Skip to main content
Log in

Continuity postulates and solvability axioms in economic theory and in mathematical psychology: a consolidation of the theory of individual choice

  • Published:
Theory and Decision Aims and scope Submit manuscript

Abstract

This paper presents four theorems that connect continuity postulates in mathematical economics to solvability axioms in mathematical psychology, and ranks them under alternative supplementary assumptions. Theorem 1 connects notions of continuity (full, separate, Wold, weak Wold, Archimedean, mixture) with those of solvability (restricted, unrestricted) under the completeness and transitivity of a binary relation. Theorem 2 uses the primitive notion of a separately continuous function to answer the question when an analogous property on a relation is fully continuous. Theorem 3 provides a portmanteau theorem on the equivalence between restricted solvability and various notions of continuity under weak monotonicity. Finally, Theorem 4 presents a variant of Theorem 3 that follows Theorem 1 in dispensing with the dimensionality requirement and in providing partial equivalences between solvability and continuity notions. These theorems are motivated for their potential use in representation theorems.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1

Similar content being viewed by others

Notes

  1. See Gonzales (1997, 2000, 2003) for details.

  2. We refer the reader to Köbberling and Wakker (2003) for a comprehensive introduction to the axioms of continuity and solvability across the topological and the algebraic registers.

  3. See Krantz, Luce, Suppes, and Tversky (1971, Chapter 1) for details as to what the subject regards as “fundamental quantities”.

  4. See Michell and Ernst (1996) for the translated version of Part 1 of Hölder’s text. Also see (Birkhoff 1967, p. 300), (Fuchs 1963, p. 45), (Krantz et al. 1971, p. 54) and the comprehensive text Moscati (2016); especially Chapters 8 and 15 on Stevens and Suppes respectively therein.

  5. In this paper, we do not investigate the behavioral implications of continuity and solvability. See Khan and Uyanik (2021) for a two-way relationship between connectedness and behavioral assumptions on preferences.

  6. Wakker and Yang (2019, 2021) provide powerful tools for analyzing concavity and convexity of utility and weighting functions by using the convexity of preferences without hinging on the continuity assumption.

  7. We thank an anonymous referee for emphasising this point.

  8. If \(n=3\) and \(A=\{1,2\}\), then for \(x=(x_{1},x_{2},x_{3})\in {\mathbb {R}}^{3}_{++},\) \(x_{A}=(x_{1},x_{2})\) and \(x_{\lnot A}=(x_{3})\) are the restriction of x to A and \(\lnot A,\) respectively. Furthermore, \((x_{-A},\alpha ),\) denotes an act where \(\alpha\) is placed on all coordinates in A. See (Page 94, Wakker (1989)) for further details.

  9. Note \(\succ\) is transitive; see Khan and Uyanik (2021) for details on the relationship between different transitivity postulates.

References

  • Aumann, R. J. (1966). Existence of competitive equilibria in markets with a continuum of traders, Econometrica, pp. 1–17.

  • Birkhoff, G. (1967). Lattice Theory (3rd Ed.), vol. XXV. New York: American Mathematical Society.

  • Chew, S. H., & Karni, E. (1994). Choquet expected utility with a finite state space: Commutativity and act-independence. Journal of Economic Theory, 62(2), 469–479.

    Article  Google Scholar 

  • Ciesielski, K. C., & Miller, D. (2016). A continuous tale on continuous and separately continuous functions. Real Analysis Exchange, 41(1), 19–54.

    Article  Google Scholar 

  • Easley, D., & Kleinberg, J. (2012). Networks, crowds, and markets: Reasoning about a highly connected world. Significance, 9, 43–44.

    Google Scholar 

  • Fleischer, I. (1961). Numerical representation of utility. Journal of the Society for Industrial and Applied Mathematics, 9(1), 48–50.

    Article  Google Scholar 

  • Fuchs, L. (1963). Partially Ordered Algebraic Systems. Pergamon Press.

  • Genocchi, A., & Peano, G. (1884). Calcolo Differentiale e Principii di Calcolo. Torino: Fratelli Bocca.

    Google Scholar 

  • Giarlotta, A., & Watson, S. (2020). A bi-preference interplay between transitivity and completeness: Reformulating and extending Schmeidler’s theorem. Journal of Mathematical Psychology, 96, 102354.

    Article  Google Scholar 

  • Gonzales, C. (1997). Additive utility without solvability on all components, In: Constructing Scalar-Valued Objective Functions, pp. 64–90. Springer.

  • Gonzales, C. (2000). Two factor additive conjoint measurement with one solvable component. Journal of Mathematical Psychology, 44(2), 285–309.

    Article  Google Scholar 

  • Gonzales, C. (2003). Additive utility without restricted solvability on every component. Journal of Mathematical Psychology, 47(1), 47–65.

    Article  Google Scholar 

  • Herstein, I. N., & Milnor, J. (1953). An axiomatic approach to measurable utility. Econometrica, 21(2), 291–297.

    Article  Google Scholar 

  • Hölder, O. (1901). Die axiome der quantität und die lehre vom mass. Teubner.

  • Karni, E., & Safra, Z. (2015). Continuity, completeness, betweenness and cone-monotonicity. Mathematical Social Sciences, 74, 68–72.

    Article  Google Scholar 

  • Khan, M. A., & Uyanik, M. (2020). On an extension of a theorem of Eilenberg and a characterization of topological connectedness. Topology and its Applications, 273, 107–117.

    Article  Google Scholar 

  • Khan, M. A., & Uyanik, M. (2021). Topological connectedness and behavioral assumptions on preferences: a two-way relationship. Economic Theory, 71(2), 411–460.

    Article  Google Scholar 

  • Köbberling, V., & Wakker, P. P. (2003). Preference foundations for nonexpected utility: A generalized and simplified technique. Mathematics of Operations Research, 28(3), 395–423.

    Article  Google Scholar 

  • Krantz, D., Luce, D., Suppes, P., & Tversky, A. (1971). Foundations of Measurement, Volume I: Additive and Polynomial Representations. New York: Academic Press.

  • Kruse, R., & Deely, J. (1969). Joint continuity of monotonic functions. The American Mathematical Monthly, 76(1), 74–76.

    Article  Google Scholar 

  • Luce, R. D., & Narens, L. (1983). Symmetry, scale types, and generalizations of classical physical measurement. Journal of Mathematical Psychology, 27(1), 44–85.

    Article  Google Scholar 

  • Luce, R. D., & Narens, L. (1986). Measurement: The theory of numerical assignments. Psychological Bulletin, 99(2), 166–180.

    Article  Google Scholar 

  • Luce, R. D., & Tukey, J. W. (1964). Simultaneous conjoint measurement: A new type of fundamental measurement. Journal of Mathematical Psychology, 1(1), 1–27.

    Article  Google Scholar 

  • Mehta, G. (1986). Existence of an order-preserving function on normally preordered spaces. Bulletin of the Australian Mathematical Society, 34(1), 141–147.

    Article  Google Scholar 

  • Michell, J., & Ernst, C. (1996). The axioms of quantity and the theory of measurement: Translated from Part I of Otto Hölder’s German Text “Die Axiome der Quantität und die Lehre vom Mass. Journal of Mathematical Psychology, 40(3), 235–252.

  • Moscati, I. (2016). Measuring Utility. Oxford: Oxford University Press.

    Google Scholar 

  • Neuefeind, W., & Trockel, W. (1995). Continuous linear representability of binary relations. Economic Theory, 6(2), 351–356.

    Article  Google Scholar 

  • Rebenciuc, M. (2008). Binary relations - addenda 1: Kernel, restrictions and inducing, relational morphisms, UPB. Sci. Bulll. Series A, 70(3), 11–22.

    Google Scholar 

  • Segal, U., & Sobel, J. (2002). Min, max, and sum. Journal of Economic Theory, 106(1), 126–150.

    Article  Google Scholar 

  • Suppes, P. (1974). The measurement of belief. Journal of the Royal Statistical Society: Series B (Methodological), 36(2), 160–175.

    Google Scholar 

  • Uyanik, M., & Khan, M. A. (2022). The continuity postulate in economic theory: a deconstruction and an integration, Working Paper.

  • Wakker, P. (1989). Additive Representations of Preferences: A New Foundation of Decision Analysis. Boston: Kluwer Academic Publishers.

    Book  Google Scholar 

  • Wakker, P. P., & Yang, J. (2019). A powerful tool for analyzing concave/convex utility and weighting functions. Journal of Economic Theory, 181, 143–159.

    Article  Google Scholar 

  • Wakker, P.P., & Yang, J. (2021). Concave/convex weighting and utility functions for risk: A new light on classical theorems, Insurance: Mathematics and Economics, 100, 429–435.

  • Willard, S. (1970). General Topology. New York: Addison Wesley Publishing Co.

    Google Scholar 

  • Wold, H. (1943–44). A synthesis of pure demand analysis, I–III, Scandinavian Actuarial Journal, 26, 85–118, 220–263; 27, 69–120.

  • Wold, H., & Jureen, L. (1953). Demand Analysis. New York: John Wiley and Sons Inc.

    Google Scholar 

  • Young, W. (1910). A note on monotone functions. The Quarterly Journal of Pure and Applied Mathematics (Oxford Ser.), 41, 79–87.

    Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Aniruddha Ghosh.

Additional information

Publisher's Note

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

While acknowledging the pioneering works of Hugo Sonnenschein and Peter Wakker, the authors thank Yorgos Gerasimou, Alfio Giarlotta, Farhad Husseinov, Rohit Parikh, Arthur Paul Pedersen, John Rehbeck, and Eddie Schlee for stimulating conversations and correspondence. The authors also thank workshop participants at the Advances in Mathematical Economics workshop at Università degli Studi di Napoli Federico II and seminar participants of the Seminar in Philosophy, Logic and Games at CUNY. Finally, they thank an erudite referee of this journal for his/her suggestions, substantive and expositional; and also acknowledge discussions with Osama Khan way back in 2017.

Appendices

Appendix

In the Appendix, we present the proofs of the results and six technical examples.

Appendix A: Proofs of the Results

Proof of Theorem 1 (a) The implication that Wold-continuity implies weak Wold-continuity follows from their definitions. Uyanik and Khan (2022, Proposition 3) prove that continuity implies Wold-continuity for finite dimensional spaces, their arguments extend to the current setting trivially.

Next, we show that weak Wold-continuity implies restricted solvability. Towards this end assume \(\succsim\) is a weakly Wold-continuous binary relation on a convex set \(X \subseteq \prod _{i\in I}X_{i}\). Let \(y=(a_{i}, y_{-i} )\), \(z= (b_{i},y_{-i} )\) and \(y\succsim x\succsim z\). If \(y\sim x \sim z,\) or \(y\succ x \sim z\), or \(y\sim x \succ z,\) then restricted solvability trivially follows. Hence, assume \(y\succ x \succ z\). From weak Wold-continuity there exists a straight line \(y\lambda z\) intersecting the indifference class of x, i.e., \(\lambda y + (1-\lambda ) z \sim x\), where \(\lambda y + (1-\lambda ) z= (\lambda a_{i}+(1-\lambda ) b_{i}, y_{-i})\). Therefore, \(\succsim\) is restricted solvable.

It remains to show that weak Wold-continuity implies Archimedean property. Towards this end assume \(\succsim\) is Wold-continuity but not Archimedean. Then, there exists \(x,y,z\in X\) such that \(x\succ y\) but for all \(\lambda \in (0,1)\), \(x\lambda z\nsucc y\) (the proof of the case where \(x\nsucc y\lambda z\) is analogous). By completeness, \(y\succsim x\lambda z\). Pick \(\lambda \in (0,1)\). If \(y\succ x\lambda z\), then \(x\succ y\succ x\lambda z\) and weak Wold-continuity imply there exist \(\delta \in (0,1)\) such that \(y\sim x\delta z\). If \(y\sim x\lambda z\), then set \(\delta =\lambda\). Then by transitivity, \(x\succ x\delta z\). By weak Wold-continuity, there exists \(\gamma \in (0,1)\) such that \(x\succ x\gamma z\succ x\delta z\sim y\). Hence, by transitivity, \(x\gamma z\succ y\). This furnishes us a contradiction with the assumption that for all \(\lambda ' \in (0,1)\), \(x\lambda ' z\nsucc y\).

(b) Continuity implies separate continuity follows from the definitions of the two continuity postulates. For the proof of the remaining implications except mixture-continuity implies separate continuity, see Uyanik and Khan (2022, Proposition 3).

It remains to show that mixture-continuity implies separate continuity. Assume \(\succsim\) is separately continuous. Pick an index i, \(x\in X\) and a sequence \((y_i^n, y_{-i})\) converging \((y_i, y_{-i})\in X\) such that \((y_i^n, y_{-i})\succsim x\). mixture-continuity implies that the set \(A=\{\lambda \in [0,1]~|~(y_i, y_{-i})\lambda y\succsim x\}\) is closed. It is clear that \(1\in A\) and for all \(\lambda >0\), \(\lambda \in A\). Hence \(1\in A\). Therefore, The proof that for all \(x\in X\) and all lines L parallel to any coordinate axis, \(A_\succsim (x)\cap L\) is closed. The proof that for all \(x\in X\) and all lines L parallel to any coordinate axis, \(A_\precsim (x)\cap L\) is closed analogously follows. Since \(\succsim\) is complete, \(\succsim\) is separately continuous.

(c) Assume \(\succsim\) is separately continuous. Let \((a_{i}, y_{-i})\succsim x\succsim (b_{i}, y_{-i})\). Separate continuity implies that the weakly worse-than and weakly better-than sets of x are closed in the subspace of X determined by the line \(L_i\) parallel to the i-\(\text {th}\) coordinate axis and passing through \(y_{-i}\), i.e. \(\succsim _{x}\upharpoonright _{L_i}\) and \(\precsim _{x}\upharpoonright _{L_i}\) are closed in \(X\cap L_i\). Since \(X\cap L_i\) is convex, it is connected. Moreover, it follows from the completeness of the relation that the intersection of these two sets is non-empty, i.e., \(\succsim _{x}\upharpoonright _{L_i}\cap \precsim _{x}\upharpoonright _{L_i}\ne \varnothing\). Hence, there exists \({\tilde{x}}\in X\cap L_i\) such that \({\tilde{x}}\sim x\). Since \({\tilde{x}}\in X\cap L_i\), there exists \(c_i\in \mathbb {R}\) such that \({\tilde{x}}=(c_i,{\bar{y}}_{-i})\). Therefore, \(\succsim\) is restricted solvable.

(d) Assume \(\succsim\) is unrestricted solvable. And assume its is not restricted solvable. Then there exists no b such that \(bq\sim ap\) whenever \(\bar{b}q\succsim ap \succsim \underline{b}q\). But this contradicts unrestricted solvability.\(\square\)


Proof of Theorem 2 We provide the proof by induction. Note that for \(n=1\), separate continuity is equivalent to continuity by definition. Now let \(n=2\) and assume without loss of generality that \(\succsim\) is weakly monotone in its first coordinate. We now show that \(\succsim\) has closed lower sections. Towards this end, assume there exist \(x\in X\) and a sequence \(y^k\rightarrow y\) such that \(y^k\precsim x\) for all k and . Then, either \(y\succ x\) or \(y\bowtie x\).

Assume \(y\succ x\). Separate continuity implies that there exists \(\varepsilon >0\) such that for all z in the \(\varepsilon\)-neighborhood of y, \((z_1, y_2)\succ x\). Pick \((z_1, y_2)\) in the \(\varepsilon\)-neighborhood of y such that \(z_1<y_1-\varepsilon /2\). Separate continuity then implies that there exists \(\delta _1>0\) such that for all \(z'\) in the \(\delta _1\)-neighborhood of \((z_1, y_2)\), \((z_1, z'_2)\succ x\). Define \(\delta =\min \{\varepsilon /2, \delta _1\}\). Since \(y^k\rightarrow y\), there exists \({\hat{k}}\in {\mathbb {N}}\) such that \(y^{{\hat{k}}}\) is in the \(\delta\)-neighborhood of y. Then, \((z_1, y_2^{{\hat{k}}})\) is in the \(\delta _1\)-neighborhood of \((z_1, y_2)\) and \(z_1<y_1^{{\hat{k}}}\). Hence, by weak monotonicity, \(y^{{\hat{k}}}\succsim (z_1, y_2^{{\hat{k}}})\). It follows from \((z_1, y_2^{{\hat{k}}})\succ x\) and transitivity of \(\succsim\) thatFootnote 9\(y^{{\hat{k}}}\succ x\). This furnishes us a contradiction.

When \(y\bowtie x\), replacing \(\succ\) with \(\bowtie\) in the paragraph above yields \(y^{{\hat{k}}}\succsim (z_1, y_2^{{\hat{k}}})\), as above. Then, \(y^{{\hat{k}}}\succsim (z_1, y_2^{{\hat{k}}})\bowtie x\). If \(x\succsim y^{{\hat{k}}}\), then transitivity of \(\succsim\) implies that \(x \succsim (z_1, y_2^{{\hat{k}}})\), yielding a contradiction. Hence . This furnishes us a contradiction.

Therefore, \(\succsim\) has closed lower sections. The proof of the closed upper sections is analogous. Moreover, the proof of the open sections have a similar construction. Finally, completeness of \(\succsim\) follows from (2021, Theorem 2).

Now assume the theorem is true for \(n-1\) dimensional spaces, \(n>2\). We next show that it is true for n dimensional spaces. Recall that \(\succsim\) is weakly monotone in all coordinates (except possibly one). Pick a coordinate i in which \(\succsim\) is weakly monotone. We now show that \(\succsim\) has closed lower sections. Towards this end, assume there exist \(x\in X\) and a sequence \(y^k\rightarrow y\) such that \(y^k\precsim x\) for all k and . Then, either \(y\succ x\) or \(y\bowtie x\).

Assume \(y\succ x\). Separate continuity implies that there exists \(\varepsilon >0\) such that for all z in the \(\varepsilon\)-neighborhood of y, \((z_i, y_{-i})\succ x\). Pick \((z_i, y_{-i})\) in the \(\varepsilon\)-neighborhood of y such that \(z_i<y_i-\varepsilon /2\). By induction hypothesis, \(\succsim\) is continuous on \(n-1\) dimensional spaces. Hence, there exists \(\delta _i>0\) such that for all \(z'\) in the \(\delta _i\)-neighborhood of \((z_i, y_{-i})\), \((z_i, z'_{-i})\succ x\). Define \(\delta =\min \{\varepsilon /2, \delta _i\}\). Since \(y^k\rightarrow y\), there exists \({\hat{k}}\in {\mathbb {N}}\) such that \(y^{{\hat{k}}}\) is in the \(\delta\)-neighborhood of y. Then, \((z_i, y_{-i}^{{\hat{k}}})\) is in the \(\delta _i\)-neighborhood of \((z_i, y_{-i})\) and \(z_i<y_i^{{\hat{k}}}\). Hence, by weak monotonicity, \(y^{{\hat{k}}}\succsim (z_i, y_{-i}^{{\hat{k}}})\). It follows from \((z_i, y_{-i}^{{\hat{k}}})\succ x\) and transitivity of \(\succsim\) that \(y^{{\hat{k}}}\succ x\). This furnishes us a contradiction.

The remaining part of the proof is analogous to the case \(n=2\) and uses the modification of the construction above for general n, hence omitted.\(\square\)


Proof of Theorem 3 Let \(\succsim\) be a complete, transitive, order dense and weakly monotonic binary relation on a convex and order-bounded set \(X \subseteq {\mathbb {R}}^n\). Theorems 1 and 2 prove all the relationships except that restricted solvability implies separate continuity. Towards this end, assume \(\succsim\) is restricted solvable but not separately continuous.

Assume there exists \(x\in X\) and a line L parallel to a coordinate axis such that \(A_\succsim (x)\cap L\) is not closed. Then, there exist \(x\in X\), an index i, and a sequence \(y^k \rightarrow y\) on the line \(L_i\) parallel to coordinate i such that \(y^k\succsim x\) for all k and \(y\prec x\). Then it follows from transitivity and weak monotonicity assumptions that \(y_i^k> y_i\) for all k. Pick \(m\in {\mathbb {N}}\). Since \(y\prec x\), order denseness implies that there exists \(x'\in X\) such that \(y\prec x'\prec x\). Then transitivity implies \(y\prec x'\prec y^m\), and hence restricted solvability implies there exists \(z\in L_i\) such that \(z\sim x'\). It follows from weak monotonicity that \(y_i<z_i<y^m_i\), and from transitivity that for all \(z_i'\in (y_i, z_i]\), \((z'_i, z_{-i})\prec x\). However, since \(y^k\rightarrow y\), there exists \(z_i'\in (y_i, z_i]\), such that \(z_i'=y_i^k\) for some k and \((z'_i, z_{-i})\succsim x\). This yields a contradiction.

The proof that for all \(x\in X\) and all lines L parallel to any coordinate axis, \(A_\precsim (x)\cap L\) is closed analogously follows. Since \(\succsim\) is complete, \(\succsim\) is separately continuous.\(\square\)


Proof of Theorem 4 We first show that Archimedean postulate implies mixture-continuity. Assume there exist \(x,y,z\in X\) such that \(\{\lambda : x\lambda y\succsim z\}\) is not closed. Then there exists \(\lambda ^n\rightarrow \lambda\) such that \(x\lambda ^n y\succsim z\) but \(z\succ x\lambda y\). By strong order-boundedness assumption, there exists \(\varepsilon >0\) such that \(p=((x\lambda y)_i +\varepsilon )_{i\in I}\in X\). By weak monotonicity, \(p\succsim p\delta (x\lambda y)\) for all \(\delta \in [0,1]\). By Archimedean, there exists \(\beta \in (0,1)\) such that \(z\succ p\beta (x\lambda y)\). Note that there exists \(\epsilon \in (0,\varepsilon )\) such that \((p\beta (x\lambda y))_i - (x\lambda y)_i > \epsilon\) for all \(i\in I\). Therefore, there exists \(N\in {\mathbb {N}}\) such that for all \(n\ge N\), \((x\lambda ^n y)_i < (p\beta (x\lambda y))_i\) for all \(i\in I\). By weak monotonicity, \(p\beta (x\lambda y)\succsim x\lambda ^n y\) for all \(n\ge N\). By transitivity, \(z\succ x\lambda ^n y\) for all \(n\ge N\). This yields a contradiction. Closedness of the lower sections follows analogously.

By the figure above, we now prove the equivalence of mixture continuity, Archimedean and weak Wold-continuity. The proof that restricted solvability implies separate continuity follows from the argument in the proof of Theorem 4 since it does not hinge on the dimension of the underlying space.

Appendix B: Examples

In this section we show that in Figure 1, the converse relationships among the continuity postulates are false. All the binary relations presented here satisfy completeness and transitivity.

The first example shows that continuity, and hence all other continuity postulates, do not imply unrestricted solvability even under weakly monotone and convex preferences.

1. Let \(\succsim\) be a binary relation defined on \({\mathbb {R}}^{2}\) as \((x_{1},x_{2})\succsim (y_{1},y_{2})\) if and only if \(f(x_{1},x_{2})\ge f(y_{1},y_{2})\) where \(f(x_{1},x_{2})=x_{2}.\) To see that \(\succsim\) does not satisfy unrestricted solvability, for all \(x_1\in \mathbb {R}\), \((2,2)\succ (x_1,1)\), hence \(\succsim\) is not unrestricted solvable in the second component. It is clear that \(\succsim\) is continuous, hence it satisfies the remaining six continuity postulates, following from Theorem 1. \(\square\)

The second example shows that unrestricted solvability and restricted solvability do not imply separate continuity, Archimedean and weak Wold-continuity, and hence any of the other continuity postulates.

2. Let the binary relation \(\succsim\) be defined on a bounded subset \(X\subset {\mathbb {R}}^{2}\) as \((x_{1},x_{2})\succsim (y_{1},y_{2})\) if and only if \(f(x_{1},x_{2})\ge f(y_{1},y_{2})\) where

$$\begin{aligned} f(x_{1},x_{2}) = {\left\{ \begin{array}{ll} 0, &{} \text { if } x_{1}+x_{2}< 1 \\ 0.8, &{} \text { if }x_{1}+x_{2}= 1 \\ 1, &{} \text { if } x_{1}+x_{2}> 1 \end{array}\right. } \end{aligned}$$

To see why this is not separately continuous and hence not continuous, pick \((x_{1},x_{2})=(1,0).\) For the binary relation to be separately continuous, the restriction of (1, 0) to any line parallel to either \(x_{1}\) or \(x_{2}\) axis should be continuous. Let \(x_{2}=0.5.\) Then \(\succsim _{(1,0)}\upharpoonright _{x_{2}=0.5}=(0.5,b]\) where b is some bound on \(x_{1}.\) This representation is neither Wold nor weak Wold-continuous (does not satisfy order-denseness). However, this binary relation is both restricted and unrestricted solvable.

This is not Archimedean. Pick \(x=(0.6,0.4)\) and \(y=z=(0.6,0.5).\) Then \(y\succ x\) but \(y\sim x\delta z\) for any \(\delta \in (0,1).\) \(\square\)

The third example shows that separate continuity does not imply weak Wold-continuity, Archimedean and unrestricted solvability, hence does not imply mixture-continuity, Wold-continuity and continuity postulates. It also shows that restricted solvability does not imply any of these continuity postulates.

3. Let the binary relation \(\succsim\) be defined on \({\mathbb {R}}^{2}\) as \((x_{1},x_{2})\succsim (y_{1},y_{2})\) if and only if \(f(x_{1},x_{2})\ge f(y_{1},y_{2})\) where

$$\begin{aligned} f(x_{1},x_{2}) = {\left\{ \begin{array}{ll} \dfrac{x_{1}x_{2}}{x_{1}^{2}+x_{2}^{2}} &{} \text { if } (x_{1},x_{2})\ne (0,0), \\ 0 &{} \text { if } (x_{1},x_{2}) = (0,0). \\ \end{array}\right. } \end{aligned}$$

In this example, we have, \((1,1)\succ (3,1)\succ (0,0).\) There exists no \(\lambda \in [0,1]\) such that \(\lambda (1,1)+(1-\lambda )(0,0) \sim (3,1).\) Hence, this is not Weak Wold-continuous. Consequently, this is also not Wold-continuous.

It is easy to show that this binary relation is not unrestricted solvable. Pick a pair (1, 1) and let the third element of the to-be-determined pair is 0. Then unrestricted solvability claims that there exists q such that \((1,1)\sim (q,0).\) In this case, there exists no q such that the indifference holds.

Finally this example illustrates that a binary relation that is restricted solvable, but not Archimedean. The function is continuous on any line parallel to a coordinate axis, hence the induced binary relation is restricted solvable. However, it is not continuous on the 45-degree line. In particular it is defined on \(\mathbb {R}^2_+\) and the function’s value is 1 on all point on the 45-degree line except 0, and at 0 its value is 0. Therefore if you pick \(x=(0,0), y=z=(1,1)\), then \(y\succ x\) but for all \(\lambda \in (0,1)\), \(y\sim x\lambda z\), hence Archimedean property fails. Similarly, it also shows that Archimedean does not imply mixture-continuity. \(\square\)

The fourth example illustrates a binary relation that is Archimedean and Wold-continuous, and hence weakly Wold-continuous and restricted solvable, but not mixture-continuous, separately continuous and continuous. It is originally presented in Uyanik and Khan (2022).

4. Let \(X=\mathbb {R}_+\) and \(f(x)=\text {sin}(1/x)\) if \(x>0\) and \(f(0)=1\). Then, it is clear that \(f(x)\in [-1,1]\) for all \(x\in X\). Define a binary relation \(\succsim\) on X as \(x\succsim y\) is and only if \(f(x)\ge f(y)\). Pick \({\bar{x}}\) such that \(f({\bar{x}})\in (0,1)\). The set \(\{x'\in X|{\bar{x}}\succsim x'\}\) is not closed since it contains a sequence \(x_n\rightarrow 0\) but \(0\succ {\bar{x}}\). Therefore, \(\succsim\) is not continuous. Moreover, since X is one dimensional, \(\succsim\) is not mixture-continuous and separately continuous. See Uyanik and Khan (2022, Proof of Proposition 3) for a proof that \(\succsim\) satisfies Wold-continuity and Archimedean properties. \(\square\)

The fifth example shows that Archimedean property does not imply restricted solvability, hence all other continuity postulates in Figure 1.

5. Let \(\succsim\) be a binary relation defined on \({\mathbb {R}}^{2}_+\) as \((x_{1},x_{2})\succsim (y_{1},y_{2})\) if and only if \(f(x_{1},x_{2})\ge f(y_{1},y_{2})\) where

$$\begin{aligned} f(x_{1},x_{2}) = {\left\{ \begin{array}{ll} \text {sin}(1/x_2) &{} \text { if } x_2>0,\\ x_1 &{} \text { if } x_1\in [0,1]\cap {\mathbb {Q}}, x_2=0,\\ 0 &{} \text { if } x_1\in [0,1]\cap {\mathbb {Q}}^c, x_2=0,\\ 1 &{} \text { if } x_1> 1, x_2=0.\\ \end{array}\right. } \end{aligned}$$

It is not difficult to show that \(\succsim\) satisfies the Archimedean property. To see that \(\succsim\) fails to satisfy restricted solvability, note that \((1,0)\succ (0,2) \succ (0,0)\) since \(f(1,0)=1, f(0,0)=0\) and f(0, 2) is an irrational number in (0,1) interval (sin() of a non-zero rational number is irrational). Hence, for all \(x_1\in \mathbb {R}_{+}\), either \((x_1,0)\succ (0,2)\) or \((x_1,0)\prec (0,2)\). \(\square\)

The sixth and final example illustrates the importance of the interiority assumption in our results by showing that Theorems 2 and 3 above are false on a general choice set; see also Karni and Safra (2015) for the use of the interiority assumption to show that the Archimedean property is equivalent to mixture-continuity under cone-monotonicity.

6. Let \(X=\{x\in (-1,1)^2: x_1=x_2\}\). Then X is an order bounded and convex set in \(\mathbb {R}^2\), it is not open in \(\mathbb {R}^2\). (Note that X is open in the one dimensional subspace containing it.) Let \(u: X\rightarrow \mathbb {R}\) be defined by \(u(x)=0\) for all \(x\le 0\) and \(u(x)=1\) if \(x_1>0\). Let \(\succsim\) be the preference relation induced by u. Then, \(A_\succsim (0.5, 0.5)=\{x\in X: x_1>0\}\) which is not closed in X, hence \(\succsim\) is not continuous. It is clear that \(\succsim\) is weakly monotone. Since the restriction of any line parallel to a coordinate axis is a singleton, \(\succsim\) is trivially separately continuous.

Note that in the example above, we can replace X by \(\{x\in [-1,1]^2: x_1=x_2\}\), hence boundaries can be included. Similarly, we can replace X by a set with non-empty interior: \(\{x\in [0,1]^2: 2x_1\ge x_2\ge 0.5 x_1 \}\). Let \(u: X\rightarrow \mathbb {R}\) be defined by \(u(0)=0\) and \(u(x)=1\) if \(x \ne 0\). Let \(\succsim\) be the preference relation induced by u. Then \(\succsim\) is weakly monotone, separately continuous and discontinuous.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ghosh, A., Khan, M.A. & Uyanık, M. Continuity postulates and solvability axioms in economic theory and in mathematical psychology: a consolidation of the theory of individual choice. Theory Decis 94, 189–210 (2023). https://doi.org/10.1007/s11238-022-09890-z

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11238-022-09890-z

Keywords

Navigation