Reducing Contrastive Knowledge


According to one form of epistemic contrastivism, due to Jonathan Schaffer, knowledge is not a binary relation between an agent and a proposition, but a ternary relation between an agent, a proposition, and a context-basing question. In a slogan: to know is to know the answer to a question. I argue, first, that Schaffer-style epistemic contrastivism can be semantically represented in inquisitive dynamic epistemic logic, a recent implementation of inquisitive semantics in the framework of dynamic epistemic logic; second, that within inquisitive dynamic epistemic logic, the contrastive ternary knowledge operator is reducible to the standard binary one. The reduction shows, I argue, that Schaffer’s argument in favor of contrastivism is compatible with a binary picture of knowledge. This undercuts the force of the argument in favor of contrastivism.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4


  1. 1.

    Schaffer usually presents contrastivism as the view that knowledge is a ternary relation between an agent, a proposition and a contrast proposition q (i.e. (K(apq))), but he takes it for granted that the two are interchangeable:

    contrastive knowledge is equivalent to question-relative knowledge: K(apq) is equivalent to K(apQ), where Q is the question ?\(\{p, q\}\) [i.e. the question whether p or q]. (Schaffer 2007a, p. 241)

  2. 2.

    There are other contrastive views, notably Dretske (1972) and Karjalainen and Morton (2003). Here I focus on Schaffer’s views.

  3. 3.

    See Schaffer (2004, 2005, 2007b, 2007a, 2008), Schaffer and Knobe (2012) and Schaffer and Szabó (2014).

  4. 4.

    See, e.g. Rysiew (2011) and Stanley (2011).

  5. 5.

    For other formal representations of Schaffer’s contrastivism, see Aloni et al. (2013), Schaffer and Szabó (2014), Hawke (2016) and Groenendijk (2007), the latter being the closest to the one I present here. None of these focus on the question of the reduction of contrastive knowledge. I leave the question of reduction of contrastive knowledge with respect to other formulations out of this paper, due to length constraints. I explain why I choose inquisitive semantics to formulate contrastivism in the end of Sect. 1.2.

  6. 6.

    It is fair to doubt the robustness of these intuitions. The issue is tested empirically in (Schaffer and Knobe 2012).

  7. 7.

    For some alternative explanations to the problem of convergent knowledge, see Aloni and Egre (2008), Aloni et al. (2013) and Gerken (2013).

  8. 8.

    For a broad overview of the framework of inquisitive semantics and its various motivations and applications, see Ciardelli et al. (2018). For a general overview of a semantic analysis of questions, see Cross and Roelofsen (2016).

  9. 9.

    For another logic that combines knowledge and questions, see van Benthem and Minică (2009).

  10. 10.

    Using the \(\sigma \) function is equivalent to the (more familiar) use of an epistemic accessibility relation R where \(\sigma (w)=\{v \in W\; |\;wRv \}\).

  11. 11.

    In some presentations of PAL (e.g. van Ditmarsch et al. 2007), the ‘box’ dual of \(\langle \varphi \rangle \), \([\varphi ]\), is taken as primitive. Apart from duality, the relation between the two is characterized by the PAL validity \(\langle \varphi \rangle \psi \leftrightarrow [\varphi ]\psi \wedge \varphi \).

  12. 12.

    For a further discussion about the relation between questions as linguistic objects and the inquisitive disjunction see, e.g. Ciardelli (2016), p. 51.

  13. 13.

    If the information state s is sufficient to resolve a question, then any stronger piece of information \(s'\) should be sufficient as well for a resolution of the question. We thus formally require that questions, understood as sets of sets of possible worlds, will be downward closed: if \(s\in Q\) and \(s' \subseteq s\) then \(s' \in Q\) (where s and \(s'\) are information states, and Q a question). The downward closure condition is omitted from Fig. 1 for graphical simplicity.

  14. 14.

    In IEL, \(\sigma (w)\) itself is obtained from a more general function, \(\Sigma \), describing the issue entertained by the agent. See the Appendix for the exact details.

  15. 15.

    The version of inquisitive epistemic logic presented here (an extension of the inquisitive logic known as InqB) cannot distinguish between the informative content of a sentence and the presuppositional content of a sentence. The two are not the same, however, and a logical analysis of presupposition should be able to distinguish them. For further discussion, including an extension of inquisitive semantics that can distinguish the two notions, see Roelofsen (2015).

  16. 16.

    See the appendix for the full formal definition of the support conditions of update clause in IDEL. For further details, see Ciardelli (2016, p. 313).

  17. 17.

    In the multi-agent case, the update \( \langle \mu \rangle \) will affect the knowledge state of all agents. This can have unintended consequences, for instance making \(\mu ^!\) common knowledge. To avoid this, one can use private announcements that will only affect the epistemic state of a distinguished subset of agents. See van Gessel (2018) for the details.

  18. 18.

    E.g. the Bush example from Schaffer (2007b) and the diamonds example from Schaffer and Szabó (2014).

  19. 19.

    The full reasons for this fact go beyond the technical scope of this paper (see Ciardelli 2016). Note that sentences of the form \(\langle \mu \rangle K \varphi \) are declaratives, and so truth conditional (see Ciardelli 2016, p. 318), meaning, roughly, that they have no inquisitive content. Furthermore, any truth conditional sentence \(\varphi \) of IDEL that does not contain the E operator (see Appendix) is equivalent to a sentence without inquisitive disjunctions (ibid. p. 209).

  20. 20.

    Further, note that it is a strength of Schaffer’s argument that it rests on simple examples where \(\varphi \) is non-epistemic. While it is true that sentences with the IDEL E modality are not equivalent to sentences without inquisitive disjunctions in general (see footnote 19), an attempt to build an entire theory of knowledge (according to which all knowledge is contrastive knowledge) from such examples will significantly weaken Schaffer’s argument.

  21. 21.

    It might indeed be the case that contrasts can explain away skeptical intuitions; the point is that these contrasts can be expressed in the language of binary knowledge.

  22. 22.

    The only difference is that given such a variant of IDEL, the reduction (or translation) happens outside of the object language. The result of the translation still holds.

  23. 23.

    Here I am slightly diverging from the formulation of IDEL in Ciardelli (2016). There, the basic dynamic modality appears in the box version as \([\varphi ] \psi \). The semantic clause is

    $$\begin{aligned} \mathscr {M}, s \models [ \varphi ] \psi \;\; \Leftrightarrow \;\; \mathscr {M}_\varphi , \; s \cap \{w: \mathscr {M}, w \models \varphi \} \models \psi \end{aligned}$$

    Ciardelli (2016, p. 313). The truth condition of \([\varphi ]\psi \) relative to a world can then be retrieved to the usual

    $$\begin{aligned} \mathscr {M}, w \models [ \varphi ] \psi \Leftrightarrow \mathscr {M}, w \models \varphi \;\; implies\;\; \mathscr {M}_{\varphi }, w \models \psi \end{aligned}$$

    Moreover, the reduction axioms for the dynamic modality (see later section) are given in terms of \([\varphi ]\psi \). In accordance to the general DEL validity \(\langle \varphi \rangle \psi \leftrightarrow [\varphi ]\psi \wedge \varphi \), I define \(\langle \varphi \rangle \psi \) in IDEL as \([\varphi ]\psi \wedge \varphi ^!\). This explains my choice for the support clause I give for \(\langle \varphi \rangle \psi \). The intuitive difference between \(\langle \varphi \rangle \psi \) and \([\varphi ]\psi \) from DEL persists in IDEL: whenever \(\varphi \) is unsupported, \(\langle \varphi \rangle \psi \) is also unsupported while \([\varphi ]\psi \) is vacuously supported. Thanks to Ivano Ciardelli for his helpful comment on this point.


  1. Aloni, M., & Egre, P. (2008). Alternative questions and knowledge attributions. Philosophical Quarterly, 60, 1–27.

    Article  Google Scholar 

  2. Aloni, M., Égré, P., & de Jager, T. (2013). Knowing whether A or B. Synthese, 190(14), 2595–2621.

    Article  Google Scholar 

  3. Baltag, A., & Renne, B. (2016). Dynamic epistemic logic. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. Winter 2016 edition.

  4. Ciardelli, I. (2016). Questions in logic. Ph.D. Thesis, University of Amsterdam.

  5. Ciardelli, I., Groenendijk, J., & Roelofsen, F. (2018). Inquisitive semantics. In Oxford surveys in semantics and pragmatics. Oxford: Oxford University Press.

  6. Ciardelli, I., & Roelofsen, F. (2015). Inquisitive dynamic epistemic logic. Synthese, 192(6), 1643–1687.

    Article  Google Scholar 

  7. Cross, C., & Roelofsen, F. (2016). Questions. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. Spring 2016 edition.

  8. Dretske, F. I. (1972). Contrastive statements. Philosophical Review, 81(4), 411–437.

    Article  Google Scholar 

  9. Gerken, M. (2013). Epistemic focal bias. Australasian Journal of Philosophy, 91(1), 41–61.

    Article  Google Scholar 

  10. Groenendijk, J. (2007). The dynamics of inquiry. In Central APA. Handout.

  11. Hawke, P. (2016). Questions, topics and restricted closure. Philosophical Studies, 173(10), 2759–2784.

    Article  Google Scholar 

  12. Hintikka, J. (1975). Different constructions in terms of the basic epistemological verbs: A survey of some problems and proposals. In The intensions of intentionality and other new models for modalities (pp. 1–25). Dordrecht: D. Reidel.

    Google Scholar 

  13. Karjalainen, A., & Morton, A. (2003). Contrastive knowledge. Philosophical Explorations, 6(2), 74–89.

    Article  Google Scholar 

  14. Lewis, D. (1982). Whether’ report. In T. Pauli (Ed.), 320311: Philosophical essays dedicated to Lennart Åqvist on his fiftieth birthday (pp. 194–206). Uppsala: University of Uppsala Press.

    Google Scholar 

  15. Roelofsen, F. (2015). The semantics of declarative and interrogative lists. Manuscript, ILLC University of Amsterdam.

  16. Rysiew, P. (2011). Epistemic contextualism. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy. Winter 2011 edition.

  17. Schaffer, J. (2004). From contextualism to contrastivism. Philosophical Studies, 119(1–2), 73–104.

    Article  Google Scholar 

  18. Schaffer, J. (2005). Contrastive knowledge. In T. S. Gendler & J. Hawthorne (Eds.), Oxford studies in epistemology (Vol. 1, p. 235). Oxford: Oxford University Press.

    Google Scholar 

  19. Schaffer, J. (2007a). Closure, contrast, and answer. Philosophical Studies, 133(2), 233–255.

    Article  Google Scholar 

  20. Schaffer, J. (2007b). Knowing the answer. Philosophy and Phenomenological Research, 75, 383–403.

    Article  Google Scholar 

  21. Schaffer, J. (2008). The contrast-sensitivity of knowledge ascriptions. Social Epistemology, 22(3), 235–245.

    Article  Google Scholar 

  22. Schaffer, J., & Knobe, J. (2012). Contrastive knowledge surveyed. Nous, 46(4), 675–708.

    Article  Google Scholar 

  23. Schaffer, J., & Szabó, Z. G. (2014). Epistemic comparativism: A contextualist semantics for knowledge ascriptions. Philosophical Studies, 168(2), 491–543.

    Article  Google Scholar 

  24. Stanley, J. (2011). Know how. Oxford: Oxford University Press.

    Google Scholar 

  25. van Benthem, J., & Minică, Ş. (2009). Toward a dynamic logic of questions. In X. He, J. Horty, & E. Pacuit (Eds.), Logic, rationality, and interaction (pp. 27–41). Berlin: Springer.

    Google Scholar 

  26. van Ditmarsch, H., van der Hoek, W., & Kooi, B. (2007). Dynamic epistemic logic. In Synthese library. Netherlands: Springer.

  27. van Gessel, T. (2018). Action models in inquisitive logic. Synthese.

Download references


I would like to thank Johan van Benthem, Ray Briggs, Ivano Ciardelli, Krista Lawlor, Lisa Modifica, Jonathan Schaffer and and two anonymous referees of this journal for many helpful comments, suggestions and corrections on earlier versions of this article.

Author information



Corresponding author

Correspondence to Michael Cohen.

Additional information

Publisher's Note

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



Inquisitive dynamic epistemic logic

Definition 1


An issue I is a non-empty, downward closed set of information states. If \(t \in I\) then t is said to resolve I.

With the notion of an issue one can define an inquisitive epistemic model.

Definition 2

Inquisitive epistemic model

An inquisitive epistemic model is a triple \(\mathscr {M}=(W,V,\Sigma )\) s.t.

  • W is a set of possible worlds,

  • V is a valuation function, \(V: \mathscr {P} \rightarrow P(W)\), and

  • \(\Sigma \) is a state map, \(\Sigma : W \rightarrow P(P(W))\), taking worlds and returning issues.

The information state of the agent at a world, \(\sigma (w)\), is defined as \(\sigma (w) = \bigcup \Sigma (w)\).

Definition 3

The language of IDEL

The entire language of IDEL is defined inductively as:

We abbreviate \(\varphi \rightarrow \bot \) as \(\lnot \varphi \), \(\lnot (\lnot \varphi \wedge \lnot \psi )\) as \(\varphi \vee \psi \), as \(?\alpha \), and \(\varphi ^! \wedge [\varphi ]\psi \) as \(\langle \varphi \rangle \psi \).

Definition 4


Let \(\mathscr {M}\) be an inquisitive epistemic model and s an information state in \(\mathscr {M}\). Then:

  1. 1.

    \(\mathscr {M},s \models p\) iff \(w \in V(p)\) for all \(w \in s\)

  2. 2.

    \(\mathscr {M},s \models \bot \) iff \(s= \emptyset \)

  3. 3.

    iff \(\mathscr {M},s \models \varphi \) or \(\mathscr {M},s \models \varphi \)

  4. 4.

    \(\mathscr {M},s \models \varphi \wedge \psi \) iff \(\mathscr {M},s \models \varphi \) and \(\mathscr {M},s \models \psi \)

  5. 5.

    \(\mathscr {M},s \models \alpha \rightarrow \varphi \) iff for any \( t \subseteq s \), if \(\mathscr {M}, t \models \alpha \), then \(\mathscr {M}, t \models \varphi \)

  6. 6.

    \(\mathscr {M},s \models K \varphi \) iff for any \(w \in s\), \(\mathscr {M},\sigma (w) \models \varphi \)

  7. 7.

    \(\mathscr {M},s \models E \varphi \) iff for any \(w \in s\) and for any \(t \in \Sigma (w)\), \(\mathscr {M},t \models \varphi \)

  8. 8.

    \(\mathscr {M}, s \models \langle \varphi \rangle \psi \) iff \(\mathscr {M}, s \models \varphi ^! \;\; and\;\; \mathscr {M}_\varphi , \; s \cap \{w: \mathscr {M}, w \models \varphi \} \models \psi \).Footnote 23

For the derived connectives the following support conditions follow:

\(\mathscr {M},s \models \lnot \alpha \) iff for any non-empty \(t, t\subseteq s\), \(\mathscr {M},t \not \models \alpha \)

\(\mathscr {M},s \models \alpha \vee \beta \) iff there are \(t_1, t_2\) s.t. \(s=t_1 \cup t_2\), \(\mathscr {M},t_1 \models \alpha \) and \(\mathscr {M},t_2 \models \beta \).

To understand clause 8., we need a definition of an updated model:

Definition 5

Updated model

An inquisitive epistemic model \(\mathscr {M}\) after the announcement that \(\varphi \), \(\mathscr {M}_\varphi =(W_\varphi , V_\varphi , \Sigma _\varphi )\) is defined as:

\(W_\varphi = W \cap \{w: \mathscr {M}, w \models \varphi \}\)

\(V_\varphi = V\) restricted to \(W_\varphi \)

\(\Sigma _\varphi (w)= \Sigma (w) \cap \{s : \mathscr {M} ,s \models \varphi \}\)

Proof of Fact 1


PAL will suffice for the proof, since IDEL is a conservative extension of PAL (Ciardelli 2016).

Assume for reductio that Fact 1 is false, and consider the sentence \(\langle p \rangle K p\). Thus, there is a sentence \(K \varphi ^*\) s.t. \(\vdash \langle p \rangle K p \leftrightarrow K \varphi ^*\). By the reduction axioms of PAL, we have that \(\langle p \rangle K p\) is provably equivalent to p, so \(\vdash p \leftrightarrow K \varphi ^*\). Consider a T model containing only w and v with the universal relation, and let p be true in w and false in v. Since \(w \models p\), \(w \models K\varphi ^*\). Since the model is S5, \(v \models K\varphi ^*\). On the other hand, since \(v \models \lnot p\), we have that \(v \models \lnot K \varphi ^*\), contradiction. \(\square \)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cohen, M. Reducing Contrastive Knowledge. Erkenn (2019).

Download citation