Walton, Douglas. Abductive Reasoning.

WALTON, Douglas. Abductive Reasoning. Tuscaloosa: The University of Alabama Press, 2004. xv + 303. Cloth, \$40.00--Since Protagoras at least, rationality has been portrayed as ineliminably subjective and, therefore, one argument as good as any other. As Douglas Walton notes, however, it is not all that long a leap to the further conclusion "that there is no such thing as rationality at all." The "sharp challenge" posed here to logic is to identify new resources for defending rationality (p. 121). Walton dedicates his new book to providing an exhaustive account of the foundations of a kind of rationality that is not confined to the things we can know and prove beyond doubt, but rather is "broad enough to include rational opinions that can be supported by good reasons that are, nonetheless, inconclusive" (p. 102).

Abductive inference is a relative newcomer in discussions of logic, Walton says. Indeed, most logic textbooks ignore it. For whereas logic is expected to be exact, abduction is decidedly inexact and, as such, suspect. It is an "uncertain and tentative" brand of reasoning, "fallible and conjectural," "variable and presumptive" (pp. 1, 4, 33). If deductive inference abstracts directly from data, while inductive inference is based on but extrapolates partially beyond data, abductive inference extrapolates still further (p. 13). Just as we speak of deductive necessity and inductive probability, so we speak of abductive expectability (p. 14). We associate with abduction what Walton calls "the judgment of likelihood" (p. 20).

C. S. Peirce conveys what is at once most precarious and compelling about abduction when he likens it to "the instincts of animals" (p. 9). Abduction draws its conclusions from an incomplete body of evidence, and to that extent, Walton confirms, counts as a guess. But it is an "intelligent guess," he insists (pp. 3, 215). The word "abduction," he proceeds to remind us, derives from ab and duco, meaning, leading back (p. 34). Accordingly, abduction amounts to reasoning from observed data to some hypothesis that would explain them. The aim of abduction, then, is, first, to generate plausible hypotheses and, second, to evaluate those hypotheses by inference to the best explanation, where the best explanation is understood to be that which "covers more of the given facts and contradicts fewer of them" (pp. 22, 174). On several occasions, Walton contrasts abduction with prediction (pp. 29, 33, 129, 193), since abduction "begins with observation and then moves to explanation of what was observed" (p. 18). This link with explanation is crucial, since we aim to explain, rather than engage in argument, in cases where there is no disagreement concerning the truth of designated propositions, but rather an urge to account most convincingly for why they are true (pp. 79-80).

What prompts an explanation is a request, not so much for information, as for understanding (pp. 80, 216). The prospects for understanding are made possible by our shared capacity for practical reasoning and heightened by our active engagement in dialogue. According to Walton, in fact, abduction is best understood as a "dialogue sequence," involving a series of questions and answers, but also the eliciting, and ultimate evaluation, of competing explanations (p. 32). He proceeds to contrast this "dialogue theory of explanation" with Hempel's deductive nomological, or covering law, model (pp. 52, 56, 78). Construing explanation as a sequence of reasoning from a system's knowledge base together with a set of laws is, as recent research in artificial intelligence has shown, inadequate. Instead, we ought to regard it as a "transaction" or "conversational exchange," an "interactive process between two agents or parties in a dialogue" (pp. 60, 69).

Walton's concern for the dialectic, or pragmatic, aspects of explanation is reflected in his repeated efforts to establish that the caliber of our reasoning is a matter as much of use or purpose as of form (pp. 100, 111, 120, 155-6, 165, 202). For instance, in the dialogue model, "an explanation is seen as a kind of inference, or at least as based on an inference or chain of reasoning" (p. 78). In such an inference, however, "the conditional ... goes from an antecedent finding of fact (observation, indicator, sign) to a consequent that postulates a best explanation of the observed fact" (p. 147). And yet, the implied modus ponens form, here, is defeasible rather than deductive, since the conditional premise functions not as a universal quanitifier, but as "the hedged quantifier of ordinary life" (p. 151).

The distinction has long been drawn in logic between efforts to assess arguments in light of standards of formal, or structural, correctness, on the one hand, and the use of logical reasoning to invent arguments, on the other. The dynamic, even creative, aspects of abduction conform to this latter ars inveniedi, or art of finding (p. 225). Abductive inference, after all, is always open to "new evidence and future developments" (p. 234). At each step in an account, new statements are added, and, through a process of "colligation," new accounts successively produced, impelling the dialogue forward (pp. 213, 228). Walton's Abductive Reasoning is an informed and tenacious effort to offer an account that reconciles this "feature of openness" with an "overarching structure representing discovery" (p. 231).--John E. MacKinnon, Saint Mary's University.
COPYRIGHT 2007 Philosophy Education Society, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.