Disappearing Diamonds: Fitch-Like Results in Bimodal Logic


Augment the propositional language with two modal operators: □ and ■. Define \(\blacklozenge \) to be the dual of ■, i.e. \(\blacklozenge \equiv \neg \blacksquare \neg \). Whenever (X) is of the form φψ, let (X\(^{\blacklozenge } \) ) be \(\varphi \rightarrow \blacklozenge \psi \). (X\(^{\blacklozenge } \) ) can be thought of as the modally qualified counterpart of (X)—for instance, under the metaphysical interpretation of \(\blacklozenge \), where (X) says φ implies ψ, (X\(^{\blacklozenge } \) ) says φ implies possiblyψ. This paper shows that for various interesting instances of (X), fairly weak assumptions suffice for (X\(^{\blacklozenge } \)) to imply (X)—so, the modally qualified principle is as strong as its unqualified counterpart. These results have surprising and interesting implications for issues spanning many areas of philosophy.

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


  1. 1.

    Chandler, H. (1976). Plantinga and the contingently possible. Analysis, 36(2), 106–109.

    Article  Google Scholar 

  2. 2.

    Chellas, B. (1980). Modal logic: an introduction. Cambridge: Cambridge University Press.

    Google Scholar 

  3. 3.

    Church, A. (2009). Referee reports on Fitch’s “A definition of value”. In J. Salerno (Ed.) New essays on the Knowability Paradox (pp. 13–20). Oxford: Oxford University Press.

    Google Scholar 

  4. 4.

    Fitch, F. (1963). A logical analysis of some value concepts. The Journal of Symbolic Logic, 28, 135–142.

    Article  Google Scholar 

  5. 5.

    Jenkins, C.S. (2009). The mystery of the disappearing diamond. In J. Salerno (Ed.) , New essays on the Knowability Paradox (pp. 302–319). Oxford: Oxford University Press.

    Google Scholar 

  6. 6.

    Kvanvig, J. (2006). The Knowability Paradox. Oxford: Oxford University Press.

    Google Scholar 

  7. 7.

    Salmon, N. (1989). The logic of what might have been. The Philosophical Review, 98(1), 3–34.

    Article  Google Scholar 

  8. 8.

    Smith, M. (2007). Ceteris paribus conditionals and comparative normalcy. Journal of Philosophical Logic, 36, 97–121.

    Article  Google Scholar 

  9. 9.

    Williamson, T. (1993). Verificationism and non-distributive knowledge. Australasian Journal of Philosophy, 71(1), 78–86.

    Article  Google Scholar 

Download references


I owe thanks to Jeff Russell, Gabriel Uzquiano, Tim Williamson, and an anonymous reviewer for their helpful comments.

Author information



Corresponding author

Correspondence to Weng Kin San.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.



Theorem 6 shows that given KTK, (4\(^{\blacklozenge }_{\Box }\)) and (5\(^{\blacklozenge }_{\Box }\)) together give rise to an S5 logic for □. We will show that though jointly sufficient, (4\(^{\blacklozenge }_{\Box }\)) and (5\(^{\blacklozenge }_{\Box }\)) are not individually sufficient to give rise to an S5 logic for □.

First, given KTK, (4\(^{\blacklozenge }_{\Box }\)) isn’t strong enough by itself to generate an S5 logic for □:

Theorem 20

The smallest normal extension ofKTKcontaining\({(4^{\blacklozenge }_{\Box })}\)doesnot extendS5K.


As before, nothing depends on □ and ■ being distinct operators. So, it suffices to show that the smallest normal monomodal extension of S4 containing □p →♢□□p does not extend S5. And that is obvious, since □p →♢□□p is already a theorem of S4. □

Now, we show, by a semantic argument, that (5\(^{\blacklozenge }_{\Box }\)) also isn’t strong enough by itself to generate an S5 logic for □. First, some preliminaries: A Kripke frame is a structure \(\mathfrak {F}=<W, R_{\Box }, R_{\blacksquare }>\), where the domain W is a non-empty set, whose elements we shall refer to as ‘worlds’, and R⊆ (W × W) and R⊆ (W × W) are binary relations on W. A Kripke model\(\mathfrak {M}=<\mathfrak {F}, V>\) is a frame with a valuation function V which maps each propositional letter to a set of worlds. If \(\mathfrak {M}=<\mathfrak {F}, V>\), we say that \(\mathfrak {M}\) is based on \(\mathfrak {F}\). A pointed Kripke model\(<\mathfrak {M},w>\) is a model \(\mathfrak {M}\) together with a world w in the domain of \(\mathfrak {M}\) (by the domain of \(\mathfrak {M}\), we mean the domain of the frame on which \(\mathfrak {M}\) is based). Satisfaction in a pointed model is defined:

$$\begin{array}{lll} \mathfrak{M},w\Vdash p & \text{iff} & w\in V(p), \text{for every}~p\in At; \\ \mathfrak{M},w\Vdash \neg\varphi & \text{iff} & \text{not}~\mathfrak{M},w\Vdash \varphi; \\ \mathfrak{M},w\Vdash (\varphi\wedge\psi) & \text{iff} & \mathfrak{M},w\Vdash\varphi~\text{and}~\mathfrak{M},w\Vdash \psi; \\ \mathfrak{M},w\Vdash \Box\varphi & \text{iff} & \text{for every}~v\in W~\text{such that}~<w,v>\in R_{\Box}, \mathfrak{M},v\Vdash\varphi.\\ \mathfrak{M},w\Vdash \blacksquare\varphi & \text{iff} & \text{for every}~v\in W~\text{such that}~<w,v>\in R_{\blacksquare}, \mathfrak{M},v\Vdash\varphi.\\ \end{array} $$

\(\mathfrak {M}\Vdash \varphi \) iff \(\mathfrak {M},w\Vdash \varphi \) for all w in the domain of \(\mathfrak {M}\). And \(\mathfrak {F}\Vdash \varphi \) iff \(\mathfrak {M}\Vdash \varphi \) for every model \(\mathfrak {M}\) based on \(\mathfrak {F}\).

It is easy to show that any frame satisfying the condition in the antecedent of the lemma below validates (5\(^{\blacklozenge }_{\Box }\)):

Lemma 21

If \(\mathfrak {F}\vDash \forall wv (wR_{\Box } v \rightarrow \exists u (wR_{\blacksquare } u\wedge \forall t(uR_{\Box } t\rightarrow tR_{\Box } v)))\) , then \(\mathfrak {F}\Vdash \Diamond p\rightarrow \blacklozenge \Box \Diamond p\) .


Theorem 22

LetLbe the smallest normal extension ofKTKcontaining\({(5^{\blacklozenge }_{\Box })}\).Then,Ldoesn’t extendS4K(and thus also doesn’t extendS5K).

Proof Sketch

Consider the model \(\mathfrak {M}\) below, where the arrows represent R (let R be the universal accessibility relation):


R is reflexive, so \(\mathfrak {M}\) is a KTK-model. Furthermore, checking that the condition in the antecedent of the previous lemma is satisfied is a tedious but straightforward exercise. Thus, the model is an L-model. However, \(\mathfrak {M},w\Vdash \Box p\wedge \neg \Box \Box p\), so \(\mathfrak {M}\) is a countermodel to (4). Thus, L doesn’t extend S4K and thus also doesn’t extend S5K. □

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

San, W.K. Disappearing Diamonds: Fitch-Like Results in Bimodal Logic. J Philos Logic 48, 1003–1016 (2019). https://doi.org/10.1007/s10992-019-09504-0

Download citation


  • Fitch’s Paradox
  • Paradox of Knowability
  • Modal logic
  • Bimodal logic