Notes
I discuss the first part, by implication, in a paper on the epistemological problem of memory which will appear in a forthcoming issue of thePhilosophical Review.
The Analysis of Knowledge and Valuation (La Salle, Ill.: Open Court, 1946), p. 358. All page references are to this book.
See Carnap,Logical Foundations of Probability (Chicago: University of Chicago Press, 1950), p. 329; and Lewis,op. cit., pp. 237–39. Lewis does not explain the logical status of such theorems on his view of probability. But since the theorem would be accepted by everyone for certain contexts, let us not question the argument on that score.
Human Knowledge, p. 188; see alsoInquiry into Meaning and Truth, 1948 edition, p. 125.
Lewis seems to suggest, in his recent reply to criticisms by Everett Nelson (“A Comment,”Philosophical Review, 63:193–96) that particular phenomenal statements rather like (1) to (8) which are logically independent of each other can be shown to support one another if they are all so related to one physical object statement that the falsity of any one of them would disconfirm it. (Suppose each phenomenal proposition were something like “There seemed to be a white patch in front of me, and when I seemed to reach out and touch, I seemed to feel a smooth surface.” According to Lewis, a proposition like this is related to “There is a piece of white paper in front of me” in such a way that if the first two clauses were true but the third false, the physical object propositions would be fairly strongly disconfirmed.) But I do not see why this should be so on a phenomenalist view (meaning by “phenomenalism” the theory intended under this title by contemporary philosophers like Ayer and Price). For, on this view, what a physical object statement means is a conjunction of complex hypothetical phenomenal statements. (Like Nelson I had, despite some puzzling passages in Lewis's book, e.g., pp. 200–2, taken Lewis to be a phenomenalist in this sense; but apparently, from what he now says, this was a mistake.) Thus a physical object statement like the one above means something like “If p, then q; if r, then s; if t, then u ...” Now I agree that if we have evidence, “p and q,” this is some support for the physical object statement, just as “p but not q” would be disconfirming for it. But is there not something wrong with saying this changes the probability of “if r, then s”? (It does seem as if it ought to change it, since the physical object statement entails “if r, then s.”) To see that there is something wrong here, let us make up a super-theological language. And let us consider a statement S in this language, which is the analogue of the physical object statement. Now let us specify that what Smeans is “If there is water, wood will float on it; if a person sees blue, he can see yellow; and if human beings die, their memories will survive and there is a God.” Now suppose we verify that there is water and that wood floats on it, and that a person who sees blue can also see yellow. So far, we have got confirmation for S; or at least, if we had found, contrary facts, we should have disconfirmed S. But does evidence verifying the first two propositions meant by S also confirm human immortality and the existence of God? Surely something is wrong here. How can the fact that we make up a language with a sentence in it which means or entails the conjunct of these three propositions, render these several propositions congruent when they were not so before?
I am grateful to Professor Patrick Suppes for helpful comments on an earlier draft of the foregoing paper.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Brandt, R.B. A puzzle in Lewis's theory of memory. Philos Stud 5, 88–95 (1954). https://doi.org/10.1007/BF02223697
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF02223697