EVIDENTIAL PARTICULARISM: MILL’S DEDUCTIVE METHOD
to understand what he calls “complex phenomena.” These are ones in which not just one cause and a law governing it are involved but various causes are operating, governed by different laws, where these causes combine to produce effects of some type or types.
How should one proceed to establish the causes and laws in such cases? Mill's “deductive method” provides the answer. It consists of three parts, all of which are necessary.The first part contains what Mill calls “inductive” inferences to general types of causes and laws governing them, using his famous “canons” of causal reasoning to do so. For Mill, each type of cause and causal law introduced needs to be defended empirically by causal-inductive reasoning involving the presence or absence of observed effects, or derived from other causal laws that are so defended. The second part of the method, which Mill calls “ratiocination,” involves combining the various causes and causal laws governing the causes introduced and calculating the consequences that follow deductively from the combination. Among these deductive consequences, it is hoped, will be ones that can be verified by experiment and observation. If they are so verified—the third part of Mill's deductive method (“verification”)—then the entire system of causes and laws employed is inductively and deductively verified, and can be used to explain the complex phenomena that prompted the investigation.
Mill's deductive method is, in essence, a combination of what Newton calls “analysis” and “synthesis.” Analysis consists in making causal and inductive inferences from observed “phenomena” to the causes of these phenomena and the laws governing them (see chapters 1 and 3). Synthesis consists in mathematically deriving, and thereby explaining, consequences from this set of causes and laws, among which will be ones that can be tested experimentally, and then testing those consequences.
Both Mill and Newton distinguish the method they advocate from the “method of hypothesis” or “hypothetico-deductivism.” As Mill notes, the method of hypothesis omits the first step—the causal-inductive argument to each cause and law—and uses only ratiocination and verification. Advocates of the method of hypothesis claim, while Mill denies, that if ratiocination from a set of unsupported hypotheses leads to experimentally verified consequences, then not only are these verified consequences explained by the set of hypotheses but also they provide evidence for the set of hypotheses used to generate them.How exactly is Mill's deductive method committed to “evidential particularism”? It is so because for each cause and law introduced it requires a causal-inductive argument from observed phenomena. (Or else it requires deduction from other laws so inferred.) To be sure, it also requires that testable consequences derived from, and explained by, the combination of causes and laws introduced be empirically verified. But the latter by itself is not sufficient for verification of the system of laws introduced. Each cause and law in that system must be individually established (Mill's “inductive” step; Newton's “analysis”).
Let me put this more broadly in evidential terms. Suppose we want to obtain evidence for a set of hypotheses H. Mill is operating with a strong sense of “evidence,” one which is such that if e is evidence that h, then e provides a good reason to believe h.6 Moreover, in his discussion of the
6. Mill speaks of “ascertaining” or “proving,” though he recognizes that this does not produce the certainty of mathematics. For him, it is empirical and probabilistic. Although it is probabilistic, it requires something much stronger than increase in probability (Bayesian B-evidence).
“deductive method,” he is concerned with evidence for a set of hypotheses H that will provide a good reason for believing the entire set. In order to do so, we must obtain evidence for each hypothesis in H.
Let et be evidence for the individual hypothesis h. in the set H. Then, on the view I am attributing to Mill, as well as to Newton, there is evidence for the system H if and only if (a) for each hypothesis h. in the system H, there is evidence e. which is not necessarily evidence for the entire system H; and (b) consequences derived from and explained by the system are verified. This is the version of evidential particularism I will consider.A lot depends on how the concept of evidence is itself to be construed. Before turning to this, however, I will construct a version of evidential holism, which is suggested to me by writings of Quine, Duhem, and Whewell. I say “suggested” because these writings do not spell out the doctrine in as complete a manner as I would like. After formulating this doctrine, in order to bring out a central difference between particularism and holism, I will invoke an example discussed by both Whewell and Mill—the nineteenth-century wave theory of light. This will help us to focus on the basic issues.
2.