Abstract
A proof-theoretic test for paradoxicality was famously proposed by Tennant: a paradox must yield a closed derivation of absurdity with no normal form. Drawing on the remark that all derivations of a given proposition can be transformed into derivations in normal form of a logically equivalent proposition, we investigate the possibility of paradoxes in normal form. We compare paradoxes à la Tennant and paradoxes in normal form from the viewpoint of the computational interpretation of proofs and from the viewpoint of proof-theoretic semantics.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In Tennant’s original account condition iii is stronger, as it is required that \(\mathscr {D}\) has a looping reduction; in Tennant (1995) the criterion was weakened: a non-terminating reduction, rather than a looping reduction, is required, in order to account for paradoxes without self-reference, such as Yablo’s.
In the examples considered by Tennant, paradoxes display exactly one redundancy.
We recall that an isomorphism between two types/formulas A, B is a pair made of a derivation of B from A and a derivation of A from B such that both their mutual compositions reduce to the identity derivation (on A and B, respectively).
More precisely, to obtain an isomorphism, one must add to the reduction rule \(\rightarrow _{\lambda }\) a dual rule
Indeed, it can be easily verified that \(\Delta K I\Delta\) reduces to \(\Delta K I \Delta\).
Recall that any T-paradox can be transformed into a closed derivation of any propositional constant.
This family of computational interpretations include: BHK, realizability, Dialectica, Curry–Howard. See Sorensen and Urzyczyn (2006).
To be more precise, the derivations of the formula \(N=(X\rightarrow X)\rightarrow (X\rightarrow X)\) are seen as representations \(\underline{n}\) of positive integers \(n\in \mathbb {N}\) and the proofs of the formula \(\underbrace{N\rightarrow \dots \rightarrow N}_{n {\text { times}}} \rightarrow N\) are seen as representations \(\underline{f}\) of (n-ary) computable arithmetical functions \(f(x_{1},\dots , x_{n})\). Hence, if all derivations of N are the same, one has that \(\underline{0}\) is the same as \(\underline{1}\) and that, for all computable function f and integers p, q, \(\underline{f}( \underline{p})\) is the same as \(\underline{f}(\underline{q})\).
For an exposition of these rules, see e.g. Negri and Plato (2001).
The fact that Russell’s paradox can be reformulated in purely semantical terms casts further doubt on the viability of a distinction between semantical and mathematical paradoxes.
More precisely, an essential net, a graphical formalism for Intuitionistic Linear Logic, Lamarche (2008).
Actually, formulas with polarities \(+,-\), corresponding to the fact that the node correspond to a positive or negative occurrence of the formula (see Lamarche 2008).
Reduction in proof-nets is just path composition, see Lamarche (2008).
Actually, the linear\(\rightarrow\)L rule, corresponding to the \(\otimes\)-rule of Multiplicative Linear Logic.
A typical elimination based clause is: “a non canonical derivation \(\mathscr {D}\) of \(A\rightarrow B\) is valid if, when modus ponens is applied to \(\mathscr {D}\) and a valid derivation \(\mathscr {D}'\) of A, the resulting derivation can be reduced into a valid derivation of B”. As this example shows, elimination-based semantics clauses recall the realizability semantics.
However, a difference between some T-paradox and the corresponding N-paradox appears when considering, rather than validity, the computational behavior of the associated \(\lambda\)-terms, reflected in how such terms are interpreted in Scott Domains. For instance, in the original model \(D^{\infty }\) proposed in Scott (1976), only solvable \(\lambda\)-terms have a non-empty interpretation. This fact reflects the intuition that solvable terms are the only to have a non trivial computational behavior, while non solvable terms correspond to everywhere undefined partial functions. Now, while the \(\lambda\)-term arising from the T-paradox Rus is non-solvable, the one arising from the N-paradox \(Rus^{*}\) (when the latter is considered as a derivation of \((\lambda \rightarrow \lambda )\rightarrow \bot\)) is solvable: its operative content is to send any function from \(\lambda\) to \(\lambda\) into the empty denotation. One can argue in a similar way in the case of N-paradoxes of conclusion \(\lambda \wedge \lnot \lambda\). Roughly speaking, both T-paradoxes and N-paradoxes correspond to partial functions, the former being, so to say, “more partial” than the latter.
References
Bainbridge E, Freyd PJ, Scedrov A, Scott PJ (1990) Functorial polymorphism. Theor Comput Sci 70:35–64
Blute R (1993) Linear logic, coherence and dinaturality. Theor Comput Sci 115(1):3–41
Coquand T, Herbelin H (1994) A-translation and looping combinators in pure type systems. J Funct Porgr 4(1):77–88
Dosen K (2003) Identity of proofs based on normalization and generality. J Symb Log 9(4):477–503
Gentzen G (1964) Investigations into logical deduction (1934). Am Philos Q 1(4):288–306
Geuvers H, Verkoelen J (2009) On fixed point and looping combinators in type theory. http://www.cs.ru.nl/~herman/PUBS/TLCApaper.pdf
Girard J-Y (1976) Three-valued logic and cut-elimination: the actual meaning of Takeuti’s conjecture. Dissertationes mathematicae
Girard J-Y (1987) Linear logic. Theor Comput Sci 50(1):1–102
Girard J-Y (1989) Geometry of interaction I: interpretation of system F. In: Ferro R, Bonotto C, Valentini S, Zanardo A (eds) Logic colloquium, Amsterdam
Girard J-Y (1990) Geometry of interaction II: deadlock-free algorithms. In: International conference on computational logic, Tallinn
Girard J-Y, Scedrov A, Scott PJ (1992) Normal forms and cut-free proofs as natural transformations. In: Moschovakis Y (ed) Logic from computer science, vol 21. Springer, Berlin, pp 217–241
Gödel K (1958) Über eine bisher noch nicht benütze Erweiterung des finiten Standpunktes. Dialectica 12:280–287
Hurkens AJC (1995) A simplification of Girard’s paradox. In: TLCA 95, second international conference on typed lambda calculi and applications, pp 66–278
Kelly GM, MacLane S (1971) Coherence in closed categories. J Pure Appl Algebr 1(1):97–140
Kleene SC (1945) On the interpretation of intuitionistic number theory. J Symb Log 10(4):109–124
Kreisel G (1960) Foundations of intuitionistic logic. In: Proceedings of the 1960 international congress on logic, methodology and philosophy of science. Stanford University Press, pp 98–210
Kreisel G, Takeuti G (1974) Formally self-referential propositions for cut free analysis and related systems. Dissertationes mathematicae 118:1–55
Lamarche F (2008) Proof nets for intuitionistic linear logic: essential nets. Technical report,<inria-00347336>, INRIA
Melliès P-A (2006) Functorial boxes in string diagrams. In: Computer science logic. CSL 2006. Lecture Notes in Computer Science, vol 4027. Springer, Berlin, pp 1–30
Mendler PF (1987) Inductive definitions in type theory. PhD thesis, Cornell University
Negri S, von Plato J (2001) Structural proof theory. Cambridge University Press, Cambridge
Petrolo M, Pistone P (2017) A normal paradox. In: Arazim P, Lavicka T (eds) The logical yearbook 2016. College Publications, London
Prawitz D (1965) Natural deduction, a proof-theoretical study. Almqvist & Wiskell, Stockholm
Prawitz D (1971) Ideas and results in proof theory. In: Fenstad J (ed) Proceedings of the 2nd Scandinavian logic symposium (Oslo), Studies in logic and foundations of mathematics, vol 63. North-Holland
Schroeder-Heister P (2006) Validity concepts in proof-theoretic semantics. Synthese 148:525–571
Schroeder-Heister P, Tranchini L (2017) Ekman’s paradox. Notre Dame J Form Log 58(4):567–581
Scott D (1970) Outline of a mathematical theory of computation. Technical report, Oxford University Computing Laboratory
Scott D (1976) Data Types as Lattices. SIAM J Comput 5(3):522–587
Scott D, Strachey C (1971) Toward a mathematical semantics for computer languages. Technical report, Oxford Programming Research Group Technical Monograph. PRG-6
Sorensen MH, Urzyczyn P (2006) Lectures on the Curry-Howard isomorphism, Studies in logic and the foundations of mathematics, vol 149. Elsevier, Amsterdam
Tennant N (1982) Proof and paradox. Dialectica 36:265–296
Tennant N (1995) On paradox without self-reference. Analysis 55:199–207
Tennant N (2016) Normalizability, cut eliminability and paradox. Synthese. https://doi.org/10.1007/s11229-016-1119-8.
Yablo S (1993) Paradox without self-reference. Analysis 4:251–252
Acknowledgements
The funding was provided by Agence Nationale de la Recherche (Grant No. ANR-14-FRAL-0002).
Author information
Authors and Affiliations
Corresponding author
Additional information
This work has been carried out as part of the ANR-DFG Project “Beyond Logic” (ANR-14-FRAL-0002). We wish to thank Luca Tranchini, Peter Schroeder-Heister, Volker Halbach and Marianna Antonutti Marfori for fruitful comments and an anonymous referee of the TOPOI journal for helpful suggestions and remarks on a first version of the paper.
Rights and permissions
About this article
Cite this article
Petrolo, M., Pistone, P. On Paradoxes in Normal Form. Topoi 38, 605–617 (2019). https://doi.org/10.1007/s11245-018-9543-7
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11245-018-9543-7