Not the optimistic type

Abstract

In recent work, Peter Hanks and Scott Soames argue for the type view, according to which propositions are types whose tokens are acts, states, or events. Hanks and Soames think that one of the virtues of the type view is that it allows them to explain why propositions have semantic properties. But, in this paper, we argue that their explanations aren't satisfactory.

Keywords::

• propositions
• semantics
• types
• Hanks
• Soames

Acknowledgements

For comments and discussion, thanks to participants in seminars at Ohio State and the University of Manitoba; to participants at talks at the CPA, CSMN at the University of Oslo, the Semantics Workshop of the American Midwest and Prairies, the Society for Exact Philosophy, SPAWN, and the WCPA; to Scott Brown, Einar Duenger Bøhn, Sam Cowling, Wesley Cray, Olav Gjelsvik, Jill Isenberg, Ali Kazmi, Sandra Lapointe, Kirk Ludwig, Cathy Muller, Anders Nes, Greg Ray, Georges Rey, Craige Roberts, David Sanson, Andreas Stokke, Judith Tonhauser, and the editors of this volume; and especially to Peter Hanks, David Liebesman, Jeff Speaks, and Joshua Spencer. Thanks also from the first author to CSMN for hospitality in Spring 2012.

Notes

^1. See, for example, Hanks 2007, 157–159; 2011, 13, 17, 41, 48; Soames 2010b, 2–3, 6–7; 2014b, 226.

^2. Some think that propositions have truth-conditional, but not representational, properties. See, for example, Speaks 2014. We think that Speaks is right about this, but we set that aside in the text. Even if propositions don't have representational properties, they still have semantic properties, since they still have truth-conditional properties; so there is still a need to explain their semantic properties.

^3. See, for example, Hanks 2007, 153; Soames 2010b, 114. For a different view of predication, see Hanks 2011, 14–16, 2013, 161–164.

^4. There are differences among Hanks's initial view (see Hanks 2007, 151–152); the view that Hanks currently favors (see Hanks 2011, 12–23, 2013, 156–165, this volume); Soames's initial view (see Soames 2010a, 116–123, 2010b, 99–107, 2012, 216–219, 2014a); and the view that Soames currently favors (see especially Soames 2014b, 241 n. 16). Hanks's initial view is one on which propositions are types whose tokens are acts and states; the view that Hanks currently favors and the view that Soames currently favors are views on which propositions are types whose tokens are only acts; and Soames's initial view is one on which propositions are types whose tokens are only events. But, when it comes to explaining the semantic properties of propositions, these differences don't matter. So, in the text, we consider Hanks's initial view, the view that Hanks currently favors, Soames's initial view, and the view that Soames currently favors together as instances of the type view, according to which propositions are act, state, or event types.

^5. King (2007, 2009, 2013, 2014) is also an optimistic naturalist. (But he does not accept the type view.) For a criticism of King's view, see Caplan and Tillman forthcoming.

^6. Hanks (2013, 161) makes a parallel claim about truth-conditions.

^7. Italics in original.

^8. Soames 2010b, 94.

^9. Hanks 2007, 159.

^10. Hanks 2011, 48.

^11. What Hanks (2007, 159–160) says is the following: ‘in asserting that Smith is tall a subject applies the property of tallness to Smith and thus represents the state of affairs that Smith is tall. The type of speech act of which this is a token, i.e. the assertive proposition that Smith is tall, inherits this representational content. Because any instance of a speaker applying the property of tallness to Smith represents a certain state of affairs, the type that unites these instances also represents that state of affairs.’ (Hanks (2007, 151–152, 2011, 16–17) distinguishes assertive, interrogative, and imperative propositions. In this paper we are focusing on what he regards as assertive propositions, which include the propositions expressed by declarative sentences such as ‘Mother Teresa supplicates’.)

^12. Soames 2010b, 107; italics in original. See also Soames 2010b, 7.

^13. Soames 2014b, 230; italics in original. See also Soames 2014b, 234 n. 5, 235, and 239.

^14. What Hanks (2011, 41) says is the following: ‘A token assertion that Le Carré is a novelist is a paradigm of the sort of thing that can be true or false, and the type [that Hanks identifies with the proposition that Le Carré is a novelist] inherits its truth-value from its tokens. … The reason the proposition that Le Carré is a novelist is true if and only if Le Carré is a novelist is that tokens of this proposition, particular assertions that Le Carré is a novelist, are true if and only if Le Carré is a novelist.’ See also Hanks 2013, 161.

^15. See, for example, Soames 2010b, 6.

^16. Soames (2014a, 96–97) talks of ‘conceivable instances.’ When he says that every (possible or) conceivable token or instance of T has F, he presumably means that the following is (impossible or) inconceivable: x is a token or instance of T, and x lacks F.

^17. This problem is noted in King 2013, 91 and Speaks 2014, 165. Speaks attributes the problem to Caplan. The problem is discussed in Tillman and Caplan 2011.

^18. Kleinschmidt and Ross (2012, 134–135) offer a related counterexample to an inheritance principle that they attribute to Liebesman 2011.

^19. Elsewhere, Hanks (this volume) might have in mind a restriction to evaluative properties. Such a restriction still faces the problem of sub-types discussed below in the text (and the problem of super-types discussed below in note 25). In addition, the worry about ad hocness might remain. See note 26.

^20. We owe this point to Jeff Speaks. See Speaks 2014, 165.

^21. Hanks suggested this reply in correspondence.

^22. See also Hanks 2013, 156–165, this volume. On the view that Hanks currently favors, not any act of referring to Mother Teresa will do; only acts of referring to Mother Teresa that are tokens of the relevant semantic reference type will do. See Hanks 2011, 26–32, 2013, 157–161.

^23.T is a sub-type of T* if and only if, necessarily, any token of T is also a token of T*.

^24. See Hanks 2011, 37, 47; 2013, 176; this volume. According to Hanks, these sub-types might not be expressed by ‘Mother Teresa supplicates’ or referred to by ‘that Mother Teresa supplicates’ relative to any context, but there are contexts relative to which ‘Sam believes that Mother Teresa supplicates’ is true if and only if Sam stands in the belief relation to one of these sub-types. See Hanks 2011, 36–38, 2013, 176–177; this volume.

^25. Nor does it seem plausible that there are contexts relative to which ‘Sam believes that Mother Teresa supplicates’ is true if and only if Sam stands in the belief relation to one of these gastronomical sub-types. (See note 24.)

Hanks (2011, 23) acknowledges that there are sub-types of propositions whose tokens are complex acts that consist in part in referring to objects in very specific ways: for example, ‘referring to George in a loud voice using one of his nicknames while holding an umbrella and standing at a train station.’ But it doesn't seem plausible that such sub-types have semantic properties. Nor does it seem plausible that there are contexts relative to which belief ascriptions are true if and only if agents stand in the belief relation to such sub-types. For a contrary position, at least about a sub-type whose tokens are complex acts that consist in part in referring to Hillary ‘during a Cabinet meeting on a Thursday,’ see Hanks, this volume.

Here's a related problem. Consider act, state, or event types in which (v) someone predicates something of Mother Teresa, (vi) someone predicates something of someone, or (vii) someone does some predicating. Every conceivable token of types (v)-(vii) has the properties representing something and having truth-conditions; so, by Semantic Inheritance, types (v)-(vii) have those properties, too. But it seems that, on the type view, types (v)-(vii) are not propositions. So presumably types (v)-(vii) should not have any semantic properties. See Speaks 2014, 165.

Hanks allows that in the vicinity of supplicate are super-types that are also propositions, where T* is a super-type of T if and only if, necessarily, any token of T is also a token of T*. But these super-types usually have to do with referring to objects using any referential device whatsoever. (See Hanks 2011, 35, 2013, 174. See also note 22.) So the super-types that Hanks discusses don't include types like (v)-(vii). (See also Hanks, this volume.)

^26. Or, if we restrict inheritance to evaluative properties (see note 19), what is it about evaluative properties in virtue of which they are inherited, while other properties need not be? This question retains at least some of its force, we think, even if types other than propositions inherit evaluative properties, provided that they need not inherit other properties.

^27. Hanks (this volume) might say that evaluative properties are inherited by all types alike (see note 19). But that leaves the problems of sub- and super-types. (On super-types, see note 25.) And, in any case, Propositional Semantic Inheritance would still be subject to the counterexamples to Semantic Inheritance that we discussed above in the text: that is, non-derivatively representing Mother Teresa as supplicating and non-derivatively being true if and only if MotherTeresa supplicates.

^28. Soames speaks of having (non-extended versions of) properties ‘in an extended sense’ where we speak of having (in a non-extended sense) extended versions of properties. We're not sure what, if anything, this difference amounts to. In any case, we don't think that anything hangs on it.

^29. See, for example, Soames 2014b, 235.

^30. Of course, we are accepting Soames's explanation merely for the sake of argument, since we don't think that propositions really have representational properties. See note 2.

^31. Entertaining is predicating: ‘What is it to entertain a proposition? It is, I suggest, to predicate something of something else.’ (See Soames 2010b, 81; italics in original.) And predicating is representing: ‘To predicate a property being P of an object o is to represent o as being P.’ (See Soames 2010b, 114; italics in original.) It seems that Soames (2014b) has not changed his mind about entertaining. See note 32.

^32. Soames (2014b, 230) accepts this consequence. He says that the types T and entertaining T are identical. See also Speaks 2014, 164.

^33. Predicating is entertaining. See note 31.

^34. Soames 2014b, 235 n. 7; italics in original.

^35. Of course, there's brute and then there's brute. On some views, each proposition is a primitive entity, and it is a brute fact about it that it has the truth-conditional properties that it does. (See McGlone 2012, Merricks ms.) By contrast, the brutal view we prefer invokes one primitive entity and one brute fact about it. (See Caplan and Tillman ms.)

Additional information

Notes on contributors

Ben Caplan

Ben Caplan is a Professor in the Department of Philosophy at Ohio State University. His first publication was in the Canadian Journal of Philosophy.

Chris Tillman

Chris Tillman is an Associate Professor in the Department of Philosophy at the University of Manitoba. His most recent publication is in the Canadian Journal of Philosophy.

Brian McLean

Brian McLean is a graduate student in the Department of Philosophy at Ohio State University. His primary research interests are in metaethics and ethics. His first publication is in the Canadian Journal of Philosophy.

Adam Murray

Adam Murray is a graduate student in the Department of Philosophy at the University of Toronto. His primary research interests are in metaphysics and the philosophy of language. He has recently published papers in the Canadian Journal of Philosophy and in Oxford Studies in Metaphysics.