|
This review was published in Canadian Philosophical Reviews, (April 1993) vol. 13, no. 2, pp. 86-89.
Make no mistake: in spite of its subtitle advertising it as "An Introduction to the Philosophy of Logic", this book is not for the uninitiated. Even those with a background in both logic and the philosophy of language will find The Norm of Truth slow going. (You might like to look, for example, for the sentence of 108 words on p. 91). In what sense, then, is the book "an introduction"? Only in the sense that sometimes Engel alludes to problems without pursuing them to completion or conclusion (see e.g. p. 270 or p. 281). The book takes its title from a sentence on p. 313: "… logic is a theory of the norm of truth." In the "Introduction", Engel explains that this book is explicitly not intended to be comprehensive and certainly is not developed around a central theme: "I have tried to deal with the various [selected] problems in a relatively autonomous way, in order to allow a separate reading of each chapter" (p. 11, gloss added). In short, the book may be regarded as a collection of papers on various topics in the philosophy of logic. (Earlier [p. 7] Engel explains his disapprobation of the term "philosophical logic".) Despite the claim that the essays are relatively self-contained, there is a great deal of 'previewing' in the earlier ones, and substantial cross-referencing throughout. Engel's "Conclusion" (pp. 321-323), is far more useful if read as a Preface. In his summing-up, Engel most clearly expresses the philosophical thrust of the book: "logic … does not commit us to any particular ontology, because it does not describe a world, but only prescribes the most general conditions for such a description. … In this sense it is formal. I have held that the 'formal contents' of logic rest on an ontology that is only minimal. By 'minimal' I do not mean that the ontological commitments of logic are those prescribed by extreme nominalists …, but rather that logic is neutral with respect to a number of ontological issues. Thus a theory of truth, as it is formulated through a Tarskian conception, is 'modest' and 'non-substantial'. In modal logic, the 'essentialism' that is prescribed by logical semantics is minimal too, and 'possible worlds' are not genuine parts of the world. The logical theory of identity only states the constraints of the individuation of substances." (p. 321) Chapter 1 examines propositions, their (presumed) roles – as truth-value bearers, as the linguistic meanings of sentences, as the content of what is said, and as the content of certain psychological states – and their possible ontological character – as (linguistic) symbols, as abstract entities (including the possibility of being collections of possible worlds), and as collections of objects or properties in the world (facts, states of affairs, etc.). Arguing that it is impossible to individuate propositions to the degree needed to preserve many of our pre-theoretical distinctions, Engel opts for a Quinean analysis of sentences as truth-bearers. But he does allow propositions a 'heuristic' value (p. 34). Already in Chapter 1, we find an early instance of what turns out to be one of the most pervasive puzzles in Engel's style. On page 29, he explains and illustrates the type/token distinction; but on the facing page, he offers neither explanation nor illustration in his invoking of the concept of a "partial sub-model". One can only wonder what sort of audience Engel imagines he is writing for, one for whom he frequently pauses to explain relatively familiar and simple concepts while plowing on through relatively unfamiliar ones. In the midst of
Chapter 2, "The meaning of propositional connectives",
Engel examines Johnson-Laird's theory of 'natural' reasoning. Here the
claim is (p. 51) that – using Carnap's and Bar-Hillel's method of
calculating 'information content' – the inference "A; therefore A or
B" is improbable in ordinary reasoning because the conclusion has less
informational content (0.25) than the premise (0.50). This example is
then contrasted with the more 'natural' "If p or q,
then r; p;
therefore r Chapters 3 and 4 deal with quantification theory. In 3, "Subject and Predicate", Engel defends a Fregean view against Sommers' 'naturalistic' account of subjects and predicates. Chapters 5 and 6 (Part 2) concern truth and meaning, and examine the theories, especially, of Tarski, Davidson, and Dummett. Chapters 7 through 10 (Part 3 "Limits of Extensionality") concern (respectively) modalities, possibles and essences; reference and propositional attitudes; identity; and vagueness. Finally, chapters 11 through 13 (Part 4 "The Domain of Logic") are given over to: the province of logic; logical necessity; and logic and rationality. Throughout one finds all the expected writers visited: (in addition to those already mentioned) Kripke, Leibniz, D. Lewis, Mill, Prior, Putnam, Quine, Russell, Wittgenstein, etc. In each case, Engel has something interesting and probing to say. On the production side, there are a very great number of typos in this book; fortunately most (although not quite all) are fairly obvious and do not prevent understanding the author's intent. The (Subject) Index is novel and dual purpose: it is in fact a Glossary with references to the text. The Name Index is inexplicably selective. Although Engel cites and discusses, for example, the work of both Rips (p. 298) and Wason (pp. 300-302, 311, 315), neither researcher is listed in the Names Index. The Bibliography, in contrast, is excellent and provides a valuable list of many important recent and historical writings in the field. Overall, the book – notwithstanding certain flaws – is first rate. Anyone interested in the philosophy of logic should want to read this volume. It goes without saying that every university library should own a copy.
Return/transfer to Norman Swartz's Philosophical
Notes
|
". The numerical values are not given; but if one
calculates them, one finds that this latter inference is no better than
the former, for the information content of the premise set is 0.75,
while that of the conclusion is only 0.50. Indeed, this is just what we
ought to expect, given that it is virtually universally conceded that
deductively valid inferences are non-ampliative. Although Engel finds
quite different fault with Johnson-Laird's suggestion, he seems not to
have seen this one.