A. Three basic theses
    1. There are (positive) universals.

    2. Theoretical terms have denotation and are analyzable by means of observation terms, logical terms, and quasi-logical terms (e.g., property, part, event, etc.); however, statements involving theoretical terms are not reducible to (equivalent in meaning to) statements about the observable properties and relations of observable objects.

    3. Nomological terms and causal terms are theoretical and not observational; therefore, they are analyzable by means of observational terms, logical terms, and quasi-logical terms, though the analysis is not reductive, in the sense that it analyzes theoretical terms merely by means of observational terms. Nonetheless, claims Tooley, theoretical statements can, by means of the Ramsey-Lewis method, be translated into statements that have no theoretical terms.

Ponder conditions 2 and 3 for analyzability on p. 26. Tooley's argument for the claim that nomological terms introduce non-extensional contexts is, I believe, flawed, though I will not argue the point here other than by pointing out that we might construe (1) as
    (1*) Water and sugar are such that it is a law that latter dissolves in the former,
which is not susceptible to his argument. However, the main argument for the analyzability of causal terms is this:
    "... even if it turns out that some non-reductionist account of causation is correct, it will still be true that there is no observable difference between a world in which all of the non-causal facts are as they would be if states of affairs were causally related, and a world in which the states of affairs in question really are causally related ... So there cannot be any properties or relations, with which one can be directly acquainted that are associated with causal terms. Consequently, neither causal terms, nor nomological terms, can be treated as primitive, however familiar some of them may be. Analysis is required" (p. 28).
Comments: First of all, we are given no general account of what is to count as an observational term; in fact, however, it seems quite clear that not only causal terms, but also natural kind terms, must be counted as theoretical terms on this account, at least if the latter have connotations about causal powers, processes, etc. Indeed, it may very well be that an Aristotelian will not know whether to accept or reject the two theses of Humean supervenience precisely because of this ambiguity. (In general, much of this book is written at such a high level of abstraction that it is hard to know what to make of it.)

B. Humean Supervenience

At any rate, in both the book and the article Tooley is concerned with undermining Humean supervenience even while hanging on to the strong form of Humean empiricism embodied in the claim that all nomological and causal terms are theoretical and not observational terms. The thesis of Humean supervenience relevant to causality goes like this:

    The truth values of all singular causal statements are logically determined by the truth values of statements of causal laws, together with the truth values of non-causal statements about particulars.
This thesis is spelled out in a bit more detail on pp. 94-96 of the article in Midwest Studies. Once again, it is hard to know exactly how to react to this thesis without knowing more about what is to count as a causal statement or a causal law. (It also seems committed to determinism as it stands, but perhaps this can be remedied).

Tooley argues against this thesis at length in the article (which corresponds to chapter 6 in the book, although in the book he comes down clearly on the side of the "intermediate position," which lies between the supervenience view and the singularist view, according to which singular causal statements are basic independently of whether or not they fall under causal laws.

The components of Tooley's considered view are these:

    (1) Causal terms and statements are theoretical rather than observational and they refer beyond what is observable; what's more, it is possible in principle to be rationally justified accepting some causal statements as true.

    (2) Causal terms and statements are analyzable.

    (3) All causal connections presuppose underlying causal laws.

    (4) Humean supervenience is nonetheless false. That is, the truth values of singular causal statements are not "logically determined" by the truth values of statements of causal laws, together with the truth values of statements about particulars.

    (5) Causation is that theoretical relation that determines the direction of the logical transmission of probabilities.

We have already seen one argument for (2). On pp. 247-249 Tooley trots out three more arguments:
    (1) The formal structural properties of causal relations can be explained only if causal concepts are analyzable. (I must confess that I don't understand this argument).

    (2) If causal concepts are unanalyzable, then the acquisition of causal knowledge presupposes the existence of a somewhat mysterious capacity.

    (3) Non-causal facts can provide evidence for causal claims.