Skip to main content

Fine on the Possibility of Vagueness

  • Chapter
  • First Online:
Kit Fine on Truthmakers, Relevance, and Non-classical Logic

Part of the book series: Outstanding Contributions to Logic ((OCTR,volume 26))

  • 191 Accesses

Abstract

In his paper ‘The possibility of vagueness’ (Fine in Synthese 194(10):3699–3725, 2017), Kit Fine proposes a new logic of vagueness, CL, that promises to provide both a solution to the sorites paradox and a way to avoid the impossibility result from Fine (Philos Perspect 22(1):111–136, 2008). The present paper presents a challenge to his new theory of vagueness. I argue that the possibility theorem stated in Fine (Synthese 194(10):3699–3725, 2017), as well as his solution to the sorites paradox, fail in certain reasonable extensions of the language of CL. More specifically, I show that if we extend the language with any negation operator that obeys reductio ad absurdum, we can prove a new impossibility result that makes the kind of indeterminacy that Fine takes to be a hallmark of vagueness impossible. I show that such negation operators can be conservatively added to CL and examine some of the philosophical consequences of this result. Moreover, I demonstrate that we can define a particular negation operator that behaves exactly like intuitionistic negation in a natural and unobjectionable propositionally quantified extension of CL. Since intuitionistic negation obeys reductio, the new impossibility result holds in this propositionally quantified extension of CL. In addition, the sorites paradox resurfaces for the new negation.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 99.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Hardcover Book
USD 129.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    Compatibilist models are thus identical to models for the modal logic B. This fact will be exploited in the next section.

  2. 2.

    The reader should consult Fine (2017) for more on the informal interpretation of the semantics.

  3. 3.

    An axiomatic presentation of the logic is given in Appendix 2. Fine (2017) notes without proof that this axiom system is weakly complete for the compatibilist semantics. In unpublished work, Fine proves strong soundness and completeness. The model-theoretic notion of consistency defined above thus coincides with its syntactic counterpart.

  4. 4.

    To see why \(\lnot (p\vee \lnot p)\) is not satisfiable, suppose for a contradiction that \(\lnot (p\vee \lnot p)\) is true at some w. Now either p is true at some v accessible from w, in which case \(p\vee \lnot p\) is true at v and hence \(\lnot (p\vee \lnot p)\) is not true at w; or p is not true at any such v, in which case \(\lnot p\) is true at w, and so \(\lnot (p\vee \lnot p)\) is not true at w, since \(p\vee \lnot p\) is then true at w and w is accessible from itself.

  5. 5.

    The terminology as well as the statement of the Possibility Theorem are adopted almost verbatim from Fine (2017); I have only replaced ‘incompatible’ by ‘inconsistent’. The change is purely terminological and is made in the interest of terminological uniformity.

  6. 6.

    Note that a collective response to \(\psi _{1},\psi _{2},\ldots ,\psi _{n}\) of the form \(\phi _{1}(\psi _{1}),\phi _{2}(\psi _{2}),\ldots \phi _{n}(\psi _{n})\) is sharp if the corresponding collective response \(\phi _{1}(p),\phi _{2}(p),\ldots ,\phi _{n}(p)\) is sharp.

  7. 7.

    A precise definition of the logic CL+ is given in Appendix 1.

  8. 8.

    The notion of a formula \(\phi \) being a (local) B-consequence of a set of formulas \(\varGamma \)—in symbols, \(\varGamma \vDash _{{\textbf {B}}}{\phi }\)—is defined in the standard way: \(\varGamma \vDash _{{\textbf {B}}}{\phi }\) iff for all B-models \(\mathscr {M}_{{\textbf {B}}}\) and for all points u in \(\mathscr {M}_{{\textbf {B}}}\), if \(\mathscr {M}_{{\textbf {B}}},u\vDash {\psi }\) for all \(\psi \in {\varGamma }\), then \(\mathscr {M}_{{\textbf {B}}},u\vDash {\phi }\).

  9. 9.

    Here, \(\sim \) and \(\rightarrow \) denote classical negation and the material conditional, respectively.

  10. 10.

    One is naturally reminded here of modal translations of intuitionistic logic into the modal logic S4. There is a result similar to Theorem 1 for these logics [see, for example, Chagrov and Zakharyaschev (1997, p. 97)]. It is worth noting, however, that the result for intuitionistic logic and S4 does not imply that classical negation can be conservatively added to intuitionistic logic. Indeed, adding classical negation to intuitionistic logic has the effect that intuitionistic negation becomes intersubstitutable with classical negation, thereby making rules like double negation elimination valid for intuitionistic negation. Let \(\lnot ^{c}\) be a classical negation operator governed by the usual classical rules of inference. Let \(\mathscr {L}_{\lnot ^{c}}\) be the language of intuitionistic logic supplemented with \(\lnot ^{c}\). We can then prove in \(\mathscr {L}_{\lnot ^{c}}\) that for every formula \(\phi \), \(\lnot ^{c}\phi \dashv \vdash {\lnot ^{i}\phi }\). Proof: By ex falso quodlibet for \(\lnot ^{c}\), we have \(\lnot ^{c}\phi ,\phi \vdash {\phi }\) and \(\lnot ^{c}\phi ,\phi \vdash {\lnot ^{i}{\phi }}\). So by reductio ad absurdum for \(\lnot ^{i}\) we get \(\lnot ^{c}\phi \vdash {\lnot ^{i}\phi }\). The other direction is proved analogously. The fact that the logic licences the substitution of logical equivalents then yields the stronger conclusion that the two copies of negation are intersubstitutable in every sentential context. See Humberstone (1979, 2011, p. 591ff) and Schechter (2011) for discussion of the question under what conditions we can have logics containing both classical and intuitionistic negation.

  11. 11.

    Incidentally, the same is true for the compatibilist conditional \(\supset \).

  12. 12.

    See, for instance, Humberstone (2011) for discussion.

  13. 13.

    Another criterion Belnap proposes is unique characterization. This criterion, although satisfied by many candidate negation operators, is not as pertinent as conservativity here, since, as mentioned above, compatibilist negation is itself not uniquely characterized. By contrast, classical and intuitionistic negation, for example, are uniquely characterized by their usual introduction- and elimination-rules.

  14. 14.

    Belnap’s own concern did involve the question of whether a connective can be understood in terms of its inferential behavior.

  15. 15.

    See Humberstone (2011, pp. 566 ff) for a discussion of some cases in which a connective might be deemed illegitimate in a logic despite its conservatively extending the logic.

  16. 16.

    Anderson et al. (1992, p. 50ff) give a similar definition of a conditional (with the symbol for relevant entailment in place of \(\supset \)) in the propositionally quantified relevant logic \(\textbf{E}^{\forall \exists p}\). They take such conditionals to express enthymematic arguments, i.e. arguments with a suppressed true premise, and claim that the defined conditional captures the meaning of the intuitionistic conditional. Interestingly, their defined conditional, just like ours, behaves exactly like the intuitionistic conditional and they show that intuitionistic logic can be exactly embedded in \(\textbf{E}^{\forall \exists p}\). We have an exactly analogous result for \(\textbf{CL}\pi \). However, I do not claim that the conditional defined by (DEF\(\rightarrow )\) captures the meaning of the intuitionistic conditional, but only that they obey the same logical principles.

  17. 17.

    It is worth noting that (DEF\(\lnot ^{i}\)) defines classical negation in the logic \(\mathbf {CL\pi +}\) in which the propositional quantifier receives its primary interpretation. A \(\mathbf {CL\pi +}\) model is an ordered quadruple \(\mathscr {M}=\langle {U,\circ ,V,\varPi \rangle }\), where the first three components are as in C(i)–C(iii), and \(\varPi \), the set of propositions, is the powerset of U. Given a \(\mathbf {CL\pi +}\)-model \(\mathscr {M}\), a proposition \(P\in \varPi \) and a propositional variable p, \(\mathscr {M}[P/p]\) is the model just like \(\mathscr {M}\) except that it assigns the proposition P to the propositional variable p. The extension \(\llbracket \cdot \rrbracket \) of the valuation function V from the propositional variables to \(\varPi \) is determined by T(i)–T(vi) together with the following clause for the propositional quantifier:

    • T\(^\prime \)(vii) \(\llbracket \exists p\phi \rrbracket ^{\mathscr {M}}=\bigcup \lbrace \llbracket \phi \rrbracket ^{\mathscr {M}[P/p]}: P\in \varPi \rbrace \)

    We may call the interpretation of the propositional quantifier given by a \(\mathbf {CL\pi +}\)-model its primary interpretation. We say that \(\mathscr {M}\) validates \(\phi \) (\(\mathscr {M}\vDash \phi \)) just in case \(\llbracket \phi \rrbracket ^{\mathscr {M}}=U\). \(\phi \) is \(\mathbf {CL\pi +}\)-valid (\(\vDash _{\mathbf {CL\pi +}}\phi \)) just in case it is validated by every \(\mathbf {CL\pi +}\)-model. The logic \(\mathbf {CL\pi +}\) is the set of \(\mathbf {CL\pi +}\)-valid formulas (\(\lbrace \phi :\;\vDash _{\mathbf {CL\pi +}}\phi \rbrace \)). The logic \(\mathbf {CL\pi +}\) is thus the logic corresponding to the primary interpretation of the propositional quantifier in \(\textbf{CL}\).

  18. 18.

    This last step follows from the fact that \(p_{0}\wedge \lnot ^{i}p_{0}\) entails \(\lnot p_{0}\).

  19. 19.

    An anonymous referee has suggested that the existence of an operator like \(\lnot ^{i}\) is only problematic if it adequately regiments negation in English. I disagree, and the arguments given above are deliberately neutral on the question of which (if any) of the operators in question adequately regiments negation in English. Even if \(\lnot \) is in fact a better candidate than \(\lnot ^{i}\) for regimenting negation in English—which is not obvious—it is unclear why this alone would make the premises in a sorites with \(\lnot ^{i}\) instead of \(\lnot \) any less plausible.

  20. 20.

    Alternatively, we could add some set of axioms and rules that uniquely characterize intuitionistic negation. The point has even more force if we already have an independent understanding of intuitionistic negation.

  21. 21.

    An analogous point applies if one wanted to argue against the significance of the Impossibility Theorem in \(\textbf{CL}\pi \) by appeal to the quantificational structure of \(\lnot ^{i}\).

  22. 22.

    This is easily verified by using the semantics given in the main text.

  23. 23.

    The proof strategy below is somewhat similar to the strategy of a similar proof given in Anderson et al. (1992, pp. 57ff.). They show that the propositionally quantified relevance logic \(\textbf{E}^{\forall \exists p}_{+}\) exactly embeds intuitionistic logic under virtually the same translation as the one given here (with \(\supset \) replaced by the symbol for relevant entailment).

  24. 24.

    The translations above are to be performed in such a way as to avoid problems with free variables in subformulas becoming bound after translation. In other words, the quantified variable p in the above translations should always be such that it doesn’t occur free in either \(\phi '\) or \(\psi '\).

References

  • Anderson, A. R., Belnap, N. D., & Dunn, J. M. (1992). Entailment: The logic of relevance and necessity (Vol. II). Princeton University Press.

    Google Scholar 

  • Belnap, N. (1962). Tonk, plonk and plink. Analysis, 22(6), 130–134.

    Article  Google Scholar 

  • Chagrov, A., & Zakharyaschev, M. (1997). Modal logic. Oxford Logic Guides (Vol. 35). Clarendon Press.

    Google Scholar 

  • Fine, K. (1975). Vagueness, truth and logic. Synthese, 30(3–4), 265–300.

    Article  Google Scholar 

  • Fine, K. (2008). The impossibility of vagueness. Philosophical Perspectives, 22(1), 111–136.

    Article  Google Scholar 

  • Fine, K. (2017). The possibility of vagueness. Synthese, 194(10), 3699–3725.

    Google Scholar 

  • Humberstone, I. L. (1979). Interval semantics for tense logic: Some remarks. Journal of Philosophical Logic, 8(1), 171–196.

    Article  Google Scholar 

  • Humberstone, L. (2011). The connectives. MIT Press.

    Google Scholar 

  • Kremer, P. (1997). On the complexity of propositional quantification in intuitionistic logic. Journal of Symbolic Logic, 62(2), 529–544.

    Article  Google Scholar 

  • Schechter, J. (2011). Juxtaposition: A new way to combine logics. Review of Symbolic Logic, 4(4), 560–606.

    Article  Google Scholar 

Download references

Acknowledgements

Thanks to Hartry Field, Harvey Lederman, Marko Malink, Stephen Schiffer, as well as two anonymous referees and the editors of this volume for their helpful suggestions and comments on an earlier draft. Special thanks to Kit Fine and Cian Dorr for many helpful discussions of this material and comments on earlier versions of this chapter.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Andreas Ditter .

Editor information

Editors and Affiliations

Appendices

Appendix 1

This appendix establishes the Impossibility Theorem from Sect. 31.2. Let \(\mathscr {L}_{{\textbf {CL+}}}\) be the language of \(\textbf{CL}\) supplemented with a unary negation operator \(\sim \). A consequence relation, \(\vdash \), for a language \(\mathscr {L}\), is a relation between sets of \(\mathscr {L}\)-formulas and \(\mathscr {L}\)-formulas that is substitution invariant and closed under the usual structural rules of reflexivity, monotonicity and cut. When \(\vdash \) is a consequence relation, we write \(\varGamma \vdash \phi \) for \(\langle \varGamma ,\phi \rangle \in \;\vdash \), and \(\phi _{1},\ldots ,\phi _{n}\vdash \psi \) for \(\lbrace \phi _{1},\ldots ,\phi _{n}\rbrace \vdash \psi \). Let \(\vdash _{\textbf{CL}}:=\lbrace \langle \varGamma ,\phi \rangle :\varGamma \vDash _{\textbf{CL}}\phi , \varGamma \subseteq \mathscr {L}_{\textbf{CL}}, \phi \in \mathscr {L}_{\textbf{CL}}\rbrace \). We define \(\vdash _{\mathbf {CL+}}\) to be the smallest consequence relation for \(\mathscr {L}_{{\textbf {CL+}}}\) that contains \(\vdash _{\textbf{CL}}\) and satisfies the following condition:

  • (RAA) If \(\varGamma ,\phi \vdash _{\mathbf {CL+}}\psi \) and \(\varGamma ,\phi \vdash _{\mathbf {CL+}}{\sim }\psi \) for some \(\psi \), then \(\varGamma \vdash _{\mathbf {CL+}}{\sim }\phi \);

Thus, \(\vdash _{\mathbf {CL+}}\) is closed under reductio ad absurdum for \(\sim \). We define the logic \(\mathbf {CL+}\) to be the consequence relation \(\vdash _{\mathbf {CL+}}\). Note that the fact that \(\mathbf {CL+}\) contains \(\textbf{CL}\) and is substitution invariant guarantees that every \(\mathscr {L}_{\mathbf {CL+}}\)-substitution instance of a valid argument of \(\textbf{CL}\) is in \(\mathbf {CL+}\). This entails, for example, that \(p,\lnot p\vdash _{\mathbf {CL+}}{\sim }p\), this being a \(\mathscr {L}_{\mathbf {CL+}}\)-substitution instance of ex falso quodlibet, which is a valid rule of \(\textbf{CL}\).Footnote 22 I will henceforth write \(\vdash \) instead of \(\vdash _{\mathbf {CL+}}\) whenever no confusion is likely to arise.

A set of formulas \(\varGamma \) of \(\mathscr {L}_{{\textbf {CL+}}}\) is inconsistent just in case \(\varGamma \vdash \phi \) and \(\varGamma \vdash {\sim }\phi \), for some formula \(\phi \). \(\varGamma \) is consistent just in case it is not inconsistent.

Lemma 2

Let \(\varGamma \) be a consistent set of formulas. Then if \(\varGamma \cup \lbrace \phi \rbrace \) is inconsistent, \(\varGamma \cup \lbrace {\sim }\phi \rbrace \) is consistent.

Proof

Suppose that \(\varGamma \cup \lbrace \phi \rbrace \) is inconsistent. Then \(\varGamma ,\phi \vdash \psi \wedge {\sim }\psi \), for some \(\psi \). By (RAA), \(\varGamma \vdash {\sim }\phi \). Now suppose for reductio that \(\varGamma \cup \lbrace {\sim }\phi \rbrace \) is inconsistent. Then \(\varGamma ,{\sim }\phi \vdash \psi \wedge {\sim }\psi \), for some \(\psi \). Hence, by (RAA), \(\varGamma \vdash {\sim }{\sim }\phi \). Hence, \(\varGamma \) is inconsistent, contrary to our hypothesis. Thus, \(\varGamma \cup \lbrace {\sim }\phi \rbrace \) is consistent.

Lemma 3

For all formulas of \(\mathscr {L}_{\mathbf {CL+}}\), \(\lnot \phi \vdash {\sim }\phi \).

Proof

By reflexivity and monotonicity, \(\phi ,\lnot \phi \vdash \phi \). By ex falso quodlibet for \(\lnot \), \(\phi ,\lnot \phi \vdash {\sim }\phi \). Now by (RAA), \(\lnot \phi \vdash {\sim }\phi \).

Note that, as a consequence of Theorem 1, the converse entailment does not hold. The definitions of the notions of an extremal response and a sharp response are as in the main text, although they should now be understood with respect to the language \(\mathscr {L}_{\mathbf {CL+}}\), and the notion of inconsistency appealed to in condition (ii) of the definition of a sharp response is now to be understood as \(\mathbf {CL+}\)-inconsistency.

We can now state and prove the Impossibility Theorem from the main text.

Theorem 2

(Impossibility) Let \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\), \(n\ge 0\), be any formulas of \(\mathscr {L}_{\mathbf {CL+}}\). Then there is no set of formulas \(\varDelta _{0}\) which is consistent with the extremal response \(\lbrace \phi _{0},\lnot \phi _{n+1}\rbrace \) and yet inconsistent with every sharp response to \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\).

Proof

The proof is similar to the proof of the Impossibility Theorem in Fine (2008). Let \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\) be any set of formulas and let \(\varDelta _{0}\) be a set of formulas consistent with the extremal response \(\lbrace \phi _{0},\lnot \phi _{n+1}\rbrace \). We show that \(\varDelta _{0}\) is consistent with a sharp response to \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\) by extending \(\varDelta _{0}\). Let \(\varDelta _{1}=\varDelta _{0}\cup \lbrace \phi _{0},{\sim }\phi _{n+1}\rbrace \). For \(k=1,2,\ldots ,n\), we define:

$$\varDelta _{k+1}={\left\{ \begin{array}{ll} \varDelta _{k}\cup \lbrace \phi _{k}\rbrace , &{}\text {if consistent} \\ \varDelta _{k}\cup \lbrace {\sim }\phi _{k}\rbrace , &{} \text {otherwise} \end{array}\right. }$$

Note that by Lemma 2, \(\varDelta _{k}\) is consistent for \(k=1,\ldots ,n+1\). The consistency of \(\varDelta _{1}\) follows from Lemma 3. For suppose \(\varDelta _{1}\) is inconsistent. Then (i) \(\varDelta _{0},\phi _{0},{\sim }\phi _{n+1}\vdash \psi \wedge {\sim }\psi \), for some \(\psi \). From this, by (RAA) and monotonicity, we obtain: (ii) \(\varDelta _{0},\phi _{0},\lnot \phi _{n+1}\vdash {\sim \sim }\phi _{n+1}\). By Lemma 3 and monotonicity: (iii) \(\varDelta _{0},\phi _{0},\lnot \phi _{n+1}\vdash {\sim }\phi _{n+1}\), in contradiction to the consistency of \(\varDelta _{0}\cup \lbrace \phi _{0},\lnot \phi _{n+1}\rbrace \). Note further that for every \(k\le n+1\), either \(\phi _{k}\in \varDelta _{n+1}\) or \({\sim }\phi _{k}\in \varDelta _{n+1}\).

We can now define a collective response \(\psi _{0}(p), \psi _{1}(p),\ldots ,\psi _{n+1}(p)\) and a corresponding collective response \(\psi _{0}(\phi _{0}), \psi _{1}(\phi _{1}),\ldots ,\psi _{n+1}(\phi _{n+1})\) to \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\). Where \(k=0,1,\ldots ,n+1\):

$$\psi _{k}(p)={\left\{ \begin{array}{ll} p, &{} \text{ if } \phi _{k}\in \varDelta _{n+1}\\ {\sim }p, &{} \text{ if } {\sim }\phi _{k}\in \varDelta _{n+1} \end{array}\right. }$$

We now get:

  1. (1)

    \(\varDelta _{n+1}\vdash \psi _{k}(\phi _{k})\), for \(k=0,1,\ldots ,n+1\).

The claim follows trivially from the reflexivity and monotonicity of \(\vdash \). Since \(\varDelta _{n+1}\) is consistent it follows from (1) that \(\varDelta _{n+1}\) is consistent with the response \(\psi _{0}(\phi _{0}), \psi _{1}(\phi _{1}),\ldots ,\psi _{n+1}(\phi _{n+1})\) to \(\phi _{0},\phi _{1},\ldots ,\phi _{n+1}\). Hence, \(\varDelta _{0}\) is also consistent with this response, since it is a subset of \(\varDelta _{n+1}\).

It remains to show that the response \(\psi _{0}(p), \psi _{1}(p),\ldots ,\psi _{n+1}(p)\) is sharp. The first condition for being a sharp response is satisfied, since \(\psi _{0}(p)=p\ne {\sim }p=\psi _{n+1}(p)\). The second condition says that any two individual responses are either the same or inconsistent. But every response is either of the form p or of the form \({\sim }p\). So the second condition is also satisfied. This concludes the proof.

Appendix 2

This appendix establishes the Embedding Theorem from Sect. 31.3. The first step is to show that with the conditional \(\rightarrow \) defined as \((\phi \rightarrow \psi )=^{df}\exists p(p\wedge ((\phi \wedge p)\supset \psi ))\), where p is the first variable not occurring free in either \(\phi \) or \(\psi \) and the negation \(\lnot ^{i}\) defined as \(\lnot ^{i}\phi =^{df}\phi \rightarrow \bot \), we can derive all the axioms and rules of \(\textbf{IL}\) in \(\mathbf {CL\pi }\). In what follows, we will often use the convention that \(\wedge \) and \(\vee \) are given precedence over \(\supset \) and \(\rightarrow \) so that, for example, \(\phi \wedge \psi \supset \xi \) abbreviates \((\phi \wedge \psi )\supset \xi \) and \(\phi \rightarrow \psi \vee \xi \) abbreviates \(\phi \rightarrow (\psi \vee \xi )\). Outer brackets of conjunctions and disjunctions are thus often omitted.

We will use the axiomatization of \(\textbf{CL}\) from Fine (2017, p. 3723f). Note that compatibilist negation \(\lnot \) is a defined connective in this axiomatization: \(\lnot \phi =^{df}\phi \supset \bot \).

figure b

In what follows, \(\vdash \) denotes derivability in \(\mathbf {CL\pi }\), which is obtained by adding the following axioms and rules for the propositional quantifier to \(\textbf{CL}\):

figure c

We need to show that all of the following are provable in \(\textbf{CL}\pi \):

figure d

We state without proof the following two elementary lemmas:

Lemma 4

\(\vdash \phi \supset \phi \)

Lemma 5

If \(\vdash \phi \), then \(\vdash \psi \supset \phi \)

The use of associativity and commutativity of conjunction and disjunction will be tacit in the proofs to follow.Footnote 23 The proofs of these principles themselves are routine. The following central lemma will be useful:

Lemma 6

\(\vdash (\phi \supset \psi )\supset (\phi \rightarrow \psi )\)

Proof

  1. 1.

    \((\phi \supset \psi )\supset (\phi \supset \psi )\)           [Lemma 4]

  2. 2.

    \(\phi \wedge (\phi \supset \psi )\supset \psi \)           [A1]

  3. 3.

    \((\phi \supset \psi )\supset (\phi \wedge (\phi \supset \psi )\supset \psi )\)           [2., Lemma 5]

  4. 4.

    \(((\phi \supset \psi )\supset (\phi \supset \psi ))\wedge ((\phi \supset \psi )\supset (\phi \wedge (\phi \supset \psi )\supset \psi ))\)           [1., 3., R2]

  5. 5.

    \((\phi \supset \psi )\supset (\phi \supset \psi )\wedge (\phi \wedge (\phi \supset \psi )\supset \psi )\)           [4., A8, R1]

  6. 6.

    \((\phi \supset \psi )\wedge (\phi \wedge (\phi \supset \psi )\supset \psi )\supset \exists p(p\wedge (\phi \wedge p\supset \psi ))\)           [\(\exists 1\)]

  7. 7.

    \((\phi \supset \psi )\supset \exists p(p\wedge (\phi \wedge p\supset \psi ))\)           [5., 6., R3]

  8. 8.

    \((\phi \supset \psi )\supset (\phi \rightarrow \psi )\)           [7., Def.]

From this and the corresponding axioms of CL we immediately get IL4, IL5, IL6 and IL7. To prove MP\(\rightarrow \), we first show:

Lemma 7

\(\vdash \phi \wedge (\phi \rightarrow \psi )\supset \psi \)

Proof

Let p be a variable not free in \(\phi \) or \(\psi \).

  1. 1.

    \((\phi \wedge p)\wedge (\phi \wedge p\supset \psi )\supset \psi \)           [A1]

  2. 2.

    \(\exists p((\phi \wedge p)\wedge (\phi \wedge p\supset \psi ))\supset \psi \)           [1. \(\exists 2\)]

  3. 3.

    \(\phi \wedge \exists p(p\wedge (\phi \wedge p\supset \psi ))\supset \exists p((\phi \wedge p)\wedge (\phi \wedge p\supset \psi ))\)           [\(\exists \wedge \)]

  4. 4.

    \(\phi \wedge (\phi \rightarrow \psi )\supset \psi \)           [3., 2., R3, Def.]

From Lemma 7 we get MP\(\rightarrow \) from R2 and R1:

Lemma 8

(MP\(\rightarrow \)) If \(\vdash \phi \) and \(\vdash \phi \rightarrow \psi \), then \(\vdash \psi \).

Next, we prove IL1 and IL3. We first state without proof the following auxiliary lemma:

Lemma 9

If \(\vdash \phi \) and \(\vdash \psi \wedge \phi \supset \xi \), then \(\vdash \psi \supset \xi \).

Lemma 10

(IL1) \(\vdash \phi \rightarrow (\psi \rightarrow \phi )\)

Proof

  1. 1.

    \(\phi \wedge (\phi \wedge \psi \supset \phi )\supset (\psi \rightarrow \phi )\)           [\(\exists 1\), Def]

  2. 2.

    \(\phi \supset (\psi \rightarrow \phi )\)           [1., A6, Lemma 9]

  3. 3.

    \((\phi \supset (\psi \rightarrow \phi ))\supset (\phi \rightarrow (\psi \rightarrow \phi ))\)           [Lemma 6]

  4. 4.

    \(\phi \rightarrow (\psi \rightarrow \phi )\)           [2., 3., R1]

Lemma 11

(IL3) \(\vdash \phi \rightarrow (\psi \rightarrow \phi \wedge \psi )\)

Proof

  1. 1.

    \(\phi \wedge (\phi \wedge \psi \supset \phi \wedge \psi )\supset (\psi \rightarrow \phi \wedge \psi )\)          [\(\exists 1\), Def.]

  2. 2.

    \(\phi \supset (\psi \rightarrow \phi \wedge \psi )\)           [1., Lemma 9]

  3. 3.

    \(\phi \rightarrow (\psi \rightarrow \phi \wedge \psi )\)           [2., Lemma 6, R1]

For IL2 and IL8 we need the following

Lemma 12

(Exp.) \(\vdash ((\phi \wedge \psi )\rightarrow \xi )\supset (\phi \rightarrow (\psi \rightarrow \xi ))\)

Proof

Let p be a variable not free in \(\phi \), \(\psi \) or \(\xi \).

  1. 1.

    \(\exists p(p\wedge ((\phi \wedge \psi )\wedge p\supset \xi ))\wedge \phi \supset \exists p((p\wedge \phi )\wedge (p\wedge (\phi \wedge \psi )\supset \xi ))\)           [\(\exists \wedge \)]

  2. 2.

    \(((p\wedge \phi )\wedge ((p\wedge \phi )\wedge \psi \supset \xi ))\supset \exists q(q\wedge (q\wedge \psi \supset \xi ))\)           [\(\exists 1\)]

  3. 3.

    \(\exists p((p\wedge \phi )\wedge ((p\wedge \phi )\wedge \psi \supset \xi ))\supset \exists q(q\wedge (q\wedge \psi \supset \xi ))\)           [2., \(\exists 2\)]

  4. 4.

    \((((\phi \wedge \psi )\rightarrow \xi )\wedge \phi )\supset (\psi \rightarrow \xi )\)           [1., 3., R3]

  5. 5.

    \((((\phi \wedge \psi )\rightarrow \xi )\wedge ((((\phi \wedge \psi )\rightarrow \xi )\wedge \phi )\supset (\psi \rightarrow \xi )))\supset (\phi \rightarrow (\psi \rightarrow \xi ))\)           [\(\exists 1\), Def.]

  6. 6.

    \(((\phi \wedge \psi )\rightarrow \xi )\supset (\phi \rightarrow (\psi \rightarrow \xi ))\)           [4., 5., Lemma 9]

Lemma 13

(IL2) \(\vdash (\phi \rightarrow \psi )\rightarrow ((\phi \rightarrow (\psi \rightarrow \xi ))\rightarrow (\phi \rightarrow \xi ))\)

Proof

  1. 1.

    \(\phi \wedge (\phi \rightarrow \psi )\supset \psi \)           [Lemma 7]

  2. 2.

    \(\phi \wedge (\phi \rightarrow (\psi \rightarrow \xi ))\supset (\psi \rightarrow \xi )\)           [Lemma 7]

  3. 3.

    \(\psi \wedge (\psi \rightarrow \xi )\supset \xi \)           [Lemma 7]

  4. 4.

    \((\phi \wedge (\phi \rightarrow \psi ))\wedge (\phi \rightarrow (\psi \rightarrow \xi ))\supset \xi \)           [A6-8, 1.-3., R3]

  5. 5.

    \((\phi \rightarrow \psi )\rightarrow ((\phi \rightarrow (\psi \rightarrow \xi ))\rightarrow (\phi \rightarrow \xi ))\)           [4., Lemma 6, R1, Exp. (twice)]

For the proof of IL8 we require another auxiliary lemma.

Lemma 14

\(\vdash (\xi \wedge \chi )\wedge (\phi \vee \psi )\supset (\phi \wedge \chi )\vee (\psi \wedge \xi )\)

Proof

  1. 1.

    \(\chi \wedge (\phi \vee \psi )\supset (\chi \wedge \phi )\vee (\chi \wedge \psi )\)          [A12]

  2. 2.

    \(\xi \wedge (\chi \wedge (\phi \vee \psi ) )\supset \xi \wedge ((\chi \wedge \phi )\vee (\chi \wedge \psi ))\)           [1., A6-8, R3]

  3. 3.

    \(\xi \wedge ((\chi \wedge \phi )\vee (\chi \wedge \psi ))\supset (\xi \wedge (\chi \wedge \phi ))\vee (\xi \wedge (\chi \wedge \psi ))\)           [A12]

  4. 4.

    \(\xi \wedge (\chi \wedge \phi )\supset (\phi \wedge \chi )\vee (\psi \wedge \xi )\)           [A7, A9, R3]

  5. 5.

    \(\xi \wedge (\chi \wedge \psi )\supset (\phi \wedge \chi )\vee (\psi \wedge \xi )\)           [A7, A8, R3]

  6. 6.

    \((\xi \wedge (\chi \wedge \phi ))\vee (\xi \wedge (\chi \wedge \psi ))\supset (\phi \wedge \chi )\vee (\psi \wedge \xi )\)           [4., 5., A11, R2, R1]

  7. 7.

    \(\xi \wedge (\chi \wedge (\phi \vee \psi ) )\supset (\phi \wedge \chi )\vee (\psi \wedge \xi )\)           [2., 3., 6., R3]

Lemma 15

(IL8) \(\vdash (\phi \rightarrow \tau )\rightarrow ((\psi \rightarrow \tau )\rightarrow (\phi \vee \psi \rightarrow \tau ))\)

Proof

  1. 1.

    \((\phi \wedge (\phi \rightarrow \tau )\supset \tau )\wedge (\psi \wedge (\psi \rightarrow \tau )\supset \tau )\supset \) \(((\phi \wedge (\phi \rightarrow \tau ))\vee (\psi \wedge (\psi \rightarrow \tau ))\supset \tau )\)           [A11]

  2. 2.

    \((\phi \wedge (\phi \rightarrow \tau ))\vee (\psi \wedge (\psi \rightarrow \tau ))\supset \tau \)           [Lemma 7, 1., R1, R2]

  3. 3.

    \(((\phi \rightarrow \tau )\wedge (\psi \rightarrow \tau ))\wedge (\phi \vee \psi )\supset \) \((\phi \wedge (\phi \rightarrow \tau ))\vee (\psi \wedge (\psi \rightarrow \tau ))\)           [Lemma 14]

  4. 4.

    \(((\phi \rightarrow \tau )\wedge (\psi \rightarrow \tau ))\wedge (\phi \vee \psi )\supset \tau \)           [2., 3., R3]

  5. 5.

    \(((\phi \rightarrow \tau )\wedge (\psi \rightarrow \tau ))\wedge (\phi \vee \psi )\rightarrow \tau \)           [4., Lemma 6, R1]

  6. 6.

    \((\phi \rightarrow \tau )\rightarrow ((\psi \rightarrow \tau )\rightarrow (\phi \vee \psi \rightarrow \tau ))\)           [5., Exp. (twice)]

It only remains to show the two negation axioms IL9 and IL10. But IL9 is just an instance of IL2, with \(\bot \) for \(\xi \). IL10 may be proved as follows:

Lemma 16

(IL10) \(\vdash (\lnot ^{i}\phi )\rightarrow (\phi \rightarrow \psi )\)

Proof

  1. 1.

    \((\phi \rightarrow \bot )\wedge \phi \supset \bot \)           [Lemma 7]

  2. 2.

    \(\bot \supset \psi \)           [A13]

  3. 3.

    \((\phi \rightarrow \bot )\wedge \phi \supset \psi \)           [1., 2., R3]

  4. 4.

    \((\phi \rightarrow \bot )\wedge \phi \rightarrow \psi \)           [3., Lemma 6, R1]

  5. 5.

    \((\lnot ^{i}\phi )\rightarrow (\phi \rightarrow \psi )\)           [4., Exp., R1, Def.]

To state the first part of our central result, we define a translation \('\) from \(\mathscr {L}_{\textbf{IL}}\) into \(\mathscr {L}_{\textbf{CL}\pi }\) by taking \((\phi \rightarrow ^{i}\psi )'=\exists p(p\wedge ((\phi '\wedge p)\supset \psi '))\) and \((\lnot ^{i}\phi )'=\exists p(p\wedge ((\phi '\wedge p)\supset \bot ))\), and translating conjunctions, disjunctions and atomic formulas homophonically.Footnote 24

Corollary 2

If \(\vdash _{\textbf{IL}}\phi \), then \(\vdash _{\mathbf {CL\pi }}\phi '\).

In fact, what we have proven is significantly stronger than Corollary 2. For we have shown that the intuitionistic axioms and rules hold unrestrictedly for all formulas of \(\mathscr {L}_{\mathbf {CL\pi }}\) (with \(\rightarrow \) and \(\lnot ^{i}\) understood as defined connectives), and not only for translations of formulas of \(\mathscr {L}_{\textbf{IL}}\).

Next we show that the converse of Corollary 2 holds as well, so that the translation of a formula of \(\mathscr {L}_{\textbf{IL}}\) is a theorem of \(\mathbf {CL\pi }\) only if its inverse image is an intuitionistic theorem. This is what justifies saying that the embedding of \(\textbf{IL}\) into \(\mathbf {CL\pi }\) is exact.

Let \(\mathbf {IL\pi }\) be the propositionally quantified intuitionistic logic whose language is the same as \(\mathscr {L}_{\textbf{IL}}\) with the addition of the propositional quantifier \(\exists \), and which results from \(\textbf{IL}\) by adding the following axioms and rules for the quantifiers:

figure e

We will make use of the following auxiliary translation from \(\mathscr {L}_{\mathbf {CL\pi }}\) to \(\mathscr {L}_{\mathbf {IL\pi }}\). Define h to be the function from \(\mathscr {L}_{\mathbf {CL\pi }}\) to \(\mathscr {L}_{\mathbf {IL\pi }}\) that replaces every \(\supset \) by \(\rightarrow ^{i}\), and leaving everything else the same (i.e. translating everything else homophonically). For example, the formula \(p\supset q\) gets translated into \(p\rightarrow ^{i}q\).

We prove the following three claims, for \(\phi \) in \(\mathscr {L}_{\textbf{IL}}\):

figure f

Lemma 17

If \(\vdash _{\mathbf {CL\pi }}\phi \), then \(\vdash _{\mathbf {IL\pi }}h(\phi )\)

Proof

The result follows immediately from the observation that every theorem of CL is a theorem of IL [Theorem 2 in Fine (2017)], the fact that \(\exists 1\) and \(\exists 2\) are common to both logics (with \(\rightarrow ^{i}\) in place of \(\supset \)) and the fact that the intuitionistic analog of \(\exists \wedge \) is provable in \(\textbf{IL}\pi \).

Lemma 18

For \(\phi \) in \(\mathscr {L}_{\textbf{IL}}\), if \(\vdash _{\mathbf {IL\pi }}h(\phi ')\), then \(\vdash _{\mathbf {IL\pi }}\phi \).

Proof

We observe that the following equivalences are easily provable in \(\mathbf {IL\pi }\):

  1. (i)

    \(\vdash _{\mathbf {IL\pi }}(\phi \rightarrow ^{i}\psi )\leftrightarrow ^{i}\exists p(p\wedge ((p\wedge \phi )\rightarrow ^{i}\psi ))\)

  2. (ii)

    \(\vdash _{\mathbf {IL\pi }}\lnot ^{i}\phi \leftrightarrow ^{i}\exists p(p\wedge ((p\wedge \phi )\rightarrow ^{i}\bot ))\)

Now suppose that \(\phi \) is in \(\mathscr {L}_{\textbf{IL}}\) and \(\vdash _{\mathbf {IL\pi }}h(\phi ')\). We obtain a proof of \(\phi \) from a proof of \(h(\phi ')\) by noting that \(\phi \) is logically equivalent to its translation \(h(\phi ')\) by virtue of (i) and (ii) and the fact that substitution of logical equivalents preserves theoremhood in \(\mathbf {IL\pi }\).

Finally, we show that \(\mathbf {IL\pi }\) is a conservative extension of \(\textbf{IL}\).

Lemma 19

For \(\phi \in \mathscr {L}_{\textbf{IL}}\), if \(\vdash _{\mathbf {IL\pi }}\phi \), then \(\vdash _{\textbf{IL}}\phi \).

Proof

We prove the contrapositive. Suppose \(\nvdash _{\textbf{IL}}\phi \). Then by the completeness of the Kripke semantics for IL there is a Kripke model \(\mathscr {M}=\langle W,\le ,V\rangle \) for IL in which \(\phi \) fails to hold: \(\mathscr {M}\nvDash \phi \). But \(\mathscr {M}\) can also be regarded as an \(\textbf{IL}\pi \)-model in the familiar way; we only need to extend the truth definition by a clause for the existential quantifier [see, for instance, Kremer (1997)]. So we still have \(\mathscr {M}\nvDash \phi \). Hence by the soundness of this semantics for \(\mathbf {IL\pi }\), we obtain \(\nvdash _{\mathbf {IL\pi }}\phi \).

Corollary 3

For \(\phi \in \mathscr {L}_{\textbf{IL}}\), if \(\vdash _{\mathbf {CL\pi }}\phi '\), then \(\vdash _{\textbf{IL}}\phi \).

Corollaries 2 and 3 imply

Theorem 3

(Embedding) \(\mathbf {CL\pi }\) exactly embeds \(\textbf{IL}\) under translation.

Rights and permissions

Reprints and permissions

Copyright information

© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this chapter

Check for updates. Verify currency and authenticity via CrossMark

Cite this chapter

Ditter, A. (2023). Fine on the Possibility of Vagueness. In: Faroldi, F.L.G., Van De Putte, F. (eds) Kit Fine on Truthmakers, Relevance, and Non-classical Logic. Outstanding Contributions to Logic, vol 26. Springer, Cham. https://doi.org/10.1007/978-3-031-29415-0_31

Download citation

Publish with us

Policies and ethics