13 pages

Underdetermination, scepticism and realism

Please download to get full document.

View again

of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
This study aims to articulate and compare the structure, presuppositions and implications of two paradigmatic sceptical arguments, i.e. arguments from underdetermination of scientific theories by observational data (UA) and Cartesian-style arguments
  1 Underdetermination, scepticism and realism 1   LADISLAV K  OREŇ  Department of Philosophy and Social Sciences. Faculty of Arts. University of Hradec Králové.  Náměstí Svobody 331 . 500 03 Hradec Králové. Czech Republic Abstract This study aims to articulate and compare the structure, presuppositions and implications of two paradigmatic sceptical arguments, i.e. arguments from underdetermination of scientific theories by observational data ( UA ) and Cartesian-style arguments ( CA ) invoking sceptical scenarios of severe cognitive dislocation. Although salient analogies between them may prompt one to think that a unified diagnosis of what   is amiss with them is called for, it will be argued that this may be a false hope, if those analogies do not underwrite a complete homology. That said, possible parallels of one promising anti-sceptical exposure of CA  are pointed out for the case of UA , which conspire together to render the problem of underdetermination less threatening than it could at first appear. Keywords: evidence; underdetermination; realism; scepticism A venerable tradition portrays science as aiming at explaining observable patterns through deeper causal mechanisms that, for all their hiddenness, are “ really out there” in  the natural world. Scientists, striving to “cut nature at its joints”, work hard to amass data via observation or controlled experiments to make it likely that their theories give increasingly more accurate and comprehensive accounts of the ready-made nature out there, including, prominently, its hidden joints. What is more, a modest epistemological optimism seems warranted. At least with our mature and impressively successful scientific theories, evidence accumulated over time warrants belief in at least their approximate and  partial truth. This familiar picture of scientific theorizing has come to be called  scientific realism  (henceforth: SR ). Despite its initial attractions, it has mobilized many able philosophical opponents, as the ongoing controversy over SR  vividly testifies. Among many objections that have been raised against it, underdetermination-style arguments  ( UA ) stand out as advancing a particularly serious challenge to its epistemological optimism, on the ground that observational data −   no matter how massive −  might be equally accommodated by empirically equivalent theories that make incompatible claims about hidden contents and structure of the natural world. 2  In this respect, as many thinkers have noted, there are salient analogies between UA and the classical Cartesian-style sceptical arguments  ( CA ). The latter threaten to undermine even our most common-sense claims to knowledge of the world around us, on the ground that the whole of our experiential evidence −  past, present or future −  can be accounted for  by experientially equivalent hypotheses of severe cognitive dislocation 3  that make incompatible claims about the contents of the outside world (and, in particular, about the causal srcin of our sensory experience). Both UA  and CA  aim to establish a radical sceptical thesis on the basis of  prima facie  appealing  premises. They have accordingly provoked several attempts to diagnose what is amiss with them but 1  W ork on this study was supported by the grant GA ČR,  Jistoty a skeptický problém , n. P401/12/P599.   2  More precisely, this applies to so-called strong underdetermination arguments, according to the distinctions introduced in Section 1.2. 3  To use an apt term coined by Wright (2002), (2004).    2 none has commanded wider acceptance. However, I believe it worthwhile to explore in some depth and detail the possibility that suggests itself here, namely, that aforementioned analogies between UA  and CA  invite analogous or even structurally unified anti-sceptical diagnoses. To prepare the ground for this issue, I first situate UA  in its proper dialectical context, explaining its motivation, structure and  possible ramifications for the controversy over SR . Having reconstructed CA  as an underdetermination  problem of a sort, I then compare both puzzles highlighting their analogies as well as their specifics. On this basis I sketch an anti-sceptical exposure of CA that I find promising, but which, owing to the specifics of UA , does not straightforwardly carry over to it. This, I argue, should come as no great surprise, because analogies between UA  and CA  do not amount to full homology. That said, I also try to show that instructive parallels of that exposure tactics can be discerned in the case of UA , which might conspire together to render the problem of underdetermination less threatening than it could at first appear. 1.   Underdetermination problem and the controversy over SR 1.1  Sources  The relevant history of UA is usually traced back to Pierre Duhem`s path-breaking critique of the hypothetico-deductive model   of confirmation of scientific hypotheses. 4 According to   this model, once a hypothesis  H is proposed, one may go on to confirm or falsify it by checking its observational consequences concerning occurrence of phenomena under specified circumstances. If a phenomenon does not occur as predicted by  H  ,  H   should be taken as falsified, hence rejected and replaced by another hypothesis. If, on the other hand, phenomena occur as  H   predicts,  H   gets (increasingly) confirmed. But things do not quite work this simple way  –   or so Duhem argued. First, scientific hypotheses have observational implications only within a larger theoretical context or system  –   if conjoined with auxiliary hypotheses and background assumptions about initial conditions, experimental setting, instruments, etc. Let us call this enlarged system  H+A ,  A  incorporating whatever relevant auxiliary hypotheses or background assumptions. Second, Duhem pointed out that observational consequences do not  –   by themselves or together with logical principles  –   determine confirmation or falsification of the hypothesis  H  . Thus a false observational consequence does not automatically falsify  H  , because it is derived from  H+A  so that what it shows is at most that  H   is false or  A  is false. Accordingly, we may save the data by keeping  H   and making appropriate revisions within  A , or the other way round. Still, data plus logic alone do not dictate what to do. Second, even if we assume that all so far checked observational consequences of  H   are true, this does not confirm  H   as likely to be true, as there is always a rival hypothesis  H* that, conjoined with its own  A* , accommodates all the data equally well,  but makes incompatible claims about the underlying reality. So data plus logic alone do not dictate a choice of  H   over  H*.  Summing up, Duhem`s holistic thesis says that larger theoretical systems, not single hypotheses, are subject to empirical test. Its corollary then seems to be that observational data together with logic are not enough to determine (a) what item is falsified vis-à-vis recalcitrant data and (b) which of two rival but empirically equivalent theoretical systems is more likely to be true, hence worthy of choice. A few decades later, Quine contributed    by two generalizations of Duhem`s claims. First, in his assault on empiricism he argued that the system (theory) to be tested against sensory experience is a whole web of inter-linked beliefs including also logical and mathematical claims, so that they too are eventually up for revision if the system faces recalcitrant data. 5  Second  ,  whereas Duhem had in mind empirical equivalence  –   hence underdetermination  –    relative to all available data (already checked observational consequences), Quine radicalized UA  to mean underdetermination of whole (perhaps even comprehensive) physical theories relative to the totality of all possible data. Importantly, the 4  See Duhem (1914/1954). Also Poincaré`s (1905/1952) famous considerations about alternative yet empirically equivalent  physical geometries of space-time are often quoted as an important milestone. 5 Quine (1951).  3 strong version of UA  excludes the possibility that future observations could eventually distinguish  H   and  H* . 6  It is arguably easier to come up with historical examples of the weak underdetermination. One notorious example put forward by Duhem himself is the transient empirical equivalence of the Ptolemaic and the Copernican system: both were able to accommodate all observed astronomical data at the time of appearance of  De Revolutionibus.  Hence, at the time, available empirical data did not support one system over the other as more likely to be true. We shall have an occasion to see that the situation is more delicate with the strong underdetermination, there being fewer genuinely interesting examples of alternative theories that are both incompatible and empirically equivalent relative to all  possible data. Let it be said that the clash of rival accounts of quantum mechanics  –   viz. Copenhagen interpretation versus Bohmian interpretation in terms of hidden parameters  –   is often cited as a  particularly up-to-date example of the strong underdetermination. Also Poincaré-style examples of rival space-time cosmologies based on different geometries seem to supply instructive examples  –   viz. a space-time theory based on the classical Euclidean geometry and a space-time theory based on the non-classical Riemannian geometry. 7  The idea behind is that all possible predictions about trajectories of physical objects made by one theory can be mimicked by the other: thus, curved trajectories of objects predicted by the space-time theory based on the Riemannian geometry can be mimicked by the space-time theory based on Euclidean geometry if we introduce into the latter theory extra-forces acting on objects as well as rulers. The two alternative theories seem equally underdetermined by all  possible data. 1.2   The role of UA in the controversy over SR It is the strong version of UA that will concern me in what follows. For one thing, it seems to be  potentially more devastating. For another, it is structurally analogous to the Cartesian-style sceptical argument with which it will be compared in due course. To reconstruct UA  in its strong form, a few distinctions are in order. 8   C belongs to   the testable basis   of a theory T   iff C   is a claim about    phenomena and T allows us to infer C  . T and T* are   empirically equivalent theories iff they have exactly the same testable basis. T and T* are   rival empirically equivalent theories  iff they are empirically equivalent theories that make incompatible claims about unobservable items. T and T* are evidentially indistinguishable theories iff they are equally well supported with respect to all possible evidence so that any possible observation that contributes to confirming (or disconfirming) one (to a given extent) confirms (or disconfirms) equally well the other (to the same extent). With this terminology in place, it is easy to formulate UA  as involving the following steps: 1.   For any scientific theory T   there is an alternative theory T*  such that the two are rival empirically equivalent theories. 6  Okasha (2002) argues that the strong version of UA   is hard to reconcile with Quine’s own holism: either a theory T   is incomplete/local and then it is always possible that future evidence will distinguish it from currently equivalent rival theories (thus we have at most the weaker Duhemian version of UA ); or the argument has to be formulated at the level of global (complete) theories  –   but then we do not quite know what would it be like to have such a theory. See also Hoefer  –   Rosenberg (1994) who distinguish local and global theories and, accordingly, local and global underdetermination, holding that only the latter is a potential threat. 7  See Poincaré (1905/1952). Reichenbach (1958) formulates a famous generalization of this procedure. 8  There is already an extensive literature on underdetermination arguments and their relation to the issue of SR . For different  perspectives on it, I refer the reader to Laudan  –   Leplin (1991), Earman (1993), Hoefer  –   Rosenberg (1994), Psillos (1999), Stanford (2001), Devitt (2002), Okasha (2002), Norton (2008).  4 2. Empirically equivalent theories are evidentially indistinguishable. 3. So no possibly relevant evidence supports T as more likely to be true than T*. 4. So no possibly relevant evidence justifies one in holding T   to be more likely to be true than not. 5. So one is never justified in holding T   to be more likely to be true than not. Simple and abstract as UA is, its ramifications for SR  are potentially far-reaching. To appreciate this, we should note that SR , as commonly conceived, involves a mixture of essential ingredients  backing up the claim that scientific theories aim to provide approximately true descriptions of the objective sort that are apt to be warranted by empirical evidence: Semantic   commitment  : sentences comprising theories  –   including those apparently about unobservable items  –   are to be taken semantically at face value, that is, as purporting to refer to and describe what they  prima facie  appear to refer to and describe (truly or falsely). Ontological commitment  : there is some stuff out there for them to be approximately true descriptions of  –   including entities, structures, etc., underlying phenomena and their regularities.  Independence commitment  : that stuff is what it is  quite independently of any linguistic, conceptual or epistemic means that can be employed in positing, classifying, describing or confirming them.  Epistemological    commitment  : empirical evidence can place us in a favourable position to hold (at least our best current) scientific theories to be (at least nearly and partially) true descriptions of the world  –    in both its observable and non-observable parts.   What UA  urges on us is that on a realistic construal of a scientific theory T     –   as concerned with a largely unobservable and ready-made world out there  –   there is a gap between the two that cannot be  bridged by empirical evidence, however varied and massive. Clearly, observational data do not entail theories, there being always room for alternative theories equally accommodating all the data in terms of incompatible claims about hidden structures. But, if UA  goes through, the data do not determine scientific theories inductively either  –   by lending them at least a reasonable degree of support  –   since empirically equivalent theories are held to be evidentially indistinguishable. Thus, if SR  with all its essential ingredients holds, then, epistemically speaking, the natural world out there seems once and for all lost! UA , I submit, can be seen as an improvement upon more traditional and controversial arguments informed by the positivist stricture to the effect that what transcends experience/observation is beyond the province of knowledge  –   to claim anything about unobservable items is to outstrip the bounds of evidence that only licenses warranted claims about the world out there. In response to arguments of this calibre, friends of SR  would do well to challenge the positivist stricture behind them  –    only what is observable or reducible to observable is knowable . Even if is it granted that scientific hypotheses and theories are evidentially controlled by the data of observation and experiment, this reasonably looking empiricist regulative does not entail the positivist stricture. For w hy, realists ask, shouldn’t theories  positing unobservable items be liable to empirical test indirectly, that is, via what they (together with auxiliary hypotheses and background assumptions) imply about phenomena? Why, then, shouldn’t we regard the fact that some theories fare excellently in predicting and explaining phenomena  –   indeed, much better than alternative theories  –   as giving considerable, if indirect, empirical support also to  5 their claims about unobservable items? 9  So, without a more powerful line of argument against the  possibility of warranted claims about unobservables, realists may feel relatively safe. Yet, UA   promises to give its opponents precisely such a principled argument, because it directly challenges the crux of   SR    –   the claim that (our mature and impressively) successful scientific theories are likely to be true (nearly and partially). Although UA  directly attacks the epistemological commitment of SR,  it could be used in drawing significant metaphysical or semantic conclusions. Indeed, scientific scepticism wholeheartedly embracing its negative conclusion is a rare reaction to UA . 10  Thus, constructivists, projectivists or internal realists would argue that realities must be theory-dependent after all. Differences aside, their goal is to close the gap between theories and reality by evidentially constraining truth-conditions or urging some form of conceptual or epistemic dependence of realities on theoretical frameworks employed to capture them. 11  Others would contend that sentences apparently about unobservables should not be interpreted literally but, as instrumentalists would have us say, as computational means (perhaps useful fictions) in the service of correlating phenomena (input-data with output-predictions). Either way, the problem of underdetermination can be mitigated, though on pain of incurring commitments which are far from unproblematic. There are other “non -re alist” strategies, to be sure. Since  this study is not primarily concerned to review them, suffice it to say, for this moment, that the friends of SR  have several lines of response available to them to show that, while UA  compromises some of its premises, it does not reduce SR  to absurdity, because SR  does not license them (or at least not all of them). I will mention some of those responses later, after having compared UA  with CA . But let me now turn to the classical Cartesian-style sceptical argument and its subversive potential. 2.   CA as an underdetermination argument It is well known  –   even well worn  –   that Descartes bequeathed to us a sceptical problem, 12  whose ghost is still with us today, in spite of the fact that he thought to have disposed of it. In its restricted form, targeting only empirical claims about the world out there, the issue arises if we take it that at least some of our best common-sense or scientific claims  –   e.g. claims about medium-sized objects observable by unaided senses under optimal conditions  –   are certain or at least likely to be true. However, the fact that we confidently take the world to be a certain way does not yet guarantee that the world really is that way. So what basis, if any, can we have for placing such confidence in our most cherished empirical claims? Sensory experience we take ourselves to have made with the outside world seems to be,  prima facie at least, the most plausible candidate. However, if experience consists of a stream of appearances  –   internal phenomena  –   that might just as well occur when we are being subject to a severe cognitive dislocation (e.g. devious delusion or compact dream), it does not seem to lend a reasonable degree of support to our empirical claims  –   still less an assurance that the world out there is, by and large, the way we take it to be. As Descartes pointed out, apparently far-fetched claims about the contents and structure of the world can accommodate the totality of given appearances  –    9   The “No miracle” argument –   sometimes dubbed the master argument for SR  –    makes this point dramatically: wouldn’t it  be a most remarkable coincidence  –   indeed, a mystery  –   if our best current theories, successfully predicting and explaining regularities of phenomena in terms of hidden structures/mechanisms, were not onto something as regards the hidden nature of the world? See Putnam (1978) and Boyd (1983) for influential defences of the claim that only SR  does not make the apparent success of science one big mystery left without any explanation. 10  A notable exception, however, is Van Fraassen`s constructive empiricism  (1981) to which I shall shortly return later. 11  Here we may assign Kuhn`s (1962/1996)  paradigm-style account of the structure of scientific change with its corollary of conceptual incommensurability of scientific frameworks employing different classificatory and epistemic styles (based on different exemplars  establishing the range of problems to be solved and the methods of solving them). On this view, realities of concern to scientists are always immanent to scientific frameworks  –    different frameworks coping with different realities. Though in many respects very different, Putnam`s internal realism  (1981) may also be reckoned here. Positions of this sort need not deny that scientific theories are concerned with observable as well as unobservable aspects of the world and that they aim to give us, in a way, accurate accounts of both. What they in different ways question is theory-independence of realities studied by science, unconditioned by conceptual and epistemic means that constitute theories or theoretical frameworks. 12  For an srcinal statement of the problem see Descartes (1640/1996).
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!