Abstract
Epistemic modal predicate logic raises conceptual problems not faced in the case of alethic modal predicate logic: Frege’s “Hesperus-Phosphorus” problem—how to make sense of ascribing to agents ignorance of necessarily true identity statements—and the related “Hintikka-Kripke” problem—how to set up a logical system combining epistemic and alethic modalities, as well as others problems, such as Quine’s “Double Vision” problem and problems of self-knowledge. In this paper, we lay out a philosophical approach to epistemic predicate logic, implemented formally in Melvin Fitting’s First-Order Intensional Logic, that we argue solves these and other conceptual problems. Topics covered include: Quine on the “collapse” of modal distinctions; the rigidity of names; belief reports and unarticulated constituents; epistemic roles; counterfactual attitudes; representational versus interpretational semantics; ignorance of co-reference versus ignorance of identity; two-dimensional epistemic models; quantification into epistemic contexts; and an approach to multi-agent epistemic logic based on centered worlds and hybrid logic.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
We do not mean to suggest that there is a consensus on the proper semantics for alethic modal predicate logic. What we have in mind here is the standard development of Kripke-style semantics for modal predicate logic (see, e.g., [12]). To the extent that we support the Conservative Approach so understood, one might expect that epistemic operators could be smoothly introduced into alethic modal predicate logic developed in other ways as well. As another point of qualification, we are not arguing for conservativeness with respect to the question of relational versus neighborhood semantics for epistemic logic (see [3]).
- 2.
Wettstein [65] uses ‘cognitive fix’ in a more demanding sense, requiring not merely a way of thinking about an object, but also accurate beliefs about what distinguishes the object from others.
- 3.
It seems that this leaves Russell without a solution to the problem of the cognitive fix, relative to logically proper names. For a discussion of this issue, see [66].
- 4.
See [53] for a fuller account of the slingshot and Føllesdal’s treatment of it.
- 5.
What if we amend (B) and (C) so that they begin with sequestered terms? Extending ‘Dthat’ to class abstracts, we have:
- (A):
-
van Benthem is a human.
- (B\({^\prime }\)):
-
\( \text {Dthat}(\left\{ x \mid x = \emptyset \, \& \right. \) van Benthem is a human\(\left. \right\} ) = \{\emptyset \}\).
- (C\({^\prime }\)):
-
\( \text {Dthat}(\left\{ x \mid x = \emptyset \, \& \right. \) van Benthem is a logician\(\left. \right\} ) = \{\emptyset \}\).
- (D):
-
van Benthem is a logician.
Could Quine still argue that (A) and (B\({^\prime }\)), and (C\({^\prime }\)) and (D), have the same modal status? (D) is clearly contingent; the multi-talented van Benthem could have been a computer scientist. Is (C\({^\prime }\)) contingent? The result of applying Dthat to a description or class abstract is supposed to be a rigid designator, or more generally, a modally loyal term in the sense defined above. Thus, evaluating
-
\( \Box \,\text {Dthat}(\left\{ x \mid x = \emptyset \, \& \right. \) van Benthem is a logician\(\left. \right\} ) = \{\emptyset \}\)
amounts to checking for every possible world whether ‘\(y= \{\emptyset \}\)’ is true there, where \(y\) is assigned the object that is designated by ‘\( \text {Dthat}(\left\{ x \mid x = \emptyset \, \& \right. \) van Benthem is a logician\(\left. \right\} )\)’ in the actual world (or the world of the context of utterance). Since the object designated by ‘\( \text {Dthat}(\left\{ x \mid x = \emptyset \, \& \right. \) van Benthem is a logician\(\left. \right\} )\)’ in the actual world is \(\{\emptyset \}\), the check succeeds, so (C\({^\prime }\)) is necessary. Hence (C\({^\prime }\)) and (D) do not have the same modal status; so the modalities do not collapse.
- 6.
- 7.
- 8.
Cf. Lewis [44, p. 543] on watching as a “relation of acquiantance”.
- 9.
While our strategy is based on the approach of Crimmins and Perry [19], those authors took cognitive fixes to be notions and took notions to be unarticulated constituents of belief-reports. Subsequently Perry has developed an account that is basically similar, but takes cognitive fixes to be notion-networks, basically intersubjective routes through notions. Of course, traditionally cognitive fixes have been taken to be individual concepts. We believe that the concept of a role provides a general framework into which all of these candidates can be fit.
- 10.
Cf. Lewis [44, p. 542]: “If I have a belief that I might express by saying “Hume was noble”, I probably ascribe nobility to Hume under the description “the one I have heard of under the name of ‘Hume’ ”. That description is a relation of acquaintance that I bear to Hume. This is the real reason why I believe de re of Hume that he was noble”.
- 11.
Cf. Lewis [45, 10f] on “relations of epistemic rapport” or “relations of acquaitance”.
- 12.
There are a few small differences. First, Fitting allows relation symbols \(Q\) (but not \(=\)) to apply to what we call role variables—his intensional variables—whereas to simplify the definition of the language (to avoid typing relations), we do not; this is why we do not count role variables among the terms \(t\) in Definition 22.1. Second, we include constant symbols in the language, whereas for convenience Fitting does not. Third, we have a bimodal language with
and 
, whereas Fitting has a monomodal language with \(\Box \). Finally, where we write \(\mathsf {P}(t,r_i)\), Fitting would write \(\mathsf {D}(r_i,t)\). - 13.
These models are almost the same as those for “contingent identity systems” in [48] (cf. [37]) and [54], but for a few differences: we follow Fitting in allowing \(F\) to contain partial functions; Parks does not deal with constants; and while Priest does deal with constants, he treats them as non-rigid. The differences between our models and Fitting’s [24, 25] are that we deal with constants, and Fitting defines \(V\) so that predicates can apply not only to elements of \(D\), but also to elements of \(F\) (cf. [37]).
- 14.
We are not suggesting that all there is to a role is a partial frunction from \(W\) to \(D\); but such a function captures an important aspect of a role, namely the players of the role across worlds.
- 15.
Some minor changes must be made, e.g., since we include constants in the language (recall note 12), but we will not go into the details here.
- 16.
One may then wish to add the assumption that for all \(f\in F\), if \(f(w)=d\), then \(d\in V(\mathsf {E},w)\), i.e., if an object \(d\) plays a role for the agent in \(w\), then \(d\) exists in \(w\), validating \(\mathsf {P}(t,r_i)\rightarrow \exists _ax \,t=x\).
- 17.
- 18.
We did not define the interpretation of role variables in Definition 22.4, since we do not officially count them as terms (recall note 12), and they only appear in the \(\mathsf {P}(t,r_i)\) clause in Definition 22.5, where the assignment \(\mu \) takes care of them directly; but the definition would be \(\left[ r_i\right] _{\fancyscript{M},\,w,\,\mu } = \mu (r_i)\).
- 19.
The exception among the authors referenced in note 17 is Carlson, who allows \(F\) to contain partial functions and uses a three-valued semantics to deal with undefined terms.
- 20.
Remember that in (27) and (28), we read
as “it is doxastically necessary that \(\varphi \)” or “in all worlds compatible with the agent’s beliefs, \(\varphi \)”. The whole of (27) gives the condition that the truth of the belief report imposes on the actual world and the space of the agent’s doxastic alternatives, so we would not read the second conjunct as “the agent believes that there exists ...”. - 21.
Belardinelli and Lomuscio [6] include both \(x\) and \(z\) variables in their multi-agent quantified epistemic logic. Instead of distinguishing two types of variables, we could instead distinguish two types of quantifiers, in the tradition of Hintikka’s [33] distinction between \(\exists y\) and \(\mathrm {E}y\). By understanding \(\exists z\) quantification in terms of agent-relative roles, we are following Perry [51].
- 22.
This point is inspired by Crimmins and Perry [19] on notion provision vs. notion contraint.
- 23.
We could have two sorts of function variables, \(r_1,r_2,\dots \) for roles and \(s_1,s_2,\dots \) for non-role functions. Or we could indicate the difference between role functions and non-role functions by a one-place predicate \(\mathsf {Role}\) whose extension contains only functions to be thought of as roles.
- 24.
According to this intuitive understanding, the extension of \(\mathsf {Stip}\) should be a functional relation: if \(\mathsf {Stip}(r,s)\) holds, then \(\mathsf {Stip}(r,s{^\prime })\) should not hold for any \(s{^\prime }\not = s\).
- 25.
The last of the quoted sentences occurs in footnote 7 of [1].
- 26.
One may try to apply a similar strategy to predicate symbols in order to model agents who do not believe/know that two (necessarily) co-extensive predicates are co-extensive.
- 27.
Compare this to the “fixedly actually” operator of Davis and Humberstone [21].
- 28.
To deal in the 1D framework with an agent who does not believe, e.g., that something contains water iff it contains H20, we would need to generalize the notion of role so that properties (understood extensionally, intensionally, or hyper-intensionally) can play roles for an agent.
- 29.
Note that (55)/(56) does not require the existence of anyone who actually plays \(r_1\). We can express a reading that requires the existence of a role-player with:
. - 30.
Aloni considers it an advantage of this more general semantics that we can have
-
(69)
\(\fancyscript{M},w\vDash _\pi \exists z^i\varphi (z^i)\wedge \lnot \exists z^j\varphi (z^j)\),
as if there is a shift in context mid-formula. Instead of doing this with one of Aloni’s models, we could consider two regular models \(\fancyscript{M}=\langle W,R_a,R_d,D,\pi (i),V \rangle \) and \(\fancyscript{M}{^\prime }=\langle W,R_a,R_d,D,\pi (j),V \rangle \), each associated with a different context, such that
-
(70)
\(\fancyscript{M},w\vDash _\mu \exists z\varphi (z)\) and \(\fancyscript{M}{^\prime },w\vDash _{\mu {^\prime }}\lnot \exists z\varphi (z)\).
Aloni’s motivation for considering (69) is the following kind of reasoning:
-
(I)
Ralph believes that the man with the brown hat is a spy.
-
(II)
The man with the brown hat is Ortcutt.
-
(III)
So Ralph believes of Ortcutt that he is a spy.
-
(IV)
Ralph believes that the man seen on the beach is not a spy.
-
(V)
The man seen on the beach is Ortcutt.
-
(VI)
So Ralph does not believe of Ortcutt that he is a spy.
Aloni concludes that
-
(71)
should be satisfiable, which it is in her semantics. However, it seems to us to be a mistake to conclude (VI) on the basis of (IV) and (V). Instead, by analogy with (I)-(III), one should conclude
- (VI\({^\prime }\)):
-
So Ralph believes of Ortcutt that he is not a spy.
Then we can express the compatibility of (III) and (VI\({^\prime }\)) by the satisfiable sentence
-
(72)
.
This is not to say, however, that there are not other good motivations for the more general semantics.
-
(69)
- 31.
Thus, we have a term-modal logic in the sense of [23].
- 32.
In English, to say “from the point of view of \(t\), \(\varphi \)”, might suggest that \(t\) believes \(\varphi \), but this it not the intended reading of \(\mathsf {pov}_t\varphi \), as its formal truth definition below makes clear.
- 33.
While we take a centered world to be any element of \(W\times D\), one may wish to only admit pairs \(\langle w,c\rangle \) such that \(c\) is an agent (in some distinguished set \(Agt\subseteq D\)) and \(c\) exists in \(w\) (using the existence predicate of Sect. 22.3, \(c\in V(\mathsf {E},w)\)), but for simplicity we do not make these assumptions here. Also for simplicity, we are not putting explicit times into the centered worlds. Adding a temporal dimension to our framework would expand its range of application to further interesting issues.
- 34.
By the truth definition, we have
iff for all \(w{^\prime }\in W\), \(c{^\prime }\in D\), if \(\langle w,[t]_{\fancyscript{M},\,w,\,\mu }\rangle R_d\langle w{^\prime },c{^\prime }\rangle \), then \(\fancyscript{M},w{^\prime },c{^\prime }\vDash _\mu \varphi \). The problem with taking the 
operators as primitive instead of 
is that we would then lose important results of modal correspondence theory. For example, requiring that \(R_d\) be reflexive (thinking of it now as an epistemic accessibility relation) would not guarantee the validity of 
, since the reflexivity of \(R_d\) would not guarantee that \(\langle w,[t]_{\fancyscript{M},\,w,\,\mu }\rangle R_d\langle w,c\rangle \). But reflexivity would guarantee the validity of 
, as desired. - 35.
One may want to define the self-role such that for all worlds \(w\) and agents \(c\), \(f_{self}(w,c)=c\).
- 36.
A similar analysis applies to other well-known problems in the theory of reference, such as Castañeda’s [16] puzzle about the first person. Surely through most of his life after 1884, Samuel Clemens believed that he wrote Huckelberry Finn. But one can imagine that in his dotage, Clemens held a copy of the book in his hand, saw that it was written by Mark Twain, but couldn’t remember that ‘Mark Twain’ had been his pseudonym and had no inclination to say “I wrote this”. Castañeda made the point, with many similar examples, that even in the latter case, we could say
-
(74)
Samuel Clemens believes that he wrote Huckelberry Finn.
since he is Mark Twain, and he believes that Mark Twain wrote Huckelberry Finn. However, in the sense in which it was true through much of his life that he believed he wrote Huckelberry Finn, at this moment late in his life, it is not. There is a reading of (74) on which it is false.
In the case we are imagining, Samuel Clemens plays (at least) two roles in Samuel Clemens’ life, the self-role \(\mathbf {r}_{self}\) and the role \(\mathbf {r}_{MT}\) of being the source of the ‘Mark Twain’ name-network that is exploited by the use of that name on the book he holds in his hands. Given this, we can distinguish between the two readings of (74), the first false and the second true, as follows:
-
(75)
; -
(76)
.
-
(74)
and 
, whereas Fitting has a monomodal language with 
as “it is doxastically necessary that 
.
.
iff for all 
operators as primitive instead of 
is that we would then lose important results of modal correspondence theory. For example, requiring that 
, since the reflexivity of 
, as desired.
;
.References
Aloni M (2005) Individual concepts in modal predicate logic. J Philos Log 34(1):1–64
Areces C, ten Cate B (2006) Hybrid logic. In: van Benthem J, Wolter F, Blackburn P (eds) Handbook of modal logic. Elsevier, Amsterdam, pp 821–868
Arló-Costa H, Pacuit E (2006) First-order classical modal logic. Studia Logica 84:171–210
Barnes KT (1976) Proper names, possible worlds, and propositional attitudes. Philosophia 6(1):29–38
Barwise J, Perry J (1983) Situations and attitudes. MIT Press, Cambridge
Belardinelli F, Lomuscio A (2009) Quantified epistemic logics for reasoning about knowledge in multi-agent systems. Artif Intell 173:982–1013
van Benthem J (1985) Modal logic and classical logic. Bibliopolis, Napoli
van Benthem J (1993) Beyond accessibility: functional models for modal logic. In: de Rijke M (ed) Diamonds and defaults. Kluwer Academic, Dordrecht, pp 1–18
van Benthem J (2010a) Frame correspondences in modal predicate logic. In: Feferman S, Sieg W (eds) Proofs, categories and computations: essays in honor of Grigori Mints. College Publications, London
van Benthem J (2010b) Modal logic for open minds. CSLI Publications, Stanford
van Benthem J (2011) Logical dynamics of information and interaction. Cambridge University Press, Cambridge
Braüner T, Ghilardi S (2006) First-order modal logic. In: van Benthem J, Wolter F, Blackburn P (eds) Handbook of modal logic. Elsevier, Amsterdam, pp 549–620
Burge T (1973) Reference and proper names. J Philos 70(14):425–439
Carlson L (1988) Quantified Hintikka-style epistemic logic. Synthese 74(2):223–262
Carnap R (1947) Meaning and necessity. University of Chicago Press, Chicago
Castañeda HN (1966) ‘He’: a study in the logic of self-consciousness. Ratio 7:130–157
Cresswell MJ, von Stechow A (1982) De Re belief generalized. Linguist Philos 5:503–535
Crimmins M (1992) Talk about belief. MIT Press, Cambridge
Crimmins M, Perry J (1989) The prince and the phone booth: reporting puzzling beliefs. J Philos 86(12):685–711
Cumming S (2008) Variabilism. Philos Rev 117(4):525–554
Davies M, Humberstone L (1980) Two notions of necessity. Philos Stud 38:1–30
Etchemendy J (1999) The concept of logical consequence. CSLI Publications, Stanford
Fitting M, Thalmann L, Voronkov A(2001) Term-modal logics. Studia Logica 69(1): 133–169
Fitting M (2004) First-order intensional logic. Ann Pure Appl Log 127(1–3):171–193
Fitting M (2006) FOIL axiomatized. Studia Logica 84(1):1–22
Fitting M, Mendelsohn RL (1998) First-order modal logic. Kluwer, Dordrecht
Føllesdal D (1961) Referential opacity and modal logic. Routledge, London (published 2004)
Frege G (1892) Über Sinn und Bedeutung. Zeitschrift für Philosophie und philosophische Kritik 100:25–50
Garson JW (2001) Quantification in modal logic. In: Gabbay D, Guenthner E (eds) Handbook of philosophical logic, 2nd edn, vol 3. Kluwer, Dordrecht, pp 267–323
Garson JW (2006) Modal logic for philosophers. Cambridge Universtiy Press, Cambridge
Grove AJ (1995) Naming and identity in epistemic logic part II: a first-order logic for naming. Artif Intell 74:311–350
Hintikka J (1962) Knowledge and belief: an introduction to the logic of the two notions. College Publications, London (Republished 2005)
Hintikka J (1969) On the logic of perception. Models for modalities. Reidel, Dordrecht, pp 151–183
Hintikka J (1970) Objects of knowledge and belief: acquaintances and public figures. J Philos 67:869–883
Holliday WH (2014) Epistemic closure and epistemic logic I: relevant alternatives and subjunctivism. J Philos Log (forthcoming). doi:10.1007/s10992-013-9306-2
Holliday WH (2014) Epistemic logic and epistemology. In: Hansson SO, Hendricks VF (eds) Handbook of formal philosophy. Springer (forthcoming)
Hughes GE, Cresswell MJ (1996) A new introduction to modal logic. Routledge, London
Israel D, Perry J (1996) Where monsters dwell. In: Seligman J, Westerståhl D (eds) Proceedings of the logic, language and computation: the 1994 Moraga, CSLI Publications, pp 303–316
Kaplan D (1968) Quantifying In. Synthese 19:178–214
Kaplan D (1989) Demonstratives. In: Perry J, Wettstein H, Almog J (eds) Themes from Kaplan. MIT Press, Cambridge, pp 481–563
Kripke S (1980) Naming and necessity. Harvard University Press, Cambridge
Lehmann S (1978) The Hintikka-Kripke problem. Philosophia 8:59–70
Lewis D (1977) Possible-world semantics for counterfactual logics: a rejoinder. J Philos Log 6:359–363
Lewis D (1979) Attitudes De Dicto and De Se. Philos Rev 88(4):513–543
Lewis D (1983) Individuation by acquaintance and by stipulation. Philos Rev 92(1):3–32
Linsky L (1979) Believing and necessity. Theory Decis 11:81–94
Ninan D (2012) Counterfactual attitudes and multi-centered worlds. Semant Pragmat 5(5):1–57
Parks Z (1974) Semantics for contingent identity systems. Notre Dame J Formal Log 15(2): 333–334
Perry J (1977) Frege on demonstratives. Philos Rev 86:474–497
Perry J (1986) Thought without representation. Proc Aristotelian Soc 60:137–151
Perry J (2009) Hintikka on demonstratives. Revue internationale de philosophie 4(250): 369–382
Perry J (2012) Reference and reflexivity, 2nd edn. CSLI Publications, Stanford
Perry J (2013) Føllesdal and Quine’s slingshot. In: Frauchiger M (ed) Reference, rationality, and phenomenology: themes from Føllesdal. Ontos Verlag, Frankfurt, pp 237–258
Priest G (2002) The hooded man. J Philos Log 31(5):445–467
Quine W (1953) Three grades of modal involvement. In: Proceedings of the 11th international congress of philosophy, vol 14. North-Holland Publishing, Amsterdam, pp 158–176
Quine W (1956) Quantifiers and propositional attitudes. J Philos 53(5):177–187
Rabinowicz W, Segerberg K (1994) Actual truth, possible knowledge. Topoi 13:101–105
Richard M (1983) Direct reference and ascriptions of belief. J Philos Log 12:425–452
Russell B (1905) On denoting. Mind 14(4):479–493
Russell B (1911) Knowledge by acquaintance and knowledge by description. Proc Aristotelian Soc 11:108–128
Schroeter L (2012) Two-dimensional semantics. In: Zalta EN (ed) The stanford encyclopedia of philosophy (winter 2012 edn)
Soames S (1989) Direct reference and propositional attitudes. In: Perry J, Wettstein H, Almog J (eds) Themes from kaplan. MIT Press, Cambridge, pp 393–419
Stalnaker RC (1984) Inquiry. MIT Press, Cambridge
Stalnaker RC (2010) Our knowledge of the internal world. Oxford University Press, Boston
Wettstein H (1988) Cognitive significance without cognitive content. Mind 97(385):1–28
Wishon D (2012) Russellian acquaintance and Frege’s puzzle (manuscript)
Yalcin S (2012) Quantifying in from a Fregean perspective (manuscript)
Acknowledgments
For helpful discussion or comments on this paper, we thank Johan van Benthem, Russell Buehler, Thomas Icard, David Israel, Ethan Jerzak, Alex Kocurek, Daniel Lassiter, John MacFarlane, Michael Rieppel, Shane Steinert-Threlkeld, Justin Vlasits, and Seth Yalcin.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer International Publishing Switzerland
About this chapter
Cite this chapter
Holliday, W.H., Perry, J. (2014). Roles, Rigidity, and Quantification in Epistemic Logic. In: Baltag, A., Smets, S. (eds) Johan van Benthem on Logic and Information Dynamics. Outstanding Contributions to Logic, vol 5. Springer, Cham. https://doi.org/10.1007/978-3-319-06025-5_22
Download citation
DOI: https://doi.org/10.1007/978-3-319-06025-5_22
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-06024-8
Online ISBN: 978-3-319-06025-5
eBook Packages: Humanities, Social Sciences and LawPhilosophy and Religion (R0)