The first step in approaching this book is to determine what Margolis means by “scientism.” In the preface and introductory chapter we learn that this is the second part of a two-volume study of American philosophy in the last half of the twentieth century; the earlier volume is Reinventing Pragmatism (Cornell University Press 2002). The present volume is a critique of a major strand of analytic philosophy, which Margolis identifies with scientism and characterizes in terms of two central themes: materialism and extentionalism. Margolis takes Davidson and Quine as his main representatives of scientism, although many other philosophers are discussed (Putnam and Rorty are his representatives of latter-day pragmatism). Margolis makes it clear that he uses “scientism” “in an unqualifiedly derogatory sense” (5), and holds that “the record of the last half-century is, philosophically, largely a record of the dawning exhaustion of an impressive vision (scientism) and the incompletely developed, still somewhat inchoate, possibilities of a promising alternative philosophy (pragmatism)” (xii). But although Margolis maintains that scientism is spent, “it hardly follows that analytic philosophy need be spent as well” (17). Margolis sometimes puts “scientism” in scare quotes and often writes of “scientisms”—practices that he also follows for other key terms. He also uses various familiar terms with special meanings. So we must do some more work to try to pin down the target of his critique.

Margolis identifies scientism with what he calls “Cartesianism”: “Broadly speaking, scientism is Cartesianism by way of the materialisms and extentionalisms by which it seeks to gain it goal” (11). “Cartesian” is a “term of art” that applies to “any realism, no matter how defended or qualified, that holds that the world has a determinate structure apart from all constraints of human inquiry, and that our cognizing faculties are nevertheless able to discern those independent structures reliably” (49). Cartesianism is opposed to constructivism—one label for the view Margolis favors. Constructivism holds that epistemology and metaphysics are necessarily mingled so that whatever we take to be “ontically independent of human inquiry” is epistemically dependent (51). Margolis emphasizes that constructivism is not a form of idealism since it is not committed to the view that all reality is mind-dependent. The claim is rather that “whatever, on epistemic grounds, we rightly judged to be ontically independent would count as such only on the sufferance of its being epistemically dependent (that is, confirmed)” (54). In other words, constructivism (I think) includes any view which holds that objective claims about the world must be made on the basis of evidence, and that our means of gathering and assessing evidence change over time. Presumably, Cartesianism holds that we have direct, unmediated access to the world. Margolis also describes his favored view as Hegelian, where this is another term of art encompassing those who “wish to integrate the analysis of truth—I would say, ‘holistically’—with the analysis of knowledge and reality, on the promise of overcoming the Cartesian paradox” (79). Apparently, this paradox consists in holding that we can come to know the world as it is while denying unmediated access.

Two more terms require attention. What Margolis intends by “materialism” is clear enough. He holds that “The pivotal issue remains (as it has always been) the public standing of the mental” (17), and materialism includes all versions of reductionism and eliminativism. The sense of extensionalism is not so straightforward. Fairly late in the book Margolis writes, in a footnote:

With this background I will sketch each of Margolis’ four main chapters.

Chapter 1 is aimed at materialism—which Margolis also identifies with a computational account of “biological, psychological, and cultural phenomena” (20). The main opponent is Chomsky (although Margolis makes positive remarks about some aspects of Chomsky's views) with side attacks on Dawkins, Dennett, Pinker, and others. Early in the chapter (20) Margolis concedes that he cannot refute the views he is opposing. Instead, he offers arguments to show that these views are not proven; then states his own views without any argument at all. This pattern recurs frequently throughout the book.

Chapter 2 deals with incommensurability—which Margolis prefers to call “incommensurabilism”. The main targets are Davidson and Putnam, who are written off as having missed the point: “The elementary point is this: semantic incommensurability is linguistically ubiquitous, utterly beyond dispute; it is also conceptually, theoretically, philosophically quite benign. Hence any attack on the bare phenomenon cannot fail to be a blunder” (53). Margolis takes incommensurability to occur whenever we lack a neutral basis for comparing predicates, or classes of predicates. He then distinguishes four common forms of incommensurability which are illustrated by the side and hypotenuse of a right triangle, lack of a standard for comparing (for example) colors of textiles with their sizes, differences of prepositions among natural languages, and results of drift over time of the meanings of terms in a natural language. Unfortunately, none of these cases involve the competing conceptual systems that are at the center of discussion in philosophy of science. Margolis does hold that, given an historicist/constructivist view of human knowledge that rejects any form of privilege, second-order “incommensurabilist epistemologies will arise from time to time within the space of garden variety incommensurablities” (53). But, he assures us, these cases are built on the ordinary forms of incommensurability (55), are all variations of the first two cases (59), and involve no particular cognitive difficulties. Alas, he never shows how we move from familiar cases to those that involve competing schemes. Margolis also claims that incommensurability has nothing to do with translation and that “Davidson unaccountably conflates Kuhn's well-known rejection of neutrality with the ‘failure of intertranslatability’ …” (56). Readers of Kuhn will recognize the he not only identified incommensurability with impossibility of translation, but continued to do so in his final papers even after he had acknowledged that multi-linguals are common, and that people working from incommensurable frameworks can learn to communicate by learning each other's languages.

Chapter 3 attacks deflationist accounts of truth, and more generally any view that treats truth in purely semantic terms. Instead, Margolis maintains, a correct analysis must be “‘holistic’: it cannot, except derivatively, segment its so-called semantic, epistemological, and metaphysical roles” (83). The main critical target of the chapter is Davidson, although there is a substantial discussion of McDowell who “salvages the epistemological import of ‘true’ … [but] only in a ‘formal’ way” (85). Towards the end of the chapter Margolis introduces a distinction between two kinds of holism—internal and external—that will play a central role in the final chapter. While both occur in many forms, the characteristic feature of external holisms is that they rely “on some form of epistemic privilege obscured in holist terms …; internal holisms are epistemologically benign” (102, cf. 108). Davidson's claim that most of our beliefs are true is given as an example of external holism. Internal holisms are found in Hegel, Nietzsche, Heidegger, Dewey, the Frankfurt Critical school, Wittgenstein, and Foucault.

Chapter 4 is largely devoted to Quine, whose doctrines of indeterminacy of translation and stimulus meaning are rejected as instances of external holism; his argument against the analytic/synthetic distinction is praised as a case of internal holism. (Wittgenstein's later doctrines of language games and forms of life are mentioned as further examples of benign internal holism.) Margolis is particularly scornful of what he describes as Quine's doctrine of “holophrastic sentences” (particularly in The Pursuit of Truth), even though Margolis admits in the introductory chapter that he does not understand this notion (3, n. 3). Part of Margolis’ problem is that Quine does not talk of holophrastic sentences, but of two different ways of understanding sentences: holophrastically (as a unit) and analytically (compositionally). The distinction is taken from C. I. Lewis (An Analysis of Knowledge and Valuation) who uses it to explain why, for example, two tautologies that are logically equivalent can still have different meanings. (Suppose you asked a student to write down DeMorgan's laws. Would you accept $p\supset p$ ?) When Margolis introduces the notion, he correctly reports that “holophrastic” refers to one way of viewing a sentence, but immediately shifts to talk of a sentence type (111). At the end of the chapter Margolis returns to eliminative materialism as exemplified by Paul Churchland, Dennett, and Sellars.

Although Margolis often states his own views without argument, there is a master argument that pervades the book. Having concluded that we lack unmediated knowledge of reality, Margolis infers that our accounts of reality are a function of our means of epistemic access—in other words, that there is no knowledge of things as they are apart from our thinking about them. But this conclusion does not follow from Margolis’ constructivist thesis. Suppose we agree that the concepts we use to think about some domain are human inventions, that the content of these concepts is at least in part a function the beliefs in which they occur and of the means we use to confirm those beliefs, and that our grounds for accepting claims about a domain depend on the theories we accept, including the theories we use to interpret our instruments. All of this does not imply that the concepts we use cannot correctly describe items that exist independently of those concepts. Nor does it imply that the process of theory testing cannot yield good, although fallible, reasons for accepting (or rejecting) a theory as a correct account of a reality that exists independently of that theory. To be sure, we usually cannot test theoretical claims one-by-one, but we can test limited constructs such as the existence of radioactive isotopes, or the double-helical structure of DNA. Such theories make testable predictions. Confirmed predictions provide reasons for accepting an entire construct as an account of the items it attempts to describe; failed predictions provide reasons for adjusting or rejecting the construct. The process is not as simple or certain as many would like, and there is no guarantee that we will ever arrive at a final theory in any specific case, but that is not the same as holding that the goal itself is somehow misconceived.

What is the upshot of this book? Not much. It does not provide any insight into science or philosophy of science, nor any reasons for doubting that science provides the best means yet developed for learning about nature. Are there questions that science cannot answer? Perhaps. Are there other means of answering such questions? Perhaps. This book does not provide any steps toward resolving these questions one way or the other, or for doubting that the continued pursuit of science is our best means for seeking to resolve them. Those who read the book's title in its most familiar sense will be disappointed. What about Margolis’ actual target—some aspects of analytic philosophy in the last half of the twentieth century? Challenges to analytic philosophy as a philosophical methodology, and replies to these challenges, exist in the literature, and this book does not add anything to the debate.