¹ So what is presented here is a Frege-inspired account of the concept of truth, in terms of cognitive value, and an not an account of Frege's view of truth. For that kind of account, see Richard G. Heck and Robert May, ‘Truth in Frege’ (manuscript).

² From this point on, I use both that-clauses and an italicization convention to make reference to thoughts. When a thought is an attribution of truth to a thought, I use both methods at once. This fixes an expressive inadequacy in the italicization convention and avoids what would otherwise be a confusing repetition of ‘that's. So instead of ‘I smell the scent of violets is true’ which is strictly speaking nonsense, or the confusing but more fundamental and correct ‘that that I smell the scent of violets is true’, I will use ‘that I smell the scent of violets is true’ to refer to the thought that I smell the scent of violets is true. This use of italicization is easily distinguished from italicization for emphasis and italicization to indicate the introduction of a technical notion, both of which I also use. I also do not pause to address niceties about the context-sensitive ‘I’ in ‘I smell the scent of violets’. I use the example despite the ‘I’ to connect with Frege's discussion below.

³ Let me acknowledge and accommodate the objection that concepts do not have cognitive values, but instead are cognitive values. Correct. The cognitive value of a concept is not something that the concept has accidentally, or even just necessarily. The cognitive value of a concept is constitutive of the concept, and individuates the concept. The idea that concepts are cognitive values is encouraged and not precluded by the idea that the cognitive value that a concept has completely determines is nature.

⁴ Kripke's Padreweski (‘A Puzzle About Belief’, in Meaning and Use (ed.) Margalit, A. (Dordrecht: Reidel, 1979)CrossRefGoogle Scholar) also shows that the difference cannot consist in the sameness or difference of linguistic expression.

⁵ The ‘understanding-based’ limits the ambition of my account of sense as cognitive value, but does not vitiate it or render my account empty. It is meant mainly to forestall objections to the account of sense as cognitive value while still allowing the account to be described in enough detail to admit of a substantial and informative application to the concept of truth.

⁶ Kit Fine in Semantic Relationism (2007, Blackwell) might be understood to be suggesting some additional cases. For example, expressing one's thought in language requires one to be able say what one thinks (86); memory requires being able to remember what one was thinking earlier (1-2); tracking an object in continuous observation requires thinking about the object as the same over time (67); reporting the sayings or attitudes of another requires being able to say what the another has said or thought (1; 87).

⁷ Cf. Evans, Gareth ‘Understanding Demonstratives’ in his Collected Papers (1985, Oxford).Google Scholar

⁸ Heck, Richard, in ‘Do Demonstratives Have Senses’ (2002, Philosopher's Imprint)Google Scholar succumbs to this temptation. For discussion see Imogen Dickie and Gurpreet Rattan, ‘Sense, Communication, and Rational Engagement’ (2010, dialectica).

⁹ I think that the real lesson of Burge's ‘Individualism and the Mental’ (1979, Midwest Studies in Philosophy)Google Scholar is to highlight the existence and individuative relevance of this kind of objective norm of testimonial knowledge.

¹⁰ There may however be important differences between that-clauses and propositional descriptions (like ‘the proposition that snow is white’). Propositional descriptions produce what Friederike Moltmann calls an ‘objectivization effect’. There are also (related) problems of intersubstitutability with that-clauses. For helpful and orienting discussion see Moltmann, ‘Propositional Attitudes Without Propositions’ (2003, Synthese) and King, Jeff, ‘Designating Proposititions’ (2002, Philosophical Review)CrossRefGoogle Scholar.

¹¹ This terminology is fairly standard, although the categorization of truth attributions involving demonstrative and named ways of thinking of propositions is less often discussed. For an exception see Soames, Scott, Understanding Truth (1999, Oxford)CrossRefGoogle Scholar.

¹² Cf. Jason Stanley, ‘Truth and Metatheory in Frege’ (1996, Pacific Philosophical Quarterly).

¹³ I do not think that the proper justification for this claim comes from thinking about the semantics of ‘that’-clauses as they occur in natural language. Rather, the main source of justification is top-down: there are ways of immediately extending knowledge that constitutively require the use of ways of thinking of thoughts that themselves constitutively require one to be able to think with the thought being thought about (see below). For related discussion, but that focuses not on truth but on propositional attitude attributions, see Tyler Burge's postscripts to ‘Frege and the Hierarchy’ in Burge'sTruth, Thought, and Reason: Essays On Frege (2005, Oxford)Google Scholar and ‘Belief De Re’ in Burge'sFoundations of Mind (2007, Oxford)Google Scholar; Peacocke's, Christopher chapter ‘Representing Thoughts’ in Peacocke Truly Understood (2008, Oxford)CrossRefGoogle Scholar; and Saul Kripke's ‘Frege's Distinction Between Sense and Reference: Some Exigetical Notes’ (2008, Theoria).

¹⁴ By ‘basic or central’ I mean to flag that the general argument applies not only to norms and rules understood in a foundationalist framework, but also in more holistic ones. I cannot argue this here.

¹⁵ This is not to say that there are no objections to the idea that conceptual analysis is an epistemic resource. This is not the place to take these objections on. But objections to the idea that conceptual analysis is an epistemic resource should be distinguished from objections to the idea that there can be any such thing as confused or incomplete understanding and objections to the idea that there is any distinctive epistemic method of conceptual analysis. If there is such a thing as confused or incomplete understanding and there is a distinct epistemic method of conceptual analysis, then conceptual analysis is an epistemic resource for improved clarity and for achieving the kind and quality of knowledge that clarity affords.

¹⁶ In other work, I explain some philosophically interesting examples in which conceptual analysis provides critical reflective justifications: how an analysis of vague concepts can give critical reflective justification for thinking that Mars was always either dry or not dry (Williamson, The Philosophy of Philosophy, 2007: 43–35Google Scholar); how a semantics for ‘average’ can give critical reflective justification for thinking that the average American can have 2.3 children even though no one has or could have 2.3 children (Kennedy and Stanley, ‘On average’, 2009, Mind).

¹⁷ There is a lot to say about this idea of a standard for standards that I cannot go into (for more than one reason) here. One important issue is that truth is a kind of formal and not substantive standard. By this I mean that truth does not provide epistemic guidance, but instead merely allows for the possibility of inquiry into the most basic standards when the most basic standards are under challenge. Another important issue is flagged in the next note.

¹⁸ There is more to say here about general issues about question-begging. These are issues about the nature of critical reflective thinking in general. This however is not the place to go into these.

¹⁹ See Anil Gupta, ‘A Critique of Deflationism’ (1993, Philosophical Topics) and Collins, John, ‘Declarative Thought, Deflationism and Metarepresentation’ in Griemann, D. and Siegwart, G. (eds), Truth and Speech Acts: Studies in the Philosophy of Language (2007, Routledge)Google Scholar.

²⁰ There is much more on the issues broached in this conclusion in my manuscript ‘Truth Incorporated’.