The Positive Metaphysics of ETMG
As discussed above, analytic metaphysicians seem largely unmoved by the critique of ETMG, and some are happy to argue that metaphysics is a priori first philosophy that underpins science, and that without it there cannot be scientific knowledge at all. On the other hand, some supporters of our negative arguments and views, such as Stanford (2010), object that we do not go far enough by failing to abandon metaphysics altogether. Despite ETMG being known to many only as a diatribe about analytic metaphysics, and sometimes being misinterpreted as a repudiation of metaphysics in general, the subtitle of ETMG makes it clear that the book advocates and develops a positive metaphysics. It also offers a general conception of naturalized metaphysics that is independent of the specific version the book develops. The critique of analytic metaphysics takes up only a part of one chapter, and the main body of the work is devoted to the construction of a positive metaphysical theory that draws on and attempts to unify our scientific knowledge of the world. The latter is the combination of Ontic Structural Realism due to Ladyman (1998) and Rainforest Realism due to Ross (2000), the avowed aim of which is to offer a theory of ontology to unify physics and the special sciences and to offer an account of causation that is compatible with fundamental physics.
It first defends the idea that insofar as there is metaphysics, it should be naturalized.
(I) Metaphysics should be naturalized and draw upon science.
In order to give this content, more must be said about what is to count as naturalized metaphysics and to draw upon science. Some philosophers may object that (I) misleadingly implies that metaphysics is distinct from science when in fact they form a continuum and all metaphysical questions are reducible without residue to theoretical questions in particular sciences. On such a view, questions about, say, natural kinds are to be resolved into questions about kinds in biology, chemistry, the social sciences, and so on. One may then wonder what is left that is distinctively metaphysical. Hence, one may endorse (I) while believing that naturalized metaphysics is nothing more than addressing traditional metaphysical questions as they arise in science. Similarly, there are debates about identity and individuality of quantum objects and spacetime points, about biological individuality, about locality, about the nature of the self or free will, that are conducted in the context of specific scientific disciplines. This is clearly a kind of naturalized metaphysics that involves addressing traditional metaphysical questions in a way that is continuous with theoretical debates in science (though one may want to say that this is just science not metaphysics that is just a terminological issue).
The book goes on to argue that
(II) Science is integrated to a very great extent, pace Nancy Cartwright, John
Dupre, and others.
Generality is the feature that distinguishes the distinctively metaphysical task from the activity that is continuous with theoretical science. Metaphysics so construed is about the big picture, and the naturalistic version builds this big picture in the light of the actual sciences that we have. What makes the account of naturalized metaphysics in ETMG distinctive is the claim that it is about the unification of science and hence it presupposes (II). This motivates a particular account of naturalized metaphysics:
(III) There is a distinctive conception of naturalized metaphysics that involves
the unification of science.
The positivists sought to demarcate science from metaphysics by providing criteria of empirical meaning. While some have not given up hope of doing so (Lutz 2012), demarcating science from metaphysics in such a way is fraught with problems. Contemporary philosophers of science tend to have learned from Quine that empirical content is spread throughout our whole conceptual scheme. High-level laws like the conservation of energy have no direct empirical implications without being accompanied by background theories and auxiliary hypotheses, and metaphysical doctrines, for example concerning locality and individuality, have played a role in science. Nonetheless, there is a big difference between most of the hypotheses discussed in current metaphysics and such high-level theoretical principles, since the former are usually completely decoupled from all empirical inquiry and not involved in corresponding scientific debates. Since it is not possible to say a priori what kind of metaphysics is required to unify science, naturalized metaphysics cannot be demarcated by its content or by a prohibition against certain kinds of hypotheses, but should instead be demarcated by whether or not it makes contact with the science of its day and serves the goal of unification expressed in (III) above. Ladyman and Ross express this as follows:
(IV) Naturalized metaphysics is demarcated by the Principle of Naturalistic Closure (PNC):
If a metaphysical claim is to be taken seriously at a time t, it should, if true, show how two or more scientific hypotheses, at least one of them specific and at least one of them both specific and drawn from fundamental physics taken seriously at t, jointly explain more than the sum of what is explained by the two hypotheses taken separately. (Ladyman and Ross 2007, 37)
First, why both the requirements in “at least one of them specific and at least one of them both specific and drawn from fundamental physics taken seriously at t”? If one says “at least one of the Xs is A and at least one of the Xs is both A and B” that is equivalent to saying “at least one of the Xs is both A and B”—the first conjunct is logically redundant. Secondly, why the reference to fundamental physics? There might be things wrong with the metaphysical claim that species are individuals not kinds, but surely it cannot be required to be unifying with respect to fundamental physics ? Thirdly, I would welcome a bit of explanation and justification. The text did not really explain what the PNC is for or why we should believe it.
This is intended as a provisional statement and to be understood in conjunction with the account of a very weak notion of the primacy ofphysics explained in the rest of the chapter. Clearly, it only has content insofar as it draws upon an understanding of “scientific hypotheses.” However, the demarcation of science is notoriously problematic, and the history of the debates initiated by the positivists—Popper, Lakatos, Kuhn, and so on—about the nature of the scientific method have not resulted in a consensus about how to characterize science.
Roughly speaking, there are three putative candidates for the nature of science, namely, the content of scientific propositions, the logic of scientific thought, and the social structure of interactions among scientists. It is clear that all these elements are important and that they are not independent of each other at least in practice; however, much philosophy of science has traditionally been focused on the first two. On the other hand, many approaches emphasizing the sociology of science are antirealist and deny the objectivity of scientific theory choice and progress. Nonetheless, there are forms of social epistemology according to which scientific knowledge is the result of social rationality and is primarily at least possessed by the community, not by individual scientists (for example, Goldman 1999). The collective nature of science is arguably what makes science self-correcting and truth-tracking to the extent that it is, and hence uniquely valued in establishing matters of fact in all other social institutions such as the law, policy making, engineering, and medicine. Ladyman and Ross take it that the only way to demarcate science and metaphysics for the purposes of their polemic is by ostension to the social institutions of science rather than by definition. This is why the PNC is to be understood with reference to “bona fide scientific institutional processes” rather than with respect to a set of necessary or sufficient conditions (Ladyman and Ross 2007, 37).
Metaphysics can similarly be characterized institutionally, and Ladyman and Ross make it clear that their target is analytic metaphysics, and they choose a few authors more or less at random who clearly exemplify that tradition. Their point is not so much that particular individuals are working in violation of the PNC but rather that the center of gravity of the discipline has shifted much too far in an a priori, intuition-led direction that has taken it out of contact with science. Of course, there are many philosophers who practice metaphysics in a way that is informed by science. The problem is that the way that core subjects like the nature of matter, space and time, causation and law, composition and individuality are framed and investigated in analytic metaphysics shows nothing of the scientific image of the world. The PNC is an attempt to say what is required for metaphysics to be informed by science, and for it to contribute to the scientific image.
Many advocates of the unity of science are reductionists, and it might be thought that naturalist metaphysics means materialist or physicalist metaphysics. The question as to how the special sciences relate to physics is of fundamental importance to the positive project of ETMG, and again the approach predominately taken to these matters in analytic metaphysics does not make sufficient contact with science. ETMG argues that 
152 Metaphysics and the Philosophy of Science
Accordingly, physics is not defined in terms of fundamental building blocks of reality but as follows:
(VI) Physics is to be characterized as the science that holds at the widest range of scales and across all of space and time.
ETMG argues that
(VII) Reductionism involving either type-type or token-token identity is not in general plausible, though it may be in specific cases.
However, in the absence of reductionism, selective eliminativism is contrary to naturalism, so (assuming some kind of scientific realism which is argued for in chapter 2 of ETMG) this motivates Rainforest Realism (following Ross 2000):
(VIII) Science offers us ontologies at many different levels. The ontological commitments of the special sciences should be taken as metaphysically on a par with those of physics. For example, there are atoms, cells, organisms, agents, social structures, and indeed tables.
These last two claims are discussed in the next section.
ETMG (especially chapters two and three) elaborates a consilience argument for the following claim:
(XI) Structural realism in its ontic not its epistemic form is the most viable position in the scientific realism debate. In particular, neither standard scientific realism nor constructive empiricism is tenable.
This consilience argument makes much of the problems with orthodox metaphysical notions of individuals in physics and other sciences, including biology. (There is of course much more to be said about the characterization of ontic structural realism and the associated debate about individuals in science; see Ladyman 2015, 2016).
It is more or less taken for granted in ETMG that
(XII) Scientific representation is often ineliminably mathematical and structural
in broad conformity with the semantic approach to scientific theories.
It is argued that rainforest realism and ontic structural realism are unified by the idea that
(XIII) Existence should be explicated in terms of the theory of real patterns.
The definition of real patterns (Ladyman and Ross, 2007, chapter 3) is as follows:
To be is to be a real pattern; and a pattern is real iff it is projectible under at least one physically possible perspective; and it encodes information about at least one structure of events or entities S, where that encoding is more efficient, in information- theoretic terms, than the bit-map encoding of S, and where for at least one of the physically possible perspectives under which the pattern is projectible, there exists an aspect of S that cannot be tracked unless the encoding is recovered from the perspective in question.
The account of real patterns in Dennett, Ross, and ETMG is articulated in information-theoretic terms as above. However, Ladyman and Ross (2013) argue that another way to understand it is in terms of statistical structures. Real patterns theory can also be developed in terms of the dynamics of phase spaces (as Jenann Ismael suggested in conversation). Hence:
(XIV) The theory of Real Patterns can be explicated in various ways, including
in information-theoretic, statistical, and dynamical terms.
Many real patterns in science are such that the compression of the information allows only probabilistic recovery of the underlying system. Real patterns are lossy, but if one ignores them, one misses out on a real feature of the world. (Consider the Carnot cycle as a real pattern with respect to the underlying statistical mechanics.) Hence, there is not even token-token identity between levels (VII).
Should a naturalistic metaphysician be a realist about natural necessity? Ladyman and Ross argue that objective modal structure is represented by science on the basis of consilience in respect of the following: induction, the nomiracles argument, novel prediction, probability, information processing and computation. Indeed, real patterns are defined modally. They are there to be discovered. Hence:
(XV) The world has an objective modal structure that is represented by causal
claims and laws in the special sciences.18
Chapter 5 of ETMG argues that causation requires a spatiotemporal frame of reference and to that extent at least it is perspectival in something like the sense ofJennan Ismael (this volume, chap. 5). Ismael also explicitly commits herself to there being an objective modal structure to reality. The idea that objective modal structure is a key component of ontic structural realism is emphasized by Ladyman (2000), and reiterated by French and Ladyman (2003). Berenstain and Ladyman (2012) argue that the standard arguments for scientific realism appeal to objective modal structure.
David Wallace (2001) advocates what he calls a “functionalist account of ontology” based on the notion of real patterns in his elucidation of the Everettian interpretation of quantum mechanics. His ontology is two-tier, in that only higher-order entities such as cats and tables are understood in terms of real patterns, whereas the wave function (or whatever else proves to be fundamental in physics) is understood in categorical rather than functional terms. On the other hand, since they hold (V), Ladyman and Ross (2007) advocate a real-patterns account of ontology across the board. There is an asymmetric relation among them due to the fact that physics is maximally general (VI). The relation is not the metaphysical composition of composition as explained in the next section. Ladyman and Ross (2007) deny strong forms of physicalism, since they hold that real patterns at different scales are emergent structure not reducible to the sum of the parts, and that the idea of physical building blocks of higher-level entities is not applicable beyond the atomic scale.