Friday, October 07, 2005

Armstrong on Modality

I read D. M. Armstrong’s new book Truth and Truthmakers. Armstrong is a great philosopher and metaphysician. He is also someone who insists on a materialist-naturalist metaphysics: there is one actual spatio-temporal world, and nothing more. I wanted to see how he handled the topics of modal truths and causality (my recent posts relating to this topic are here and here). I came away very disappointed. So, below you will find the spectacle of an untrained layperson criticizing a brilliant and prominent scholar on a blog. I ask anyone interested to please be willing to point out where I’m mistaken here.

First, let me offer some praise and briefly lay some groundwork. Armstrong is engaged in realist metaphysics in a serious way. The idea of truthmakers is a great one: every proposition, if true, has a truthmaker which makes it true. There must be “some way the world is in virtue of which these truths are true” (p.1). By considering different kinds of propositions, we build up and fill out a metaphysical structure which provides the truthmakers needed. The book introduces and describes the truth-truthmaking relation and his methodology for using it for metaphysics.

Next, Armstrong reviews his building blocks for truthmakers, in particular, his “states of affairs”. His version of this idea, introduced in previous papers and books, is central to the system. I’ll sketch it here and mention a concern with this approach, before I get to the specific issues of modality and causality.

Armstrong asks: What is the truthmaker for a proposition that a particular has some property? It’s probably not enough to say that the truthmaker just is that particular. What if it has other properties besides the one in question – do they come into play? What if it has relational properties which involve other particulars – do they come into play? How do we combine and delineate the relevant parts needed for a truthmaker?

Armstrong digresses to briefly outline all the combinations of approaches to particulars and their properties which have been proposed: (are properties universals or tropes, do they adhere to a particular’s substance or is a particular just a bundle, etc.) Armstrong notes some of the advantages and disadvantages but also says not all of these are important for truthmaker theory (he happens to favor immanent universals as properties, by the way). The most crucial issue appears to involve how to think of the relation between the particular and its properties. Armstrong says there “must exist states of affairs to get us beyond the ‘loose and separate’ entities” and thereby provide the truthmaker. He says: “States of affairs must be introduced as additions to the ontology.” (Emphasis original p.49). So, states of affairs are introduced to hold the particular and its properties together by fiat.

Moving from old-fashioned individuals to states of affairs brings in extra ontology for free. It seems to me this is going beyond traditional materialism. I was impressed by Bill Vallicella’s in-depth critique of this notion in his recent book, but I won’t try to do justice to his argument now. I’ll just express my feeling that Armstrong’s binding of properties into individuals via states of affairs seems too easy.

You might also appreciate this when you see how Armstrong builds up to larger states of affairs. Larger states of affairs can encompass and bind smaller ones to provide appropriate truthmakers when needed. To provide a truthmaker for negative truths about the world (e.g. there are no unicorns), Armstrong utilizes the notion of a general or maximal state of affairs for the whole world. This mega-fact sets limits for what is actually in the world. The fact that there is no unicorn supervenes on the highest-order state of affairs. And just as we bound a property to a particular at no ontological cost, Armstrong asserts that the mega-state of affairs encompasses the conjunction of all its smaller constituent states of affairs with no additional cost, despite being “something more than the mereological sum of its constituents” (p.72). (there is an obvious regress issue here, but Armstrong argues it is a benign one.)

The issue of contingency vs. necessity comes up in a few contexts in the book, for instance in whether the connection between a particular and its property is contingent or necessary. But chapter 7 is explicitly on the question of what the truthmakers are for modal truths.

He starts it off thus: “It seems to me very surprising that so many good philosophers consider that huge metaphysical commitments must be made in order to give an account of these truths.” (p.83). He mentions David Lewis and Alvin Plantinga specifically. In Lewis’ system (simplified), something is necessarily true, if it is true in all possible worlds; it is possibly true if it is true in a subset of worlds. Something in our world is contingent if it is possibly not true in other worlds. Famously, Lewis says these possible worlds all exist concretely. Armstrong says “...we ought to be looking for quite modest truthmakers, fairly deflationary truthmakers, for these fairly unimportant truths of mere possibility.” (pp.83-4)

Here’s his truthmaker argument for a contingent truth (my paraphrase-see p.84 for this in his notation):
1. Assume that (state of affairs) ‘T’ is the truthmaker for truth ‘p’.
2. Assume p is contingent (my emphasis).
3. p entails 'it is possible that not-p' (from #2 and the definition of contingency).
4. Therefore, T is also a truthmaker for 'it is possible that not-p', and so T is a truthmaker for a contingent truth p. (This is via #1, #3, and from something Armstrong calls the 'entailment principle' that he argues for earlier in the book).

That's it: there's no more. But shouldn't truthmakers for modal truths distinguish a contingent truth from a necessary one? We could use this argument's structure equally well for the case where p is assumed to be necessary. Therefore there is nothing in the truthmaker which determines the truth to be contingent vs. necessary. Therefore it is NOT an adequate argument for a modal truthmaker at all!

I was stunned by what seemed to me to be the obvious inadequacy in the argument here. Could I have missed something? I have read the chapter more than once, but I'll keep at it. As I said in my introductory paragraph I invite anyone reading this that is familiar with this to “set me straight”.

Here’s something Armstrong says to close this section of the chapter (7.2, p.86): “…I still favor the hypothesis that particulars, properties, states of affairs and laws of nature are all of them contingent beings.” There is no argument for this, it is just a preference. While it is a hypothesis that certainly seems consonant with our human intuitions, it in no way follows from his system. With no way to distinguish the contingent from the necessary, I would think it would be more consistent and simpler for him to give up the intuition and argue that all of these constituents and relations are necessary.

I will skip over more material and just briefly touch on Chapter 10 which deals with causation. Armstrong wants to affirm real causal connections, not just Humean regularities. He terms this “singular” causation. How does one do this? He acknowledges the difficulty of the problem and mentions some possible approaches. You can’t just add causal connections between states of affairs without seeming to add still more gratuitous ontology. He favors an idea that the universals which are parts of states of affairs might be vehicles of causal or “cause-like” connections at a higher level than the state of affairs involved. Using universals to do the work would help explain the existence of lawful regularities across the states of affairs of the world.

This work on causal structure within his framework of states of affairs seems fine on its own. But I hold that without an explanation of modality, the discussion is relatively empty. I have thought the heart of real causality is the idea that things could have been some other way. So the earlier lack of adequate treatment of modality makes Armstrong’s discussion of causality of minor interest to me.


Anonymous said...

He is also someone who insists on a materialist-naturalist metaphysics: there is one actual spatio-temporal world, and nothing more.

This is kind of ancillary to your post, but that claim alone is highly questionable, considering the likelihood that space and time are actually emergent properties of some more fundamental reality.

Steve said...

Right. For the purposes of the issues here, he might say: "OK, substitute whatever other microphysical structure you want, and the specific arguments still hold".

But in the bigger picture, I'm sympathetic to the idea that the impulse to preserve a materialist metaphysics depends on a simple spatio-temporal mechanistic concept of physics which may be (further) undermined by physicists themselves as they search for a more fundamental theory.

Jennifer said...

How does one qualify quantums of power, if not via materialist or idealist subsets of some unidentifiable whole?

Perhaps Nietzsche subsumes Lukacs in this sense, since reification (at the level of idealist abstraction) becomes, without a doubt, a materialist force, and so, all is subsumed under a generic force which is represented by the paradigm of an organic "will to power".


Unknown said...

If I am understanding you correctly, you are saying that Armstrong's system invokes truthmakers that work equally well for necessary events and possible events. The failure to demonstrate how we can distinguish between a truthmaker for a necessary event and for a possible event makes his explanation of causality incomplete. There is no way to account for modality. Is this your claim?

Unfortunately, I did not read this work of Armstrong's so I cannot really get into it. What I am curious about is the last paragraph of your post. You say: "I have thought the heart of real causality is the idea that things could have been some other way." Can you please expound on that?

Steve said...

Yes, that is my claim.

When something happens, we seem to know that, counterfactually, something else could have happened instead. If it really couldn't have, then there is no "real" possibility. Modality is so crucial to our reasoning that I think it is real and must be accounted for in a metaphysics.

With regard to that last statement about causality, I think an interesting model which combines modal ontology and causation is one where there is a space of possibilia which are resolved into concrete events, and that this process of actualization is causation.

Anonymous said...

The argument works because truthmaker only says that we need truthmakers for true propositions-
So if ever we have 'P is true', then we need to find a truthmaker for P. But if we ever have 'P is contingent', then we need not find a 'contingent maker'.
So assume T makes P true,
and that P is contingent,
then 'possibly not P' is true,
and so via entailment thesis T is a truthmaker for this modal truth.

Perhaps you could challenge the argument by saying that it gets things topsy-turvy. Surely it is that 'possibly not p' and 'possibly p' are both true, which makes P contingent. Thus given that P is true and that T is a truthmaker for P, we are still in need of a truthmaker for 'possiby not P' before P is contingent.

After all what is the truthmaker for 'It is true that P is contingent'? It is not simply P, but the state of affairs of P's being contingent. But if this is an acceptable truthmaker- then surely it is acceptable for the state of affairs of P's being true to be a truthmaker for 'P is true'.
In which case the whole truthmaker project is ruined.

Steve Esser said...

Thanks for the comment - that strikes me as a good brief take on the problem.