Naturalistic metaphysics and science
Naturalistic metaphysicians theorize about the natural world—a world also studied by science—by conceiving, assessing, and supporting various theories about it. Many of these theories are admittedly quite speculative. I am interested in the brand of naturalistic metaphysics that recommends, as a methodological dictum, turning to science to see how theories are best assessed and supported. Many naturalists have happily followed Quine in accepting the following three broad methodological judgments. First, in science we find a broad methodological injunction to let theoretical virtues guide theorizing, at least regarding matters that are unobservable to our “naked senses.” Secondly, in science we can witness (a degree of) confirmation holism: a theory can be related to evidence only as a part of a wider web of theoretical beliefs, not in and of itself. And thirdly, science (suitably interpreted) has all the answers to questions of ontology: there is no “first philosophy” concerning questions of existence. These three naturalist tenets lead to two broad strategies to support explanationism in metaphysics.
The first strategy capitalizes on methodological similarity between metaphysics and science in the spirit of the first tenet above. Consider how this tenet is exemplified in the apologia for metaphysics by Sider, Hawthorne, and Zimmerman:
Scientists must regularly choose between many theories that are consistent with the observed data. Their choices are governed by criteria like simplicity, comprehensiveness, and elegance. This is especially true in very theoretical parts of science, for instance theoretical physics... . Just like scientists, metaphysicians begin with observations, albeit quite mundane ones: there are objects, these objects have properties, they last over time, and so on. And just like scientists, metaphysicians go on to construct general theories based on these observations, even though the observations do not logically settle which theory is correct. In doing so, metaphysicians use standards for choosing theories that are like the standards used by scientists (simplicity, comprehensiveness, elegance, and so on). (2008, 7)
Sider, Hawthorne, and Zimmerman stress the idea that in metaphysics, just as in physics, theory choice is governed by theoretical virtues. In effect, they hold that metaphysics is but a degree apart from the more speculative reaches of empirical science, and regardless of its highly speculative character, the methodological similarity with science nevertheless justifies the rationality and meaningfulness of metaphysics as a theoretical endeavor. This is a very broad idea, of course, and there are many ways to make it more precise. In particular, one can refine it in explanationist terms, leading to a vindication of explanationism in metaphysics. Thus, Swoyer (2008, 15), for example, appeals to inference to the best explanation in framing a methodological similarity between his arguments for Platonism, on the one hand, and science on the other.
My suggestion is that we should (re)construe arguments for the existence of abstract entities as inferences to the best over-all available ontological explanation... . Inference to the best explanation plays a central role in daily life and, according to many philosophers, in science. . [M]any maintain, without inferences to the best explanation science, and much of ordinary life, would be impossible.
This is a good example of the strategy that aims to justify explanationism in metaphysics by reference to a methodological uniformity (see also Paul 2012).
As an aside, it is worth nothing that a similar line of thought has been popular in the scientific realism debate in philosophy of science. Here naturalism has been taken to imply that an explanationist argument for scientific realism should exemplify the same method as scientists themselves employ: the realist’s inference to the best explanation—empirical success is best explained by approximate truth—should be viewed as a further application of scientific explanationism. The realist inference thus arguably exemplifies the same method as various explanatory inferences in science—a fact that arguably provides support for scientific realism via a rule- circular justification of the realist’s inference (cf. Psillos 1999, chap. 4; 2011). This, too, is an example of the same strategy for justifying explanationism outside science.
Let us now move on to the second strategy that capitalizes on confirmational holism. This flows out of the three tenets of naturalism as follows. In answering questions of ontology, naturalists turn to our best scientific theories, and confirmation holism recommends belief in all theoretical assumptions that are responsible for the successes—whether predictive, explanatory, or whatever— that provide realism-eliciting evidence for a given theory. Our grasp of theoretical virtues further shapes our understanding of those successes. This gives rise to a Quinean “indispensability argument”: we ought to have commitment to whatever indispensably contributes to our best theories being the best. This then leads to a justification of metaphysical views when it is argued that assuming realism about paradigmatic scientific posits (electrons, quarks, etc.) we should also be committed to paradig- matically philosophical assumptions (e.g., existence of abstracta, the presentist view of time) on the basis of their allegedly similar role in contributing to our best theories.
This broad idea of confirmational holism can be further refined in different ways. In particular, the notion of indispensability at stake is sometimes framed in explanationist terms, leading to a vindication of explanationism with respect to metaphysical claims. For example, the advocates of the so-called explanatory indispensability argument have recently argued that metaphysical posits (e.g., mathematical and other abstracta) can indispensably contribute to some of our best scientific explanations in a way that is ontologically committing (by scientific realist lights).11 Such arguments aim to establish that paradigmatically metaphysical views (e.g., Platonism) can enjoy a degree of empirical confirmation by virtue of the relevant assumptions (e.g., regarding numbers) playing an appropriate explanatory role—a role that is appropriately continuous with those played by paradigmatically scientific assumptions.
The scene is now set for a more critical discussion. I will not be taking issue with naturalism or the broad idea that science and metaphysics can be viewed “of a piece.” Rather, through a closer reflection on explanationism in science, I will only take issue with the explanationist spin that has been put on these two strategies for vindicating naturalistic metaphysics. An incentive for this springs from within the naturalistic conception of philosophy. As said, naturalism recommends turning to science to see how theories are best assessed and supported. But exactly which scientific disciplines should we turn to in forming a view of the successful scientific methodology and its limits ? I think we should follow Quine himself and understand ‘science’ quite broadly, so that all scientific findings concerning science and scientists should be taken on board, including those from the history of science and the relevant areas of psychology.  After all, there is every reason to expect that these areas of enquiry have potential to inform us of our capacity and reliability in explanatory reasoning. With this in mind, let us now have a closer look at the two strategies for vindicating explanationism, in the light of the study of explanation and explanatory inferences in psychology and history and philosophy of science.
-  For Quine, the same theoretical virtues are also important for justifying our views concerning common senseobjects.
-  Note that these philosophers are patently not claiming that metaphysical theories are equally well supportedby evidence as our scientific theories are. Rather, the claim is that the kind of justification is arguably similar totheoretical domains of science. Metaphysics is undeniably much more speculative than science, and one shouldnot object to explanationism in metaphysics—as, e.g., Ladyman (2007, 2012) partly does—on the grounds thatit does not have a probativeforce on a par with the explanatory considerations arguably operative in science. Forthis is just to object to metaphysics’ speculative nature, not to explanationism in metaphysics per se.
-  As Putman puts it, scientific realism should be “viewed ... as part of the only scientific explanation of the success of science” (1975, 73). Similarly, Psillos maintains that naturalistic philosophers of science “should employno methods other than those used by the scientists themselves” (1999, 78), and Boyd argues that “the epistemology of empirical science is an empirical science” (1989, 13).
-  Note that here, too, a naturalistic philosopher is not committed to claiming that her philosophical theory (aboutscience) can be supported to the same degree that scientific theories themselves are supported. One should notobject to explanationism in this context merely on the grounds that it does not have probative force on a parwith the explanatory inferences in science. This mistake is made by Frost-Arnold (2010), who argues that therealist explanation (and the related inference to the best explanation) fails to satisfy scientific demands for a goodexplanation (and good explanatory inference), and therefore fails the tenets of naturalism. This demand is basedon too strict a conception of naturalism. The realist explanation can be purely philosophical in the sense of notenjoying the degree of evidence enjoyed by paradigmatically good scientific explanations. The mode of inferencecan be the same in the two cases, while the overall evidence (or “epistemological standard”) is not.
-  See, for example, Colyvan (2006, 229): “[IBE] is a special case of the indispensability argument. [T]his is astyle of argument that the scientific realist accepts. Mathematical entities surely feature prominently in variousexplanations.” See also Baker (2009) and Psillos (2011) for arguments in the same spirit.
-  See, e.g., Quine (1995, 49): “I use science broadly, including not only the ‘hard sciences’ but also ‘softer sciences,’from psychology and economics through sociology to history.”