A World of States of Affairs. - book reviews

Mind, July, 1998 by Phil Dowe

A World of States of Affairs, by D. M. Armstrong. Cambridge: Cambridge University Press, 1997. Pp. xiii 283. P/b 14.95 [pounds sterling].

David Armstrong's most recent (and in some senses ultimate) book is a defence of his view that the world is a world of states of affairs. States of affairs, known to many philosophers as facts, are things like a particular having a property or standing in a relation to another particular, so their constituents are particulars, properties and relations. Not all states of affairs are first order: there are facts about facts--for example, central to the picture are "strong" laws of nature. The constituents of "molecular" states of affairs include "atomic" or simple states of affairs, but Armstrong leaves open the issue of atomism, that is, whether there are any atomic states of affairs. This view, that the world is a world of states of affairs, Armstrong attributes to Wittgenstein and Russell, and also Skyrms, but reserves the strongest notice of influence for his teacher at Sydney, John Anderson, and his view that reality, though mind independent, has an essentially propositional structure.

States of affairs are contingent existents. Armstrong holds the Humean and Tractarian thesis of the independence of first order state of affairs, that no first order state of affairs entails or precludes another distinct first order state of affairs. But he argues against the view that all states of affairs are independent exceptions include Russellian facts of totality (e.g. that such and such a collection of states of affairs is all there is) and laws of nature which Armstrong takes to be "contingent connections between state-of-affairs types". States of affairs are the truth-makers of all contingent and necessary truths. Mere possibilities are fictional combinatorial reconstructions of actual states of affairs. Numbers are internal relations. Classes are types of possible states of affairs. This world of independent states of affairs is tied together by causation--an all-embracing causal net which gives the world a unity other factualists have had difficulty providing.

In what follows I will focus on just one theme in Armstrong's book--powers and causation. If there is a brief characterisation of how Armstrong's metaphysics has matured over the years then it could be this: causation has come into clearer and clearer focus as the fundamental issue of metaphysics.

Clearly there is a link between properties and powers. Particulars have powers in virtue of their properties, and having a power is in any case a property of that particular. Categoricalism (which Armstrong still defends) says properties and relations have natures distinct from the powers they bestow.

One can distinguish between the dispositional property itself, the initiating circumstances (which may not occur), and the manifestation of the power (which may not occur). When a particular does manifest a power, we have a case of singular causation. Then the total cause--which will include the particular having the relevant power, and the initiating circumstances--brings about the manifestation in some standard way (not by a "deviant" chain). Armstrong assumes that every token of the causal relation is governed by some law.

Since singular causal relations are necessary for the manifestation of a power, it follows that a dispositionalist account of the causal relation is not available. Dispositionalists therefore need to make an exception for the relation of causation--causation must be categorical. Also, the consequents of conditionals associated with dispositions must be causal statements, which means that a conditional account of causation must be ruled out since it would then be circular.

When a particular has an unmanifested power, the particular is not related to the potential manifestation (all terms of a relation must obtain for a relation to exist). Also, since Armstrong rejects negative universals in general, omissions are not causes. "Lack of water caused his death" may be true, but does not express a causal relation, and no relevant power has been manifested. Also, where the law is indeterministic, and the total cause occurs and the manifestation of the power does not occur, then again, we do not have a case of causation. This means that propensities do not involve probabilistic causation as such, but just a probability of causing.

According to Armstrong the concept of singular causation is conceptually primitive. For example, as already mentioned, causation cannot be analysed in terms of conditionals. The singular causal relation "C causes E" does not entail the counterfactual "if not C then not E", although it is often true, and so is supported in some sense. What does entail the counterfactual is the conjunction of "C causes E" with the state of affairs that there is no C' such that C' causes E, and the state of affairs that E does not occur without a cause. But the additional requirements involve singular causation, so causation cannot be analysed in terms of such conditionals. (Although this conditional might be thought to analyse necessary causation, whereas the conditional relevant to sufficient causation is "if C then E".)