What are Propositions?

Abstract

This paper defends the view that propositions – that is, what are picked about by complement clauses and the range of quantifiers like that in ‘Sanna believes all that Matti said' – are states of affairs. States of affairs – and thus propositions – are not, in the primary sense, representational; what is representational and what is true or false in the first instance are mental states and sentence tokens that represent propositions. There is, it is argued, a derivative sense in which propositions are bearers of truth, but truth in that sense is a derivative, non-explanatory notion. This view is contrasted with views like the one Scott Soames develops in What is Meaning?. It's argued that this view is superior to Soames' in various ways.

Keywords::

• propositions
• representation
• propositional attitudes
• truth
• states of affairs

Notes

^1. Henceforth, I often use ‘property’ as a blanket term for properties and relations.

^2. One might think of the matter as follows. For a state of affairs to obtain is for various objects to instantiate various properties. (Better: for various objects (and properties) to instantiate various properties (higher order ones, in the case of properties)). The properties that are states of affairs will thus be individuated as finely as the configurations of objects and properties that realize those properties: A state of affairs is the property of this configuration of objects and properties obtaining. In general, then, necessarily equivalent states of affairs are distinct.

Of course the state of affairs that Twain is Twain is the state of affairs that Twain is Clemens, just as the Russellian proposition that Twain is Twain is the Russellian proposition that Twain is Clemens. If one – as I am about to – identifies propositions with states of affairs, one will have to explain how the propositional attitude ascriptions ‘Mary believes that Twain is Twain’ and ‘Mary believes that Twain is Clemens’ can differ in truth value. I have discussed this issue in considerable detail elsewhere. (Richard, 1990, 2013) Though those discussions don't identify propositions with states of affairs, what is said there about the semantics of attitude ascriptions transfers straightforwardly to the view of propositions taken here.

^3. The modality here is metaphysical.

^4. I ignore context sensitivity throughout.

^5. Alternatively, one might associate ways of instantiating with connectives and such, and give an inductive definition of the state of affairs represented by a sentence. Assuming that we already have a definition of truth for the language we're studying, there's no advantage I can see in doing this.

^6. Why do I say ‘typically tied to its behavior’? Isn't it definitional that beliefs tend to interact with desire to produce the relevant sort of behavior?

I don't think that this is true, much less definitional. There are various pathologies of belief. (Davies and Coltheart (2000) give a nice discussion of some of them.) What is striking about such pathologies is that in them, states we are inclined to call beliefs and desires do not interact in normal ways. In particular, when someone suffers from such a pathology, their belief (-like) state often does not interact with standing desires to produce behavior. The patient suffering from Cotard syndrome who thinks she is dead and wants the dead to be cremated does not report to a crematorium. (Note that this does not seem to be due to ‘lack of integration’ of the attitudes in any normal sense of integration.)

I think the idiom of belief is the right one to use in such cases, though I grant that for some such cases it may be indeterminate whether they are cases of belief. But if these cases are cases of belief (or even if it is simply indeterminate whether they are), their existence shows that the attitudes are not to be given crisp definitions on which the claim that something is a belief entails that it realizes a stereotypical belief+desire → action functional organization.

I'm indebted here to discussions with Bob Matthews.

^7. The counterfactual here is not to be understood in terms of a possible worlds semantics, with the whole true if the consequent is true at the ‘closest antecedent worlds’. I don't see this as a problem, at least not one for the proposal I am making about truth. Possible worlds semantic accounts of conditionals are useful heuristics. That doesn't mean they are literally correct.

^8. Soames, 2010, 102. Subsequent references to Soames 2010 are indicated parenthetically in the text.

^9.Soames, ms. A.

^10. Why, you may be wondering, does Soames think it absurd to say that a property can have truth conditions? According to Soames

ascriptions to propositions of what can be said of act types, as well as ascriptions to act types of what can be said of propositions, strike us as bizarre or incoherent. …it would be incoherent to say “*the proposition that John is brilliant is what I just did” or “*What I plan to do (when I plan to predicate brilliance of John) is false”. …it would be absurd to say “What I believe, and what Gödel proved, is something I just did.” (101–102, italics in original)

But surely it is exactly as ‘absurd’ to say ‘the proposition that John is brilliant just happened / occurred’ or ‘What happened in the bathroom at 2 pm was false’; likewise for ‘What I believe, and what Gödel proved, is something that happened in the bathroom.’ This sort of linguistic evidence doesn't allow us to decide that event types, instead of act types, are propositions.

Soames himself says, of the idea that propositions are events

Since this is a new way of thinking of propositions…it may be a bit surprising to be told that they are things that occur. However, I don't see this consequence as incoherent…. the theoretical advantages of thinking of propositions in this way are substantial, and justify this modest extension of our view of them. (103–104)

Fair enough. But exactly the same thing might be said about the idea that propositions are act types. And exactly the same sort of thing ought be said of the idea that propositions are states of affairs.

^11. As I observed in note 7, the interpretation of the modal here is somewhat delicate. What's meant, of course, is that a state of affairs P is true iff (were one to have a belief whose content was P, the object of one's belief would actually obtain). More generally, to say that P is true at w is to say that were one to believe P, the content of one's belief would obtain at w.

^12. Do not interpret me as here endorsing a minimalist notion of truth, but rather as agreeing with the minimalist's observation that the (propositional) notion of truth has enormous descriptive utility.

^13. ‘thinking’ here is supposed to be a generic mental attitude, one common to entertaining, believing, assuming, doubting, knowing, and so on; in this sense, one thinks that grass is green when (for instance) one thinks that if grass in green, then hay is beige.

^14. Soames himself poses such an objection to a ‘deflationary’ account of propositions that takes them to be nothing more than theoretical posits introduced by the theorist. On such an account, a proposition might be something along the lines of an ordered pair of a property and the right number of constituents; to entertain it is to ascribe the property to the constituents. This means that the proposition that A and B is < the property of being jointly true>, < the proposition that A, the proposition that B>; believing it is ascribing joint truth to the propositions.

The deflationist's propositions are abstract, set theoretical structures. Soames objects that we don't think about such things when we think, for example, that snow is not crimson, and thus can't be credited with making them objects of predication. As he puts it:

…it is not obvious that agents can correctly be said to have any of the theorist's abstract structures [any of those she might identify with propositions within the context of her theory] in mind in a sense sufficiently robust to make them the targets of the agents' acts of predication. (97)

Soames concludes that a deflationary account of propositions is not in offing: we must adopt a ‘realist‘ account of propositions, on which they are ‘parts of the reality being modeled, rather than merely components of the model.’ (98)

What is not obvious, I would say, is why propositions must be something that an agent ‘has in mind in a robust sense’. Their primary purpose, as I see it, is a classificatory one. My point in this section, as will become clear, is that it is implausible that there is a single ‘robust’ sense in which someone needs to have to have a proposition ‘in mind’, in order to be correctly said to believe or entertain it.

^15. Much the same point applies to (quite complicated) account Soames offers at the end of Soames, 2010 for what is involved in beliefs expressed by sentences involving quantification, structured singular terms (like those involving function symbols), and complex predicates. Thanks to the editors for suggesting that I note this.

^16.Soames, ms. B.

^17. This is just the objection to the deflationary account of propositions discussed in note 14.

Soames presumably made the relations R& and RV relations to n-tuples so that he could avoid variations to standard objections to Russell's multiple relation theory of belief. The multiple relation theory, as Russell presented it, requires an infinity of belief relations: the three place relation I bear to you and happiness when I believe that you are happy, the four place relation I bear to you, admiring, and Frege when I believe that you admire Frege, the five place relation ….. This seems objectionable because it leaves us without a single relation to assign to the verb ‘believes‘. If the number of constituents in p and q controls the relation used to form their conjunction, it's not altogether clear that one could give a finitary theory of propositions that would, for example, assign to each conjunction the correct conjunctive propositions.

Even if this problem and the one in the text can be evaded, the resulting theory would still be liable to the objections lodged below.

^18. Whether it's reasonable (and thus, I would say, correct) to say that something has a belief depends on factors that vary across the situations in which we ascribe (or withhold) ascription of the attitude. I discuss this in Richard 2013.

^19. We will, for example, generalize from the paradigm in one way if we are trying to give an account of everyday attitude ascription, perhaps in quite another if we are doing cognitive science.

^20. This is a revised version of comments on Soames, 2010 delivered at an APA Author Meets Critics session in December 2011. I'm grateful to Soames for his comments on that version of the paper, as well for comments from David Liebesman, Susanna Siegel, Sebastian Watzl, the participants at a workshop on propositions at the École Normale Superior in April 2012 and the editors of this volume.

Additional information

Notes on contributors

Mark Richard

Mark Richard is a Professor of Philosophy at Harvard University. He is the author of Propositional Attitudes, When Truth Gives Out, Meaning in Context: Context and the Attitudes, and the forthcoming Meaning in Context: Tense, Truth, and Tribulation.