Abstract

In lieu of an abstract, here is a brief excerpt of the content:32. BAYESIANISM AND ANALOGY IN HUME'S DIALOGUES Wesley Salmon has recently focussed attention on Hume's consideration of the argument from design for the existence of God in the Dialogues Concerning Natural Religion, by construing it according to a Bayesian account of inductive inferences to causal hypotheses. Salmon argues that an interpretation of the argument from design, considered by Philo and Cleanthes in the Dialogues, as an appeal to analogy is not a correct interpretation of Hume's argumentation. I would like here to argue that Salmon's account is too far afield of Hume's actual pattern of arguing to be taken as an accurate account of the discussion in the Dialogues. I would also like to characterize Hume's account of causal arguments in the Dialogues more closely, and to show how it is connected with his notion of resemblance. In doing so, I shall defend the thesis that an appeal to analogy is after all the fundamental form of argument that is considered in the Dialogues. I. Salmon argues that the discussion among the characters in the Dialogues, principally between Philo and Cleanthes, centers about the various probability terms entering into Bayes' s Theorem in the form: P(C/A&B).P(B/A) P(B/ASC) = ---------------------------------, P(C/ASB) -P(B/A) + PÍC/A&B) -P(B/A) where in the context of the Dialogues, "A" stands for any event of coming-into-being, "B" stands for any event that is a product of intelligence, and "C" stands for any event that 2 exhibits order or design. The various probabilities that enter into Bayes's Theorem can, then, be characterized in the present context more thoroughly. For example, P (B/A&C) is the probability that an event of coming-into-being which displays order is the product of intelligence. This 33. probability, if high, would tend to support the argument from design, and, if low, would tend to undermine it. In order to evaluate P(B/A&C) by using Bayes' s Theorem, we need to be in a position to find the following three probabilities: P(B/A): This is the probability that any event of coming-into-being is a product of intelligence. If we know its value, it is easy to calculate P(B/A), the probability that any event of coming-into-being is not a produce of intelligence. For P(B/A) = 1 - P(B/A). Both these probabilities are called the prior probabilities of the problem. P(C/A&B): This is the probability - which Salmon typically calls one of the two likelihoods of the problem - that an event of cominginto -being which is a produce of intelligence does exhibit order or design. P(C/A&B): This is the probability - which Salmon calls the other likelihood of the problem that an event of coming-into-being which is not a produce of intelligence exhibits order or design. Now Salmon is not merely arguing - though he is also arguing - that Bayes' s Theorem, with its constituent probabilities so understood, is relevant for evaluating the argument from design. He is arguing that Bayes' s Theorem is relevant for interpreting Hume's thoughts about the argument from design. Salmon admits that "Hume must have been unaware of Bayes" s Theorem." But he claims that Bayes' s Theorem is nevertheless relevant for interpreting Hume. He suggests a number of reasons for this. First, the argument considered by Philo and Cleanthes is weakened, he says, if it is 34. 4 represented as a mere appeal to analogy. Secondly, he says that the correct strategy for analyzing any causal hypothesis is Bayesian, and undoubtedly Philo and Cleanthes are debating about causal hypotheses. Thirdly, Philo and Cleanthes show by the end of the second dialogue that they are aware the argument is more complex than a simple appeal to analogy. Fourthly, with respect to the probabilities one needs to know in order to use Bayes' s Theorem in the relevant way, Hume's characters and especially Philo "devote considerable attention to each of the three types of probabilities in their discussion." Fifthly, Hume just does in fact recognize that the argument from design is "something deeper and more subtle than a simple appeal to D analogy." We can greatly...