankment) events that are simultaneous with respect to one of the frames are simultaneous to the other as well. There is one common absolute time for every frame of reference. Different observers agree about simultaneity of events with respect to different frames of reference absolutely. This is the principle that is actually revised through the thought experiment. It is fundamental principle of the classical physics and in the subsequent relativistic paradigm is substituted with the relative simultaneity principle. I will not argue In detail for the apriority of both AS and RS, for the present purposes I will only follow Friedman’s line regarding fundamental constitutive principles of natural science as proposed in Dynamics of Reason and related recent articles. Both principles, in spite of having significant differences still agree in their constitutive role, namely, to define space of empirical possibilities by setting general correspondence between the mathematical part of the theory and the concrete empirical phenomena in such a way that empirical laws could receive empirical meaning at all. As such the constitutive principles could Finally, an important remark has to be done with respect to a recent criticism against the role of thought experiments in science that points in a different direction. The classical debate regarding the status of the thought experiments embraces the claim, in recent times most notably put by John Norton24 that thought experiments are noting but logical arguments in disguise. This objection claims that the result of the thought experiment is a kind of a logical truth. On this view, the result of the thought experiment is conclusion of an argument with the assumptions being the premises of the argument. Premises are taken as grounded in experience. Provided that above we have dismissed successfully with the possibility for the assumptions in the argument to be grounded in experience this objection looses its foundation and starts to look peculiar. First, if the train thought experiment were just logical argument in disguise this would mean that every other physicist with a good sense for logical rules could compose and solve the argument. However, as it is known no one else before Einstein proposed similar argument. Yet, the opponent could claim here that just because Einstein was a rare instance of logical genius only he could reach for the conclusion basing on the available, presumably for every other physicist, premises. This claim does not survive closer investigation though. For one thing, the “premises” in the way they function in the thought experiment were in a sense not available to everyone else. And this is not because the contemporary physicists did not have access to such type of vital information but because no one before i) accepted the principle of relativity and in its domain of validity in the way Einstein did it and ii) no one combined the PR with the LP in this way and, consequently, no one detected the apparent contradiction between them (in the light of the classical rule for addition of velocities). For whatever the contemporary physicists were they were definitely not poorer logicians than Einstein. That mere logical ability is not sufficient in order to reach for the “conclusion” of the argument is apparent from the fact that the rules of combination of (physical) premises are not logical rules. And rules that tell us which ones actually could be premises of such an argument, the rules of providing the premises, also could not be mere logical rules. From logic alone we could not detect a contradiction between the assumptions since we had before that to accept PR and LP as universally valid principles with pretentious for physical significance. Thus, the act of acceptance is not a mere logical act. Also, the proposed identity between the result of the thought experiment and a kind of logical truth would not be sufficient for the purposes of the physicist. The main reason is that the mere validity of such argument will not have any physical force. Only when the argument is sound and not merely valid we would have a meaningful statement regarding the physical world. The soundness of the premises PR and LP however, could not come from logic alone25. Still, if we believe that the reasoning proposed above in defense of the a priori character of the assumptions made in the thought experiment were correct, the opponent loses its claim that they were grounded in experience. The only possibility left would be that somehow Einstein grasped a priori universal principle with physical validity. This, unlike the opponent’s proposal would mean that the thought experimenter has gone beyond the old data available and has received genuine new knowledge about the physical world but not by just following deductive or inductive rules of inference from empirical assumptions26. If we accept Friedman’s model of constitutive framework built up of a priori revisable principles, I believe we can find support for their a priori character in the conception of a priori revision. My claim was that in the case with scientific principles that underlie paradigms of Kuhnian type the aprioristic revision could be done through thought experiments. The role of the a priori revision, as it can be seen from multiple and historically influential thought experiment could be significant in at least two aspects. First, epistemologically to enforce the claim about the apriority of the scientific principles and second, to play substantial role for the shifts in the scientific Thought experiments in science have historically successful role in revising fundamental principles in science. Small number of them can serve as paradigmatic examples of a priori revision of equally a priori fundamental principles of science. For that purpose I argued in defense of the following three claims: first, that some fundamental principles underlying science are a priori. Support for this claim comes from the scientific model proposed by Michael Friedman, which I assume for the present purposes. Second, some thought experiments in science are conducted a priori. As an illustration and support for this claim I proposed reconstruction and analysis of Einstein’s train thought experiment. I argued that it is a historical example of influential thought experiment that has been conducted a priori and which revises fundamental claim of natural science – the principle of absolute simultaneity. If successful this will help to provide defense against basic claim in Friedman’s model that is a target for an empirical criticism, namely that in some cases fundamental principles of natural science cannot be revised empirically but still can be revised a priori. This will help to affirm the status of these principles as a priori and secure them against the possibility to turn out to be empirically false 1 Kuhn, Thomas S. A Function for Thought Experiments in The Essential Tension: Selected Studies in Scientific Tradition and Change. Chicago: University of Chicago Press, 1977, p. 263 2 Benaceraff, Paul Mathematical Truth, in The Journal of Philosophy, 70, 1973, pp. 661 - 79 3 Friedman, M. Dynamics of Reason: The 1999 Kant lectures, 2001, CSLI Publications, Stanford, CA 4 Reichenbach, Hans The Philosophy of Space and Time, Dover , NY, 1958 5 Lukas, J.R. Euclides ab omni naevo vindicatus, British Journal For the Philosophy of Science. 20, 1969 6 Cassirer, Ernst The problem of knowledge : philosophy, science, and history since Hegel, translated by William H. Woglom and Charles W. Hendel., New Haven: Yale University Press, 1950 7 Schiller, F.C.S. Non-Euclidean Geometry and the Kantian a priori, Philos. Review, vol. V, N. 2, p.176 8 Brown, James Robert The Laboratory of The Mind, London, Routledge, 1991, part 1. 9 Casullo, Albert Revisability, Reliabilism, and A priori Knowledge in Philosophy and Phenomenological Research, Vol. XLIX, No. 2, Dec. 1988, p. 198 - 199 10 Reconstruction by Einstein, A. Relativity: The Special and the General Theory, London, Routledge, 1994, pp. 12 11 Comprehensive collection of illustrations for this is available in Holton, Gerald Einstein, Mickelson, and the “Crucial” Experiment, Isis 60 (1969). 12 Polanyi, M. Personal Knowledge. Chicago: University of Chicago Press, 1958, pp. 10 – 11. Also see Cutting, G. Einstein’s Discovery of Special Relativity, 13 Wertheimer, M. Productive Thinking. Enlarged edition. New York: Harper and Brothers, 1959, p. 214 14 Grünbaum, A. The Genesis of the Special Theory of Relativity in “Current issues in the Philosophy of Science”, Feigl, H. and Maxwell, G. (eds). New York: Holt, Rinehart, and Winston, 1961, p.45 15 Cutting, G. Einstein’s Discovery of Special Relativity, Philosophy of Science, Vol. 39, No. 1. (Mar., 1972), p.54 16 Holton, Gerald Einstein, Mickelson, and the “Crucial” Experiment, Isis 60 (1969), p.195 17 Polanyi, M. Personal Knowledge. Chicago: University of Chicago Press, 1958, pp. 10 – 11. 19 Cutting, G. Einstein’s Discovery of Special Relativity, Philosophy of Science, Vol. 39, No. 1. (Mar., 1972), p.58 20 Gribbin. J. Q is for Quantum, Phoenix Press, London, p.281 for Maxwell’s equations and p.288 for Mickelson’s 21 Audi, R. Epistemology Routledge, London, 2003, p.102 22 Cutting, G. Einstein’s Discovery of Special Relativity, Philosophy of Science, Vol. 39, No. 1. (Mar., 1972), p.58 23 Einstein, A. Autobiographical notes, in Albert Einstein: Philosopher Scientist. Schilpp, P. A. (ed.) NY, Harper 24 Norton, John Thought Experiments in Einstein’s work, in Thought Experiments in Science and Philosophy, T. Horowitz and G.J. Massey (eds.) Savage, MD: Rowman and Littlefield, pp.129 - 148 25 For brief but informative discussion see Cutting, G. Einstein’s Discovery of Special Relativity, 26 This point is made by James Brown in his Thought Experiments, in A Companion to the Philosophy of Science, W.H. Newton-Smith (ed.), Blackwell Publishers, 2001, pp. 529, with respect to Galileo’s falling bodies thought

Source: http://www.personal.ceu.hu/students/03/Boris_Grozdanoff/A%20priori%20revisability%20and%20Thought%20Experiments%20in%20science.PDF