Causality and Explanation.

Bayer, Benjamin

SALMON, Wesley C. Causality and Explanation. New York: Oxford University Press, 1998. xiv + 434 pp. Cloth, $65.00; paper, $29.95--In Causality and Explanation, Wesley Salmon has assembled two decades of his essays on scientific explanation and causality, many of which were previously unpublished or hard to find. Offering introductory essays for beginners in the philosophy of science, as well as advanced material on technical and applied topics, this collection traces the gradual development and modification of Salmon's views. Throughout this development the central spirit of Salmon's project shines through: to "put the `cause' back in `because'" (p. 103).

Although Salmon opens the collection with a tribute to Carl Hempel, he wishes to distance himself from Hempel's deductive-nomological, and inductive-statistical conceptions of scientific explanation ("A Third Dogma of Empiricism," 1977). According to each, a scientific explanation is a form of argument front a set of premises (universal or statistical generalizations and antecedent conditions) to a conclusion. Salmon takes the explanation-as-argument view as inadequate, and seeks to offer his "statistical relevance" model as an alternative ("Causality and Theoretical Explanation," 1975). On this model, an explanation is not an argument, but an assemblage of the total set of factors relevant to an effect. Salmon construes statistical relevance through Reichenbach's principle of the common cause: when two events occur together more often than each would independently of the other, a common cause must explain both (as when all of the lights in a room go off because of one switch).

Salmon does not believe that statistical relevance alone is sufficient to put the "cause" back in "because." To that end, he wishes to show that it is precisely because there are continuous physical processes linking diverse effects with their common cause that we observe their statistical relevance. In other words, the improbably coincidental cries out for an explanation in terms of real causal connections. By invoking numerous scientific examples (such as the seemingly miraculous convergence of measurements of Avogadro's number in the analysis of Brownian motion and electrolysis), Salmon makes a strong case for the existence of theoretical entities--the agents of spatiotemporal continuity--and for the validity of scientific realism.

Salmon, it turns out, has much to say about causal processes. He argues that to play a role in scientific explanation, they must be distinguished from "pseudo-processes" (p. 16). In an oft-repeated but clarifying example, Salmon asks us to picture a rotating beacon in a sports stadium. According to special relativity, no object may exceed the speed of light, though the rotating spot of light on the wall can go arbitrarily fast, depending upon the radius of the stadium. The moving spot is a pseudo-process, and nothing like the propagation of a real physical signal through space. In earlier essays, Salmon argues that the hallmark of a causal process is its ability to transmit a mark. For instance, one might place a red filter along a spot of the circumference of the beacon, producing a red spot on the opposing wall. The filter transmits a mark to the spot on the wall, but nothing can be done at that spot to make adjacent spots red.

In later essays (for example, "Causality without Counterfactuals," 1994), Salmon reveals that he has come to doubt that causal processes can be explicated entirely in terms of mark transmission. The notion of the ability to transmit a mark turns out to involve reference to counterfactuals, which Salmon had originally sought to avoid (p. 252). His commitment to the role of causal processes in explanation remains steadfast, however, as he draws on the work of Phil Dowe to argue that causal processes can instead be described in terms of conserved quantities, such as momentum or energy.

By emphasizing causal processes as important elements of scientific explanation, Salmon has distanced himself from the Humean emphasis on events ("Causal Propensities," 1990). In doing so he feels that he has formulated a solution to Hume's problem of causality: the much sought-after connection between cause and effect is nothing but a real physical process, continuous in space and time--even if it is not a necessary connection, as modern physics suggests. Whether or not the process is the right element to focus on, Salmon's departure from a Humean universe of disjointed events is a welcome breath of fresh air.

It is modern physics, in fact, that is something of a stumbling block for Salmon. Though he seems enthusiastic to accept indeterminism--for which he designs his theory of probabilistic causality (pp. 3, 5)--he is less enthusiastic about the "spooky action at a distance" of EPR phenomena, which directly contradicts his emphasis on continuous physical processes (p. 23). Clearly something has to give: important aspects of his theory, or modern interpretations of physics. Salmon seems to realize this, however, which we cannot help but appreciate in the context of an otherwise impressive work.

--Benjamin Bayer, The University of Illinois at Urbana-Champaign.