domingo, 13 de noviembre de 2011

Ontic Structural Realism and Scientific Realism: On Ladyman and Ross, Sellars and Brassier


On Ladyman and Ross, Sellars and Brassier

   Ladyman and Ross go a long way in debasing Kuhn-inspired relativisms about science by showing how discontinuity in scientific theories at the level of content is underwritten by continuity at the level of structure. In doing so, they also build on the epistemological (rather than ontological) valence of the distinction between observables and unobservables as posited in theoretical physics. Their brand of ontic structural realism predates on the semantic approach to theory-modeling which situates the relationship between mathematical structures as primary as opposed to the partial-interpretation of theoretical terms on the basis of observables (the 'syntactic' approach advocated by Carnap). Similarly to Badiou's deflation of the empiricist notion of content (the 'third dogma of empiricism' he criticizes apropos Quine in The Concept of Model), L&R seek to undermine the ontological valence of the entities postulated within the manifest image ("Scholastic" and "neo-Scholastic" metaphysics included) and its reference to middle-sized objects and properties, the better to assert (again, like Badiou or Meillassoux) the reality of primary mathematical properties. These, however, do not demand commitment to the positive ontological status of imperceptible particles; the  underdetermination of objects obtained in avowing the reality of processes remains ontologically agnostic about any entity in current science. In this regard, L&R seek to anchor their realism in scientific predictive success (the so-called 'no-miracles argument'), while at the same time having leverage to resist the underdetermination of realism by instrumentalism in consideration of theory-change (the so-called 'underdetermination problem'). The latter challenge applies even to positions such as Van Fraassen's constructive empiricism. Below I will explain what I find most problematic about this restriction of the realist commitment to structure, but provisionally we can gauge that if Meillassoux's metaphysical argument against the frequentialist implication is correct, then the predictive success of science is no less miraculous than its potential disruption by the sudden change of these laws ex nihilo. Intra-systemic predictability at the level of local laws provides no less a secure foundation for realism than unpredictable anomalies (in Kuhn's sense) force us into accepting pragmatic instrumentalism about science.

Of course, L&R's principled reluctance to make metaphysics refractory to empirical science would no doubt resist the very epistemological coordination between thought and reality in terms of a correlation between subject and object: ontic structural realism is precisely meant to revise the epistemological framework to dislodge traditional representational analogy in favor of structural isomorphy (or homomorphy). Yet what I find most unpersuasive so far about the book is the early defense in Chapter I of scientism, since in a certain sense it seems to me to rest, on similar grounds to the Churchlands, on an opaque sense of super-empirical virtue tethered to a rather thin pragmatism. Brassier above all has shown the deficiencies in appeals to the super-empirical in Churchland's  neurocomputational idealism through pragmatism, much like Badiou criticizes Quine's naturalism for reifying science on alleged pragmatic grounds (again, this point is belabored in The Concept of Model). In L&R's account, the metaphysical subordination to science, and physics in particular, is entirely sketched on pragmatic grounds under what they programmatically label the "Primacy of Physics Constraint", or PPC in short. The following quote provides the basic position:

"Special science hypotheses that conflict with fundamental physics, or such consensus as there is in fundamental physics, should be rejected for that reason alone. Fundamental physical hypotheses are not symmetrically hostage to the conclusions of the special sciences.

This, we claim, is a regulative principle in current science, and it should be respected by naturalistic metaphysicians. The first, descriptive, claim is reason for the second, normative, one." (Pg. 45)

This reinforces the earlier, more general "Principle of Naturalistic Closure", which states that:

"Any new metaphysical claim that is to be taken seriously at time t should be motivated by, and only by, the service it would perform, if true, in showing how two or more specific scientific hypotheses, at least one of which is draw from fundamental physics, jointly explain more than the sum of what is explained by the two hypothesis taken separately, where this is interpreted by reference to the following terminological stipulations:

Stipulation: 'scientific hypotheses' are understood as hypotheses that are taken seriously by institutionally bona fide science at t

Stipulation: A specific scientific hypothesis is one that has been directly investigated and confirmed by institutionally bona fide scientific activity prior to t or is one that might be investigated at or after t, in the absense of constraints resulting from engineering, physiological, or economic restrictions, or a combination, as the primary object of attempted verification, falsification, or quantitative refinement, where this activity is a part of an objective research project fundable by a bona fide scientific research funding body

Stipulation: An objective research project has the primary purpose of establishing objective facts about nature that would, if accepted, on the basis of such a project, be expected to continue to be accepted by inquirers aiming to maximize their stock of true beliefs, notwithstanding shifts in the inquirers practical, commercial, or ideological preferences" (Pg. 38)

Notice the appeal to the communitarian 'consensus' evoked here, which is doubtlessly strange considering that L&R later go on to attack scientific realism (which they distinguish from their own OSR) on the grounds that it is dependent on subjective matters of consensus and pragmatic factors, e.g the disposition of specific scientists or the scientific communities. Their arguments against these accounts in grappling with the problem of the reference of theoretical terms are of great value.

Yet the avowal of scientism is there transparently tethered to an explicitly pragmatic, normative subordination of metaphysics to science, without any further epistemological labor or explanation. Although I endorse the idea of making metaphysics continuous with science, as well as their arguments for structural realism in order to account for theory change, it seems to me that the early stages of the book do not advance a sufficiently robust account about the rational necessity for making science metaphysically authoritative, failing thus to clarify the position from which the ontological prerogative of science derives. Perhaps this is where a Sellarsian approach has something to contribute still, specifically insofar as it insists on the necessity of a methodological dualism keeping the normative-register of the manifest image as a necessary condition of possibility for the epistemic entitlement endowed to science, and for the logic of revision itself. In this regard, perhaps the question concerning the ontological status of unobservables in L&R's account vis a vis their discussion of Van Fraasen's constructive empiricism makes my concern particularly salient. I would quote the following passage from the book:

"Opponents of scientific realism, such as Van Fraassen, deny that the local defenses of realism about specific unobservables are compelling, arguing that they can in each case be reinterpret in pragmatic terms as inferences to the empirical adequacy of the explanation in question, plus a commitment to continue theorizing with the resources of the theory... If unobservable entities merely happened to be around when certain phenomena were occurring then their presence would not be explanatory. Hence, Van Fraassen argues, scientific realism relies upon some kind of metaphysical theory of laws of nature, singular causation, or essential natures. For Van Fraassen, this means it ultimately rests on explanation by posit. Here we reach an impasse with the scientific realism insisting on the need for explanations where the antirealist is content without them." (Pg. 74)

And yet, if I understand Van Fraassen correctly, his point is not that explanation is no good, but that explanation by posit appealing to the abovementioned metaphysical items (laws of nature, singular causation...) is what is left unexplained qua posited by the scientific realist, and as such they would require a justification not circularly defined terms which would simply presuppose the pragmatic requirement that we ought to continue using the theory in dispute. At that point Van Fraassen seems to think the 'realist' wager becomes fatally undermined by its tacit pragmatic instrumental appeals. Similar to Meillassoux's hijacking of the correlationist argument to unearth a tacit realism in the form of the factial, Van Fraassen hijacks the scientific realist to unearth a tacit instrumentalism in their argument. Yet as we have seen L&R do not, despite their claims to the contrary, escape from instrumentalism by witholding epistemic commitment about individuals, and circumscribing realism to structural properties. For whether the prescription to philosophize in continuity with contemporary physics is wed to a notion of individuated objects or to merely relational structural properties is strictly speaking irrelevant to Van Fraassen's point, which is that one obviates explanation by normative prescription/proscription, and as such cannot but be instrumental on those grounds. Even if OSR can explain theory change while remaining metaphysically agnostic about individuals, the same questions raised apropos the latter in classical scientific realism reappear in OSR with regard to the prescription to endorse the ontological valence of structural properties. And since this endorsement seems to be postulated in pragmatic terms, in principle rather than established by argument, it is not clear Van Fraasen's hijacking of the scientific realist argument to unearth its tacit circular appeals to authority is not reproduced in OSR. Pragmatism is idealist in tenor since it subordinates the reality or ontological valence of a given set of postulates, in this case scientific individuals and structure, to the normative injunction to 'do science'.

In this regard, it is important to notice that structure fares no better than individuals in L&R's account: both are finally undermined by the pragmatic claim that physics ought to be metaphysically endorsed since it cannot be rendered explanatorily virtuous and non-miraculous without assuming their reality in principle. If regularity provides grounds to motivate realism this cannot be surreptitiously postulated in order to avoid its instrumentalization, since in doing so one ends up doing just that, whether we tether our metaphysics to structure or individuals. Unyoking scientific modeling from observation and inference fares no better than mathematical structure if it must be assumed relative to the unexplained necessity to endorse realism on the basis of the scientific capacity to explain and predict phenomena, i.e. the latter remains just as susceptible to instrumentalization as the syntactic scientific realism about individuals was susceptible to deflation into empiricist constructivism. Now, ironically, L&R themselves attempt to return Van Fraassen the favor by claiming that constructive empiricism must commit to a minimum metaphysical endorsement of objective modal relations; i.e. it must ground its assertion of the probable-predictable adequacy of phenomena to theoretical posits in the reality of their possibility or actuality, even if it remains agnostic about their other properties. Thus L&R claim:

"If science tells us about objective-modal relations among the phenomena (both possible and actual), the occasional novel predictive success is not miraculous but to be expected." (Pg 153)

However, this misses the deflationary point raised by Van Fraassen, because to be a realist about modal objectivity is not sufficient to be a realist about the mind-independent reality, or about scientific phenomena in any orthodox sense. We know, at least since Kant, that it is perfectly possible to avow the causal efficacy of phenomena as constrained to our experiential field, just like in Hegel's objective idealism the ontological determinations that yield objectivity do not for that reason escape idealism. Thus irrespect6ive of whether pragmatism must be underwritten by some metaphysical commitment or other, the point, which Brassier stresses brilliantly apropos the Churchlands, is that they cannot but be complicit with an idealist miraculous congruence between thought and being. This follows transparently once we realize that just like Kant reactivates the valence of causal efficacy at the price of inflecting objectivity as a function of subjective synthesis, L&R operate under the frequentialist implication whose motivation is the 'no-miracles' argument. The latter is meant to justify the endorsement of naturalist metaphysics. This move, contrary to its pretension, relativizes the explanatory demand of scientific practice to the pragmatic concern to explain predictive success, rather than to legitimate the relation between our concepts and the objects presumably existing independently of the former. Both Van Fraassen and L&R may endorse modal objectivity, but since both finally accept a prescriptive rather than explanatory legitimation for the valence of science, it is the constructive empiricist idealization Van Fraassen openly accepts which seems to be tacitly presupposed by L&R's OSR, and not the latter's realism that is to be found in the former. At a loss for an epistemological account which distinguishes mathematical relations from properly physical structure, L&R simply assume a miraculous congruence between thought and the real through physics, supported by their two founding principles. 
Now, consider L&R's appeals to recent developments in GR and QM which attempt to ground objectual relations in terms of diffeomorphic transformations, nested in an explanation of symmetry understood in group theory as inclusion to equivalence classes with the appropriate rules for preservation. This allows one to explain structural continuity in purely formal terms. It endows the OSR to ground objective modality in structural symmetry. But how does the latter mathematical structure relate to physical structure at all? What relation holds between the two, i.e. what epistemic grounds solicit the postulate of the mind independence of the phenomena described by structure, once we have disowned appeals to the noumenal or qualia? This, again, L&R fail to do, folding back on their convenient prescriptive principle PNC, at a loss for any justification. Thus if OSR is indeed a realism in any way about science it is because the objective modal structure it ontologically asserts is in fact separate from the relational mathematical expression and yet indistinguishable from it all the same. This point is made by Van Fraassen (2006) and it dramatically entails that OSR cannot properly explain what separates mathematics from physics, they can only postulate it. In fact, they have no shame in admitting this much:

"When theories are empirically adequate they tell us about the structure of the phenomena and this structure is (at least in part) modal structure. However there is still a distinction between structure and non-structure. Merely listing relations among locators does not state anything with modal force. Therefore it doesn't specify structure in our sense and it isn't yet scientific theory as we've defended it. Physical structure exists, but what it it? If it is just a description of the properties and relations of some underlying entities this leads us back to epistemic structural realism. What makes the structure physical and not mathematical? This is a question that we refuse to answe. In our view, there is nothing more to be said about this that doesn't amount to empty words and venture beyond what the PNC allows. The 'world-structure' just is and exists independently of us and we represent it mathematico-physically via our theories." (Pg. 158)

 The emergent result seems to be, quite predictably, either a kind of mathematical realism or neo-Pythagoreanism (Maddy, Rednik) purported by mathematicians themselves, or philosophers of science now insisting, like their repudiated Continental 'mystics', on an unobjectifiable excess to structure grounding the realist locus for naturalist metaphysics; which now starts sounding a whole lot more like Heidegger than Hegel, if not Scheller. Take the following passage which sums up the ordeal:

"Of course, all the considerations from physics to which we have appealed do not logically compel us to abandon the idea of a world of distinct ontologically subsistent individuals with intrinsic properties. As we noted, the identity and individuality of quantum particles could be grounded in each having a primitive thisness, and the same could be true of spacetime points. What we can establish is that physics tells us that certain aspects of such a world would be unknowable... On our view, things in themselves and qualia are ideal wheels in metaphysics and the PPC imposes a moratorium on such purely speculative philosophical toys... we take it that such a gap between epistemology and metaphysics is unacceptable. Given that there is no a priori way of demonstrating that the world must be composed of individuals with intrinsic natures, and given that our best physics puts severe pressure on such a view, the PNC dictates that we reject the idea altogether." (Pg. 154)

If our considerations hold here the PNC and PPC both merely aggravate the gulf between epistemology and ontology rather than palliate it. At a loss for justification beyond the super-empirical call for theoretical 'unification' restricted to scientific practice and their instituional approval, both principles are laid to rest on the pragmatic grounds criticized above, and end up soliciting the much repudiated invocations of the 'noumenal' and 'qualia' that L&R associate with philosophical spooks; the specter that Hegel had long since warned against: the uncanny coincidence of all merely 'contingent' (pseudo) philosophical accounts which obviate dialectical necessity by subordinating themselves to the empirical, as well the reification of abstraction in the form of a pure non-conceptual externality, a pure 'thisness' or immediacy intractable by conceptual means. This is yet another dimension in which the kind of pragmatic idealism bolstered by so-called 'hardcore' philosophers who claim to be continuous with science, following Kant, end up enacting, like Meillassoux diagnoses, a Ptolemaic counter revolution of sorts. Although L&R seek to escape this mystical evacuation of realism into a kind of neo-Kantian endorsement of noumenal ineffability, they end up bolstering it through their pragmatism.

Perhaps this is where the dialectical interplay between the scientific and manifest images advanced in the classical Sellarsian account can be put to work in some sense, by securing the relation between the two relatively autonomous registers (the conceptual-normative, and the natural-causal), and so avoiding the gratuitous demand for scientism which renders it epistemologically, if not metaphysically, rest on dubious grounds. Here the question would be to track the dialectic of concept revision by describing the porous frontier between the observable and the unobservable, and thus between the reviseable manifest register, still tethered to the reality of objects, as well as the scientific register's description of both elementary imperceptibles and non-perceptual structural properties all the same. Below I will indicate why Brassier particularly underlines the Sellarsian endorsement of secondary-properties as ontological valences.

The other point which seems of paramount interest is of course going back to the semantic-approach for theory modeling which reactivates the questions raised by Frege and Hilbert, and whether such an approach can be satisfactorily representational in a rigorous sense by means of structural isomorphy without having to invoke the privilege of intuition (in this regard Badiou seems to have taken the side of Hilbert in denouncing any notion of empirical 'content' as grounding structural isomorphy between domains and axiomatics, whereas the scientific realist- empiricist Sellarsian might want to salvage a role for sensibilia in some form or other). I take it that the revisionist naturalism endorsed by Brassier is an attempt to reconcile the rationalism from the Badiouean/Hilbertean deflation of self-grounding intuitions by underlining the relational autonomy of our conceptual economy, while insisting along Sellars' naturalism that an ontological role can be preserved for sensa as long as the latter are understood as pertaining to the autonomous domain of causal efficacy, and so as mediated processes themselves. This would simultaneously revoke the privilege of the manifest image to render scientific postulates of non-phenomenologizable content subordinate to perception (thus against instrumentalizations of science in terms of analogical modeling or pragmatic postulation), while at the same time explaining how sensibilia ontologically conditions conceptual mediation in resolutely non-conceptual terms; that is, in natural-causal terms, as neurophysiological processes. Furthermore, Brassier deems this as a crucial move in order to salvage the epistemic priority of science which, beyond purely mathematical structure, is capable of remaining anchored in the world. At this juncture the idea of 'real patterns' might be metaphysically useful, even if epistemologically still on dubious grounds. The problem with a blunt realism about ontic structure with a physicalist bent is that it adjudicates the privilege of science while remaining open to the kind of instrumentalist-pragmatic seizure of the kind Van Fraassen proposes, in which case we fare no better than Quine or Churchland in grounding the metaphysical labor. I take it this is why Brassier considers the unavoidable juncture of epistemology and metaphysics to remain the problematic moment for thought, and the one in which Sellars can help.

Correspondence From Ray Brassier on this Issue
I share your reservations about their attenuated conception of explanatory virtue and their pragmatic justification for scientism. Ultimately, L&R are too willing to throw out the rationalist baby with the metaphysical bathwater. What is frustrating about the book is they nowhere address the obvious Sellarsian-rationalist rejoinders to van Fraasen's empiricism; which is frustrating given that there was a substantial debate between Sellars and Van Fraassen on this very issue back in the 1970s (and van Fraassen was Sellars' student). In fact, there is only one reference to Sellars in the whole book, at the very beginning of Chapter 1 when they cite approvingly from "Philosophy and the Scientific Image of Man". But one can imagine a Sellarsian critique of OSR given Sellars' remarks in his 1966 critique of Feyerabend "Scientific Realism or Irenic Instrumentalism": The idea that the framework of common sense has a rock bottom does not require that this rock bottom consists of sense-impressions. The framework of common sense is a framework of (among other things) colored physical objects extended in space and enduring through time. And while objects which are red on the facing surface have the power to cause normal observers in the standard conditions to have sense impressions of red, this is incorrectly taken to mean that the physical property of being red on the facing side is to be analyzed in terms of this causal power.

As Berkeley, Kant, and Whitehead, among others, have pointed out, physical objects cannot have primary qualities only for structural and mathematical properties presuppose what might be called content qualities. And unless one falls into the trap of thinking of the framework of physical objects as a common sense theory evolved with unconscious wisdom to explain the manner in which sense data occur, it will scarcely do to say that the content qualities of physical objects are conceived, by a common sense use of analogy, to be the physical counterparts of the qualities of sense data (i.e. to play in the realm of physical things the content-role played in sense data by sense qualities). For, if the conceptual space of common sense physical
objects is underived, their content qualities must be directly rather than analogically conceived, for it is only in terms of perceived, and therefore conceptualized, qualitative difference that form and content can be distinguished.

1. The abandonment by scientists of the conceptual framework of common sense physical objects would involve either the abandonment of the conceptual space of color tout court, or the retention of this conceptual space as it reappears in its analogical offshoot, the conceptual space of sense impressions. The latter would be cut off from its foundation and left to wither on the vine. In either case, the conceptual space of the qualities of sense (secondary qualities) in one use of this phrase) would disappear from the public observation base of science. It would enter science only in linguistics, in the
study of the structure of the language of non-scientists and of scientists only to the extent that their sense impression talk
continued to reflect the pre-revolutionary framework of common-sense physical objects.

2. Only when the conceptual space of sense impressions has acquired a status which is not parasitical on the framework of common sense physical objects. In other words, only with the development of an adequate scientific theory of the sensory capacities of the central nervous system could the framework of common sense be abandoned without losing conceptual contact with a key dimension of the world. (Scientific Realism or Irenic Instrumentalism, pp. 175-178)

The exceptional strength of Sellarsian rationalism lies in its reconciliation of realism and empiricism: neither manifest objects nor unobservable posits are merely "useful fictions": one severs the link between the sensible and the intelligible at one's peril. By the same token, the notion of "real patterns" seems to me too weak to bear the burden of adjudicating between appearance and reality, or to bridge the gap between epistemology and metaphysics.


So, to sum up:

The view that philosophers would benefit from learning actual science does not transparently entail philosophical sobriety, nor does it overcome by itself the critical injunction. Let us disambiguate:

1) The relation between metaphysics and science is not only ontological, or of ontological choice, but fundamentally epistemological. And in this particular regard, it is the question about the relationship between the two levels: what there is cannot be disassociated from the question about how we know what there is, i.e. the question of normative standards for epistemology cannot be dislodged from the ontological question about the structure of reality lest we revert to a form of dogmatic metaphysics. The question about the relationship between philosophy or metaphysics and science is necessarily at the juncture of questions about what we can know about the world, the relation between concepts and objects.

2) Ladyman and Ross explicitly subscribe to the idea of making metaphysics continuous, and even subordinate, to science, and physics in particular. Theirs is a brand of scientism which undermines the ontological valence of objects qua individuated particulars by arguing that physics teaches us to remain ontologically agnostic about them, the better to avow the reality of structure and primary relations, tracked by mathematical structural patterns. In this regard, L&R are very much informed by work done not just in the philosophy of science, but in science itself; they perform a restricted 'bracketing' about ontological commitments about individuals the better to anchor realism on structure.

3) However, without a proper epistemological footing, naturalist metaphysics cannot be appropriately legitimated on realist grounds, and that in obviating this demand, L&R are forced to appeals on the principled authority of science which devolves in a tacit instrumentalization of science and scientific phenomena, as in Van Fraassen's constructive empiricism, despite their claims to the contrary. The ontological agnosticism about objects is not remedied by an ontological commitment to structure if the authority of the latter merely follows circularly from the 'avowal of science' in principled grounds. In this regard Brassier's remark apropos the Churchlands and Badiou's apropos Quine can be reiterated about L&R, and any variety of scientific realism which is taken to draw its legitimation from a super-empirical standpoint which authoritatively prescribes rather than explains. Pragmatism is complicit with idealism insofar as it adjudicates science from the unexplained principled prescription of the normative practice of science in its current canonical forms.
4) In that regard, L&R's attempts to reshuffle epistemological terms apart from the relation between subjects and objects vis a vis analytic "neo-scholastic" metaphysics is analogous to the criticism against Meillassoux advanced by Gabriel Catren's observation that contemporary physics does not make appeals to inference on the basis of induction from observational terms. This again reiterates the deflation of the terms of the manifest image, and secondary properties in particular, in favor of the mathematicity of structure. More importantly, this reflects L&R's adoption of the semantic approach to theory modeling which dislodges the role reserved for observables vis a vis partial-interpretation in the syntactic approach.  We can also say thus that the scientific revision of epistemological norms by ontic structural realism would itself undermine the construal of the relation between science and the world in terms of subjective inference from observational instances. But, in doing the opposite move, that is, by simply modeling epistemology on a physicalist metaphysics, their philosophy already runs on an unjustified epistemological commitment which endows ontological valence to the postulates of science in pragmatic, principled terms. And this is, like we surmised above, subject to instrumentalist relativization, since the argument cannot be formulated in intra-theoretical terms which would simply insist on the importance of thinking in continuity with the theory in question lest we fall on circularity.

In this regard, the authoritarian validation of science on pragmatic grounds fares no better philosophically than classical metaphysics in its dogmatic pretense to uncritically yield descriptively the nature of reality. Saying we must be continuous with science because science helps us predict the world with accuracy does not get us beyond constructive empiricism, and so it is insufficient for a robust metaphysical naturalism with a realist bent. Realism can not rest on the satisfaction of the no-miracles argument plus a commitment to structure. The autonomy of the real does not follow from predictive success, any more than predictive failure reinforces the dependence of the real on thought. Saying the structure can be metaphysically avowed because there is concensus in bona fide science is both a) motivated by subordinating subordinating metaphysical to the institutional legitimation arbitrarily posited in the PPC and PNC. But through the latter the priority of physics turns out to be prescribed rather than explained. This is made particularly salient in the refusal by L&R to explain the relation between theoretical mathematical posits and 'real' physics structure.

5) Following Sellars, we must claim that the dialectic of concept revision must itself be the condition of possibility for the epistemic endowment of scientific claims. This is what Sellars describes as the methodological, non-ontological, independence of the normative-conceptual register of the manifest image with respect to the natural-causal register of the scientific image. Their relation is not one of plain subordination or undermining, but of a perpetual negotiation. Thus eliminativism does not follow from naturalism; the manifest image retains a methodological independence. The autonomy of the conceptual cannot be eliminated since it is the condition of possibility for the endowment of epistemic entitlement and concept revision itself: it alone tells us on what epistemic basis we should revise our theories and metaphysical commitments.