Abstract
Quine often argued for a simple, untyped system of logic rather than the typed systems that were championed by Russell and Carnap, among others. He claimed that nothing important would be lost by eliminating sorts, and the result would be additional simplicity and elegance. In support of this claim, Quine conjectured that every many-sorted theory is equivalent to a single-sorted theory. We make this conjecture precise, and prove that it is true, at least according to one reasonable notion of theoretical equivalence. Our clarification of Quine’s conjecture, however, exposes the shortcomings of his argument against many-sorted logic.
Similar content being viewed by others
Notes
See in particular Quine (1951, pp. 69–71), Quine (1960, pp. 209–210), and Quine (1963, pp. 267–268). He explains the conjecture as follows: “We can always reduce multiple sorts of variables to one sort if we adopt appropriate predicates. Wherever we might have used a special sort of variable we may use instead a general variable and restrict it to the appropriate predicate” (Quine 1963, p. 268). Quine (1969, p. 92) expresses the same idea when he remarks that “notations with one style of variables and notations with many are intertranslatable.” Quine (1937, 1938, 1956) provides support for the conjecture by describing a method of “translating” between many-sorted and single-sorted logic and applying it to NBG set theory and Russell’s theory of types. Of course, the fact that sorts can be eliminated (or, better, unified) was also discussed in the pure logic literature. For example, see Schmidt (1951).
See Quine (1975), Sklar (1982), Halvorson (2012, 2013, 2015), Glymour (2013), Fraassen (2014), and Coffey (2014) for general discussion of theoretical equivalence in philosophy of science. See Glymour (1977), North (2009), Swanson and Halvorson (2012), Curiel (2014), Knox (2011), Knox (2014), Barrett (2015), Weatherall (2015a, b, c), and Rosenstock et al. (2015) for discussion of whether or not particular physical theories should be considered theoretically equivalent. Finally, see de Bouvére (1965), Kanger (1968), Pinter (1978), Pelletier and Urquhart (2003), Andréka et al. (2005), Friedman and Visser (2014), and Barrett and Halvorson (2015a, b) for some results that have been proven about varieties of theoretical equivalence.
We will use the notation \(\exists _{\sigma =n}x\,\phi (x)\) throughout to abbreviate the sentence “there exist exactly n things of sort \(\sigma \) that are \(\phi \).”
Quine (1975) proposed his own criteria for equivalence of theories. This criterion suffers from some serious shortcomings, however, so we will not discuss it here (Barrett and Halvorson 2015a). Definitional and Morita equivalence each capture a sense in which two theories are intertranslatable (Barrett and Halvorson 2015a, b). Theories that are equivalent according to these criteria, therefore, can be “mutually interpreted” into one another. One has good reason to prefer these stricter notions of equivalence over mutually interpretability, however, because the latter does not preserve various important theoretical properties. See Visser (2015) for examples of some of these properties, Barrett and Halvorson (2015b) for an introduction to Morita equivalence, and Andréka et al. (2008) and Mere and Veloso (1992) for discussion of closely related ideas.
The model \(A^+|_\Sigma \) is the \(\Sigma \)-structure obtained from the \(\Sigma ^+\)-structure \(A^+\) by “forgetting” the interpretations of symbols in \(\Sigma ^+-\Sigma \). One can show that the model \(A^+\) exists and is unique up to isomorphism (Barrett and Halvorson 2015b, Theorem 4.1).
This construction recalls the proof that every theory is definitionally equivalent to a theory that uses only predicate symbols (Barrett and Halvorson 2015a, Prop. 2). Quine (1937, 1938, 1956, 1963) suggests the basic idea behind our proof, as do Burgess (2005, p. 12) and Manzano (1996, pp. 221–222). The theorem that we prove here is much more general than Quine’s results because we make no assumption about what the theory T is, whereas Quine only considered Russell’s theory of types and NBG set theory.
Note that if there were infinitely many sort symbols in \(\Sigma \), then we could not define the \({\widehat{\Sigma }}\)-sentence \(\phi \) in this way.
Intuitively, two translations F and G between theories are “almost inverse” if both \(F\circ G\) and \(G\circ F\) map every formula to a provably equivalent (but not necessarily equal) formula. The reader is invited to consult Barrett and Halvorson (2015a) for a precise definition. See Friedman and Visser (2014) for alternative characterizations of definitional equivalence and intertranslatability.
Note that if we remember that \(T_1\) and \(T_2\) use different sort symbols (as we did in the example above), then we can recognize that they have a common conservative extension. In this case, \(T_1\) implies that “there is one thing of sort \(\sigma _1\)”, while \(T_2\) implies that “there are two things of sort \(\sigma _2\)”, and these implications no longer contradict one another.
Manzano (1996, pp. 221–222) puts our attitude nicely: “It is well-known that many-sorted logic reduces to [single-sorted] logic, and this approach is the one commonly used in textbooks. [...] What is not usually said in textbooks is that the reduction has a price.”
See our definition of the sentence \(\phi _p\) and the surrounding discussion in the construction of the theory \(\hat{T}\).
Of course, there are exceptions, such as Turner (2010).
References
Andréka, H., Madaraśz, J., & Németi, I. (2008). Defining new universes in many-sorted logic (Vol. 93). Budapest: Mathematical Institute of the Hungarian Academy of Sciences.
Andréka, H., Madarász, J. X., & Németi, I. (2005). Mutual definability does not imply definitional equivalence, a simple example. Mathematical Logic Quarterly, 51(6), 591–597.
Barrett, T. W. (2015). On the structure of classical mechanics. The British Journal for the Philosophy of Science, 66(4), 801–828.
Barrett, T. W., & Halvorson, H. (2015a). Glymour and Quine on theoretical equivalence. Journal of Philosophical Logic.
Barrett, T. W., & Halvorson, H. (2015b). Morita equivalence. arXiv preprint.
Barrett, T. W., & Halvorson, H. (2016). From geometry to conceptual relativity.
Borceux, F. (1994). Handbook of categorical algebra (Vol. 1). Cambridge: Cambridge University Press.
Burgess, J. P. (2005). Fixing frege. Princeton: Princeton University Press.
Coffey, K. (2014). Theoretical equivalence as interpretative equivalence. The British Journal for the Philosophy of Science, 65(4), 821–844.
Curiel, E. (2014). Classical mechanics is Lagrangian; it is not Hamiltonian. The British Journal for the Philosophy of Science, 65(2), 269–321.
de Bouvére, K. L. (1965). Synonymous theories. In: Symposium on the Theory of Models, pp. 402–406. Amsterdam: North-Holland Publishing Company.
Friedman, H. M., & Visser, A. (2014). When bi-interpretability implies synonymy. Logic Group Preprint Series, 320, 1–19.
Glymour, C. (1971). Theoretical realism and theoretical equivalence. In PSA 1970, pp. 275–288. Berlin: Springer.
Glymour, C. (1977). The epistemology of geometry. Noûs, 11, 227–251.
Glymour, C. (1980). Theory and evidence. Princeton: Princeton University Press.
Glymour, C. (2013). Theoretical equivalence and the semantic view of theories. Philosophy of Science, 80(2), 286–297.
Halvorson, H. (2012). What scientific theories could not be. Philosophy of Science, 79(2), 183–206.
Halvorson, H. (2013). The semantic view, if plausible, is syntactic. Philosophy of Science, 80(3), 475–478.
Halvorson, H. (2015). Scientific theories. In Oxford Handbooks Online.
Hodges, W. (2008). Model theory. Cambridge: Cambridge University Press.
Kanger, S. (1968). Equivalent theories. Theoria, 34(1), 1–6.
Knox, E. (2011). Newton-Cartan theory and teleparallel gravity: The force of a formulation. Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics, 42(4), 264–275.
Knox, E. (2014). Newtonian spacetime structure in light of the equivalence principle. The British Journal for the Philosophy of Science, 65(4), 863–880.
Mac Lane, S. (1971). Categories for the working mathematician. New York: Springer.
Manzano, M. (1996). Extensions of first order logic. Cambridge: cambridge University Press.
Mere, M. C., & Veloso, P. (1992). On extensions by sorts. Monografias em Ciências da Computaçao, DI, PUC-Rio, 38, 92.
North, J. (2009). The ‘structure’ of physics: A case study. The Journal of Philosophy, 106, 57–88.
Pelletier, F. J., & Urquhart, A. (2003). Synonymous logics. Journal of Philosophical Logic, 32(3), 259–285.
Pinter, C. C. (1978). Properties preserved under definitional equivalence and interpretations. Mathematical Logic Quarterly, 24(31—-36), 481–488.
Quine, W. V. O. (1937). New foundations for mathematical logic. The American Mathematical Monthly, 44(2), 70–80.
Quine, W. V. O. (1938). On the theory of types. The Journal of Symbolic Logic, 3(04), 125–139.
Quine, W. V. O. (1951). On Carnap’s views on ontology. Philosophical Studies, 2(5), 65–72.
Quine, W. V. O. (1956). Unification of universes in set theory. The Journal of Symbolic Logic, 21(03), 267–279.
Quine, W. V. O. (1960). Word and object. Cambridge: MIT Press.
Quine, W. V. O. (1963). Set theory and its logic. Cambridge: Harvard University Press.
Quine, W. V. O. (1969). Existence and quantification. In Ontological relativity and other essays (pp. 91–113). New York: Columbia University Press.
Quine, W. V. O. (1975). On empirically equivalent systems of the world. Erkenntnis, 9(3), 313–328.
Quine, W. V. O., & Carnap, R. (1990). Dear Carnap, Dear Van: The Quine-Carnap correspondence and related work. Berkeley, CA: University of California Press.
Rosenstock, S., Barrett, T. W., & Weatherall, J. O. (2015). On Einstein algebras and relativistic spacetimes. Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics, 52, 309–316.
Schmidt, A. (1951). Die Zulässigkeit der Behandlung mehrsortiger Theorien mittels der üblichen einsortigen Prädikatenlogik. Mathematische Annalen, 123(1), 187–200.
Sklar, L. (1982). Saving the noumena. Philosophical Topics, 13, 89–110.
Swanson, N., & Halvorson, H. (2012). On North’s ‘The structure of physics’.
Turner, J. (2010). Ontological pluralism. The Journal of Philosophy, 107(1), 5–34.
Van Fraassen, B. C. (2014). One or two gentle remarks about Hans Halvorson’s critique of the semantic view. Philosophy of Science, 81(2), 276–283.
Visser, A. (2015). On Q. In Logic Group Preprint Series, p. 330.
Weatherall, J. O. (2015a). Are Newtonian gravitation and geometrized Newtonian gravitation theoretically equivalent? Erkenntnis, 1–19.
Weatherall, J. O. (2015b). Categories and the foundations of classical field theories. In E. Landry (Ed.), Forthcoming in categories for the working philosopher. Oxford: Oxford University Press.
Weatherall, J. O. (2015c). Understanding gauge. Philosophy of Science.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
The objective of this appendix is to prove Theorem 2. We prove a special case of the result for convenience. We will assume that \(\Sigma \) has only three sort symbols \(\sigma _1, \sigma _2,\sigma _3\) and that \(\Sigma \) does not contain function or constant symbols. A perfectly analogous (though more tedious) proof goes through in the general case.
We prove the result by explicitly constructing a “common Morita extension” \(T_{4}\cong {\widehat{T}}_{4}\) of T and \({\widehat{T}}\) to the following signature.
The symbol \(\sigma _{12}\in \Sigma ^+\) is a sort symbol. The symbols denoted by subscripted \(\rho \) are function symbols. Their arities are expressed in the following figure.
The symbols \(i_1\), \(i_2\), and \(i_3\) are function symbols with arity \(\sigma _1\rightarrow \sigma \), \(\sigma _2\rightarrow \sigma \), and \(\sigma _3\rightarrow \sigma \), respectively.
We now turn to the proof.
Proof of Theorem 2
The following figure illustrates how our proof will be organized.
Steps 1–3 define the theories \({\widehat{T}}_1,\ldots , {\widehat{T}}_{4}\), steps 4–6 define \(T_1,\ldots , T_{4}\), and step 7 shows that \(T_{4}\) and \({\widehat{T}}_{4}\) are logically equivalent.
-
Step 1 We begin by defining the theory \({\widehat{T}}_1\). For each sort \(\sigma _j\in \Sigma \) we consider the following sentence.
The sentence \(\theta _{\sigma _j}\) defines the symbols \(\sigma _j\) and \(i_j\) as the subsort of “things that are \(q_{\sigma _j}\).” The auxiliary axioms \(\phi _{\sigma _j}\) of \({\widehat{T}}\) guarantee that the admissibility conditions for these definitions are satisfied. The theory \({\widehat{T}}_1={\widehat{T}}\cup \{\theta _{\sigma _1}, \theta _{\sigma _2},\theta _{\sigma _3}\}\) is therefore a Morita extension of \({\widehat{T}}\) to the signature \({\widehat{\Sigma }}\cup \{\sigma _1,\sigma _2,\sigma _3, i_1,i_2,i_3\}\).
-
Step 2 We now define the theories \({\widehat{T}}_2\) and \({\widehat{T}}_3\). Let \(\theta _{\sigma _{12}}\) be a sentence that defines the symbols \(\sigma _{12}, \rho _1, \rho _2\) as a coproduct sort. The theory \({\widehat{T}}_2={\widehat{T}}_1\cup \{\theta _{\sigma _{12}}\}\) is clearly a Morita extension of \({\widehat{T}}_1\).
We have yet to define the function symbols \(\rho _{12}\) and \(\rho _3\). The following two sentences define these symbols.
The sentence \(\theta _{\rho _3}\) simply defines \(\rho _3\) to be equal to the function \(i_3\). For the sentence \(\theta _{\rho _{12}}\), we define the formula \(\psi (x,y)\) to be
$$\begin{aligned} \exists _{\sigma _{1}} z_1 \big ( \rho _{1}(z_1)=x\wedge i_{1}(z_1)=y\big ) \vee \exists _{\sigma _{2}} z_2\big ( \rho _{2}(z_2)=x\wedge i_{2}(z_2)=y\big ) \end{aligned}$$We should take a moment here to understand the definition \(\theta _{\rho _{12}}\). We want to define what the function \(\rho _{12}\) does to an element a of sort \(\sigma _{12}\). Since the sort \(\sigma _{12}\) is the coproduct of the sorts \(\sigma _1\) and \(\sigma _2\), the element a must “actually be” of one of the sorts \(\sigma _1\) or \(\sigma _2\). (The disjuncts in the formula \(\psi (x,y)\) correspond to these possibilities.) The definition \(\theta _{\rho _{12}}\) stipulates that if a is “actually” of sort \(\sigma _j\), then the value of \(\rho _{12}\) at a is the same as the value of \(i_j\) at a. One can verify that \({\widehat{T}}_{2}\) satisfies the admissibility conditions for \(\theta _{\rho _3}\) and \(\theta _{\rho _{12}}\), so the theory \({\widehat{T}}_3={\widehat{T}}_{2}\cup \{\theta _{\rho _3}, \theta _{\rho _{12}}\}\) is a Morita extension of \({\widehat{T}}_{2}\) to the signature
$$\begin{aligned} {\widehat{\Sigma }}\cup \{\sigma _1,\sigma _2, \sigma _3, \sigma _{12}, i_1,i_2, i_3, \rho _1, \rho _2,\rho _3, \rho _{12}\} \end{aligned}$$ -
Step 3 We now describe the \(\Sigma ^+\)-theory \({\widehat{T}}_{4}\). This theory defines the predicates in the signature \(\Sigma \). Let \(p\in \Sigma \) be a predicate symbol of arity \(\sigma _{j_1}\times \cdots \times \sigma _{j_m}\). We consider the following sentence.
The theory \({\widehat{T}}_{4}={\widehat{T}}_3\cup \{\theta _{p}:p\in \Sigma \}\) is therefore a Morita extension of \({\widehat{T}}_3\) to the signature \(\Sigma ^+\).
-
Step 4 We turn to the left-hand side of our organizational figure and define the theories \(T_1\) and \(T_{2}\). We proceed in an analogous manner to the first part of Step 2. The theory \(T_1=T\cup \{\theta _{\sigma _{12}}\}\) is a Morita extension of T to the signature \(\Sigma \cup \{\sigma _{12}, \rho _1,\rho _2\}\). Now let \(\theta _\sigma \) be the sentence that defines the symbols \(\sigma , \rho _{12}, \rho _3\) as a coproduct sort. The theory \(T_{2}=T_{1}\cup \{\theta _{\sigma }\}\) is a Morita extension of \(T_{1}\) to the signature \(\Sigma \cup \{\sigma _{12}, \sigma , \rho _1, \rho _2, \rho _3, \rho _{12}\}\).
-
Step 5 This step defines the function symbols \(i_1\), \(i_2\), and \(i_3\). We consider the following sentences.
The sentence \(\theta _{i_3}\) defines the function symbol \(i_3\) to be equal to \(\rho _3\). The sentence \(\theta _{i_{2}}\) defines the function symbol \(i_{2}\) to be equal to the composition “\(\rho _{12}\circ \rho _{2}\).” Likewise, the sentence \(\theta _{i_{1}}\) defines the function symbol \(i_{1}\) to be “\(\rho _{12}\circ \rho _{1}\).” The theory \(T_3=T_{2}\cup \{\theta _{i_1},\theta _{i_2}, \theta _{i_3}\}\) is a Morita extension of \(T_{2}\) to the signature \(\Sigma \cup \{\sigma _{12}, \sigma , \rho _1, \rho _2, \rho _3, \rho _{12}, i_1, i_2, i_3\}\).
-
Step 6 We still need to define the predicate symbols in \({\widehat{\Sigma }}\). Let \(\sigma _j\in \Sigma \) be a sort symbol and \(p\in \Sigma \) a predicate symbol of arity \(\sigma _{j_1}\times \cdots \times \sigma _{j_m}\). We consider the following sentences.
These sentences define the predicates \(q_{\sigma _j}\in {\widehat{\Sigma }}\) and \(q_p\in {\widehat{\Sigma }}\). One can verify that \(T_3\) satisfies the admissibility conditions for the definitions \(\theta _{q_{\sigma _j}}\). And therefore the theory \(T_{4}=T_3\cup \{\theta _{q_{\sigma _1}}, \theta _{q_{\sigma _2}}, \theta _{q_{\sigma _3}}\}\cup \{\theta _{q_{p}}: p\in \Sigma \}\) is a Morita extension of \(T_3\) to the signature \(\Sigma ^+\).
-
Step 7 It only remains to show that the \(\Sigma ^+\)-theories \(T_{4}\) and \({\widehat{T}}_{4}\) are logically equivalent. One can verify by induction on the complexity of \(\psi \) that
$$\begin{aligned} T_{4}\vDash \psi \leftrightarrow {\widehat{\psi }}\text { and } {\widehat{T}}_{4}\vDash \psi \leftrightarrow {\widehat{\psi }}. \end{aligned}$$(5)for every \(\Sigma \)-sentence \(\psi \). One then uses (5) to show that \(\text {Mod}(T_{4})=\text {Mod}({\widehat{T}}_{4})\). The argument involves a number of cases, but since each case is straightforward we leave them to the reader to verify. The theories \(T_{4}\) and \({\widehat{T}}_{4}\) are logically equivalent, which implies that T and \({\widehat{T}}\) are Morita equivalent. \(\square \)
Rights and permissions
About this article
Cite this article
Barrett, T.W., Halvorson, H. Quine’s conjecture on many-sorted logic. Synthese 194, 3563–3582 (2017). https://doi.org/10.1007/s11229-016-1107-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11229-016-1107-z