Skip to main content

A Modal Logic for Supervised Learning


Formal learning theory formalizes the process of inferring a general result from examples, as in the case of inferring grammars from sentences when learning a language. In this work, we develop a general framework—the supervised learning game—to investigate the interaction between Teacher and Learner. In particular, our proposal highlights several interesting features of the agents: on the one hand, Learner may make mistakes in the learning process, and she may also ignore the potential relation between different hypotheses; on the other hand, Teacher is able to correct Learner’s mistakes, eliminate potential mistakes and point out the facts ignored by Learner. To reason about strategies in this game, we develop a modal logic of supervised learning and study its properties. Broadly, this work takes a small step towards studying the interaction between graph games, logics and formal learning theory.

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

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


  1. Generally speaking, to define the existence of winning strategies for players, we need to extend SLG with some fixpoint operators. We leave this for future inquiry.

  2. For instance, in sabotage games, we use \(\Diamond \) to capture actions of Learner in formulas describing winning strategies (if they exist). See Gierasimczuk et al. (2009).

  3. In contrast, one extreme case of non-cooperative variants of SLG might be that Learner is allowed to stay at her current position in each round: she makes no efforts to reach the goal node.

  4. By abuse of notation, for any \(\varphi \in \mathcal {L}_{\blacklozenge \langle + \rangle }\), \(\varphi ^\star \) is a formula of the bridge modal logic.

  5. The four modalities used in its proof can be reduced to two by a standard argument Kracht and Wolter (1999), but we will omit the details because of the syntactic cost involved in writing the formulas.

  6. For more discussions on the applications of SG-style frameworks to paradigms of learning theory, we refer to Gierasimczuk et al. (2009), whose arguments also apply to SLG after minor modifications.


  • Angluin, D. (1987). Learning regular sets from queries and counterexamples. Information and Computation, 75, 87–106.

    Article  Google Scholar 

  • Areces, C., Fervari, R., & Hoffmann, G. (2012). Moving arrows and four model checking results. In Ong, L., & Queiroz, R., editors, WoLLIC 2012, volume 7456 of LNCS, pp 142–153. Springer.

  • Areces, C., Fervari, R., & Hoffmann, G. (2015). Relation-changing modal operators. Journal of the IGPL, 23(4), 601–627.

    Article  Google Scholar 

  • Areces, C., Fervari, R., Hoffmann, G., & Martel, M. (2018). Satisfiability for relation-changing logics. Journal of Logic and Computation, 28, 1143–1470.

    Article  Google Scholar 

  • Areces, C., Figueira, D., Figueira, S., & Mera, S. (2011). The expressive power of memory logics. The Review of Symbolic Logic, 4, 290–318.

    Article  Google Scholar 

  • Aucher, G., van Benthem, J., & Grossi, D. (2018). Modal logics of sabotage revisited. Journal of Logic and Computation, 28(2), 269–303.

    Article  Google Scholar 

  • Baltag, A., Bezhanishvili, N., Özgün, A., & Smets, S. (2019). A topological approach to full belief. Journal of Philosophical Logic, 48(2), 205–244.

    Article  Google Scholar 

  • Blackburn, P., de Rijke, M., & Venema, Y. (2001). Modal logic. Cambridge University Press.

  • Blackburn, P., & Seligman, J. (1995). Hybrid languages. Journal of Logic, Language and Information, 4(3), 251–272.

    Article  Google Scholar 

  • Blando, F.Z., Mierzewski, K., & Areces, C. (2020). The modal logics of the poison game. In Liu, F., Ono, H., & Yu, J. editors, Knowledge, Proof and Dynamics, Logic in Asia: Studia Logica Library, pp 3–23. Springer.

  • Chandra, A., & Merlin, P. (1977) Optimal implementation of conjunctive queries in relational databases. In Proceedings of 9th ACM STOC, pp 77–90.

  • Chang, C.C., & Keisler, H. J. (1973). Model Theory. Studies in Logic and the Foundations of Mathematics. North-Holland.

  • Dégremont, C., & Gierasimczuk, N. (2011). Finite identification from the viewpoint of epistemic update. Information and Computation, 209(3), 383–396.

    Article  Google Scholar 

  • Gettier, E. (1963). Is justified true belief knowledge? Analysis, 23(6), 121–123.

    Article  Google Scholar 

  • Gierasimczuk, N. (2009). Bridging learning theory and dynamic epistemic logic. Synthese, 169(2), 371–384.

    Article  Google Scholar 

  • Gierasimczuk, N. (2009). Learning by erasing in dynamic epistemic logic. In Dediu, A. H., Ionescu, A. M., & Martin-Vide,C. editors, Proceedings of LATA09, volume 5457 of LNCS, pp 362–373. Springer.

  • Gierasimczuk, N., & de Jongh, D. (2013). On the complexity of conclusive update. The Computer Journal, 56(3), 365–377.

    Article  Google Scholar 

  • Gierasimczuk, N., Kurzen, L., & Velázquez-Quesada, F.R. (2009) Learning and teaching as a game: A sabotage approach. In He, X., Horty, J., Pacuit, E. editors, Proceedings of LORI 2009, volume 5834 of LNCS, pp 119–132. Springer.

  • Gold, E. (1967). Language identification in the limit. Information and Control, 10, 447–474.

    Article  Google Scholar 

  • Grossi, D., & Rey, S. (2019) Credulous acceptability, poison games and modal logic. In Agmon, N., Taylor, M. E., Elkind, E., & Veloso, M. editors, Proceedings of AAMAS 2019, pp 1994–1996.

  • Harel, D. (1985). Recurring dominoes: Making the highly undecidable highly understandable. In Selected papers of the international conference on “foundations of computation theory” on Topics in the theory of computation, pp 51–71.

  • Kelly, K. T., Schulte, O., & Juhl, C. (1997). Learning theory and the philosophy of science. Philosophy of Science, 64(2), 245–267.

    Article  Google Scholar 

  • Kracht, M., & Wolter, F. (1999). Normal monomodal logics can simulate all others. Journal of Symbolic Logic, 64, 99–138.

    Article  Google Scholar 

  • Lange, S., Zeugmann, T., & Kapur, S. (1996). Monotonic and dual monotonic language learning. Theor. Comput. Sci., 155, 365–410.

    Article  Google Scholar 

  • Li, D. (2020). Losing connection: The modal logic of definable link deletion. Journal of Logic and Computation, 30, 715–743.

    Article  Google Scholar 

  • Li, D., Ghosh, S., Liu, F., & Tu, Y. (2021). On the subtle nature of a simple logic of the hide and seek game. In Silva, A., Wassermann, R., & de Queiroz, R. editors, Proceedings of WoLLIC2021, volume 13038 of LNCS, pp 201–218.

  • Mukouchi, Y. (1992) Characterization of finite identification. In Jantke, K. P. editor, Analogical and Inductive Inference, volume 642 of LNAI, pp 260–267. Springer.

  • Thompson, D. (2020). Local fact change logic. In Liu, F., Ono, H., & Yu, J. editors, Knowledge, Proof and Dynamics, Logic in Asia: Studia Logica Library, pp 73–96. Springer.

  • van Benthem, J. (1984). Correspondence theory. In Gabbay, D., & Guenthner, F. editors, Handbook of Philosophical Logic: Volume II: Extensions of Classical Logic, pp 167–247. Springer.

  • van Benthem, J. (2014). Logic in games. The MIT Press.

  • van Benthem, J., & Liu, F. (2020) Graph games and logic design. In Liu, F., Ono, H., & Yu, J. editors, Knowledge, Proof and Dynamics, pp 125-146. Springer.

  • van Benthem, J., Mierzewski, K., & Blando, F. Z. (2020). The modal logic of stepwise removal. The Review of Symbolic Logic, pp 1–28.

Download references


We thank Johan van Benthem, Fenrong Liu, Fernando R. Velázquez-Quesada, Nina Gierasimczuk and Lena Kurzen for their inspiring suggestions. We also wish to thank the two anonymous reviewers for their very useful comments for improvement. Dazhu Li is supported by the Major Program of the National Social Science Foundation of China [17ZDA026].

Author information

Authors and Affiliations


Corresponding author

Correspondence to Dazhu Li.

Additional information

Publisher's Note

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

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Baltag, A., Li, D. & Pedersen, M.Y. A Modal Logic for Supervised Learning. J of Log Lang and Inf 31, 213–234 (2022).

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Formal learning theory
  • Modal logic
  • Dynamic logic
  • Undecidability
  • Graph games