Abstract
The paper sets up a general framework for defining the notion of verisimilitude. Popper’s own account of verisimilitude is then located within this framework; and his account is defended on the grounds that it can be seen to provide a reasonable structural or Pareto criterion, rather than a substantive criterion, of verisimilitude. Some other criteria of verisimilitude that may be located within the framework are also considered and their relative merits compared.
Similar content being viewed by others
Notes
I should like to thank a number of anonymous referees, whose comments led me to clarify and to correct some material from an earlier version of the paper, and I am especially grateful to Graham Oddie for his comments.
Pt satisfies (i). For by (Dt), it suffices to show Ptt ≥ Pt, which is (b). Also, Pt satisfies (ii). For suppose Q ≥t P, which, by (Dt), implies Qt ≥ Pt. By (a), Q ≥ Qt and so, by the transitivity of ≥, Q ≥ Pt. Finally, note that any proposition R satisfying (i) and (ii) is identical to Pt. For R ≥t P by (i); and so it follows from (ii) (putting R for Q and Pt for R) that R ≥ Pt. Similarly, Pt ≥ R; and so, by anti-symmetry of ≥, R = Pt.
It is interesting to note that the present definitions and lines of reasoning can be formalized within the modal system S5 with propositional quantifiers [as described in Fine (1970), for example]. Thus the above definition will take the form: ∀R(R ∧ □(Q ⊃ R) ⊃ □(P ⊃ R)). For the equivalencies below, we will need the axiom, ∃P(P ∧ ∀Q(Q ⊃ □(P ⊃ Q)), asserting the existence of a true world proposition.
First suppose P+ ≥ Q+. Take any true R for which Q ≥ R. Since R is true, Q+ ≥ R; since P+ ≥ Q+, P+ ≥ R; and since P ≥ P+, P ≥ R. Hence P ≥t Q. Now suppose P ≥t Q. Q+ is true and Q ≥ Q+. So P ≥ Q+; and since Q+ is true, P+ ≥ Q+.
A similar account of Pt is given in Miller (1978, p. 419) within the context of a Boolean algebra.
First suppose P− ≥ Q−. Taken any false R for which Q ≥ R. Then Q is also false. So Q− = Q ≥ R. Since P− ≥ Q−, P− is also false; and so P = P− ≥ Q− = Q ≥ R. Now suppose P ≥f Q. If Q is false, Q− = Q and, since Q ≥ Q−, P ≥ Q−. But P is then also false, and so P− = P ≥ Q−. If Q is true, then Q− = ⊤ and so P− ≥ Q−.
A discussion of some of them is to found in Oddie (2014).
Suppose Q ≥t Qʹ. To show P ∨ Q ≥t P ∨ Qʹ, it suffices to show, given (Dft), that (P ∨ Q)t ≥ (P ∨ Qʹ)t (the f-case is similar). By distribution, (P ∨ Q)t = (Pt ∨ Qt) and (P ∨ Qʹ)t = (Pt ∨ Qʹt). But given Q ≥t Qʹ, it follows from (Dft) that Qt ≥ Qʹt; and so (Pt ∨ Qt) ≥ (Pt ∨ Qʹt).
Zwart and Franssen (2007) have also drawn on the example of social choice theory in discussing the concept of verisimilitude, but in a completely different way and in a very different context.
In particular, we do not assume that closeness is linear or that limits to closeness or farness always exist. We also only take into account closeness to the actual world. Oddie (1990) subsumes Hilpinen’s proposal under the rubric of a “power relation”, as developed by Brink and Heidema (1987). This is in keeping with our own presentation of his proposal and is the rubric under which we shall develop an alternative truthmaker approach in other work.
Think of closer as more actual. Hence we use w \({ \succcurlyeq }\) v, not w ≼ v, to indicate closeness.
This result was not available to Hilpinen since he assumed, in effect, that the ranking relation is linear.
A somewhat different tabulation of alternative approaches to verisimilitude is given in Miller (2006, §10.1).
References
Brink, C., & Heidema, J. (1987). A verisimilar ordering of theories phrased in a propositional language. The British Journal for the Philosophy of Science, 38, 533–549.
Burger, I. C., & Heidema, J. (2002). Merging inference and conjecture by information. Synthese, 131(2), 223–258.
Fine, K. (1970). Propositional quantifiers in modal logic. Theoria, 36(3), 336–346.
Fine, K. (2016). Angellic content. Journal of Philosophical Logic, 45(2), 199–226.
Fine, K. (2019). Verisimilitude and Truthmaking. Erkenntnis. https://doi.org/10.1007/s10670-019-00152-z.
Goldstick, D., & O’Neill, B. (1988). Truer. Philosophy of Science, 55, 583–597.
Hilpinen, R. (1976). Approximate truth and truthlikeness. In M. Przelecki, K. Szaniawski, & E. R. Wojcicki (Eds.), Formal methods in the methodology of empirical sciences (pp. 19–42). Dordrecht: Ossolineum, Wroclaw, and Reidel.
Kuipers, T. A. F. (1982). Approaching descriptive and theoretical truth. Erkenntnis, 18, 343–378.
Kuipers, T. A. F. (1992). Naive and refined truth approximation. Synthese, 93(3), 299–341.
Miller, D. (1974). Popper’s qualitative theory of verisimilitude. The British Journal for the Philosophy of Science, 25, 166–177.
Miller, D. (1978). On distance from the truth as a true distance. In J. Hintikka, et al. (Eds.), Essays on mathematical and philosophical logic (pp. 415–435). Dordrecht: Reidel.
Miller, D. (2006). Out of error: further essays on critical rationalism. Burlington, VT: Ashgate.
Niiniluoto, I. (1978). Truthlikeness in first-order languages. In J. Hintikka, I. Niiniluoto, & E. Saarinen (Eds.), Essays on mathematical and philosophical logic (pp. 437–458). Dordrecht: Reidel.
Oddie, G. (1986). ‘Likeness to Truth’ Western Ontario series in philosophy of science. Dordrecht: Reidel.
Oddie, G. (1987). Truthlikeness and the convexity of propositions. In T. Kuipers (Ed.), What is closer-to-the-truth? (pp. 197–217). Amsterdam: Rodopi.
Oddie, G. (1990). Verisimilitude by power relations. British Journal for the Philosophy of Science, 41, 129–135.
Oddie, G. (2013). The content, consequences and likeness approaches to verisimilitude: compatibility, trivialization and underdetermination. Synthese, 190(9), 1647–1687.
Oddie, G. (2014). ‘Truthlikeness’. In N. Zalta (Ed.), The Stanford encyclopedia of philosophy, Fall 2001. Retrieved from http://plato.stanford.edu/archives/fall2001/entries/truthlikeness.
Popper, K. R. (1963). Conjectures and refutations. New York: Harper & Row.
Tichy, P. (1974). On Popper’s definitions of verisimilitude. British Journal for the Philosophy of Science, 25, 155–160.
Zwart, S. D., & Franssen, M. (2007). An impossibility theorem for verisimilitude. Synthese, 158(1), 75–92.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
The present appendix is not self-contained but presupposes some of the definitions already given in the main body of the text.
We first establish the sufficiency of Popper’s criterion given principles P1–P5 from Sect. 3. Given P1–P5, the following facts will obtain:
-
(1)
True(P) & P ≥ Q ⇒ P ≥t Q
Proof
Suppose P ≥ Q and that P (and hence Q) is true. By P1, @ ≥t Q. So by P4, P ∨ @ ≥t P ∨ Q. But, given that P is true, P ∨ @ = P and, given that P ≥ Q, P ∨ Q = Q. Hence P ≥t Q.□
-
(2)
Q+ ≥t Q.
Proof
By P1, @ ≥t Q. By P4, Q ∨ @ ≥t Q ∨ Q. But Q ∨ @ = Q+ and Q ∨ Q = Q. Hence Q+ ≥t Q.□
-
(3)
True(P) & P ≥ Q+ ⇒ P ≥t Q.
Proof
Suppose P is true and P ≥ Q+. P ≥t Q+ by (1). But Q+ ≥t Q by (2); and so P ≥t Q by P5.□
-
(4)
True(P) ⇒ Q ≥f P.
Proof
By P2 and P4, P ∨ Q ≥f P ∨ @. By P3 and P4, ⊥ ∨ Q ≥f P ∨ Q; and so by P5, ⊥ ∨ Q ≥f P ∨ @. But ⊥ ∨ Q = Q and, given that P is true, P ∨ @ = P. Hence Q ≥f P.□
From these facts, we can establish the sufficiency of Popper’s criterion:
Theorem 1
Given P1–P5 and the right-to-left direction of (Dfv),
-
P ≥v Q according to Popper’s criterion ⇒ P ≥v Q.
Proof
We easily confirm that it suffices to show:
-
True(P) & P ≥ Q+ ⇒ P ≥v Q
since this covers both the TT and the TF case.
So suppose P is true and P ≥ Q+. By (3) above, P ≥t Q; by (4), Q ≥f P; and so by the right-to-left direction of (Dfv), P ≥v Q.□
We turn to the characterization of Hilpinen’s criterion (Sect. 4) in terms of truth- and falsity-content.
Lemma 2
-
(i)
P \({ \succcurlyeq }\)t Q iff P↑ ⊆ Q↑
-
(ii)
P \({ \succcurlyeq }\)f Q iff P↓ ⊆ Q↓
Proof
(i) Suppose P \({ \succcurlyeq }\)t Q and pick a wʹ ∈ P↑. Then wʹ \({ \succcurlyeq }\) w for some w ∈ P. So there is a v ∈ Q such that w \({ \succcurlyeq }\) v. By \({ \succcurlyeq }\) transitive, wʹ \({ \succcurlyeq }\) v; and so wʹ ∈ Q↑.
Now suppose P↑ ⊆ Q↑ and pick a w ∈ P. Since P ⊆ P↑, w ∈ Q↑. So for some v ∈ Q, w \({ \succcurlyeq }\) v, as required.
(ii) Similar.□
We now consider the relationship between Popper’s and Hilpinen’s criterion for verisimilitude:
Lemma 3
(Soundness) If P ≥t,f,v Q in the world space S then P ≥t,f,v Q in any ranked space \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Proof
The case for v will follow from the cases for t and f. For if P ≥v Q in S then P ≥t Q and P ≥f Q in S; so P ≥t Q and P ≥f Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\); and so P ≥v Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Let us first consider the case in which P ≥t Q. We distinguish two subcases.
Subcase 1
P false or Q true. Suppose P ≥t Q in S. Then P ⊆ Q. So P↑ ⊆ Q↑; and hence, by lemma 2(i), P ≥t Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Subcase 2
P true, Q false. Suppose P ≥t Q in S. Then P ⊆ Q+. So P↑ ⊆ Q+↑. But Q+↑ = (Q ∪ {@})↑ = Q↑ since @ ∈ Q↑; and so P ≥t Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Let us now consider the case in which Q ≥f P. We distinguish three sub-cases:
Subcase 1
P true. Suppose Q ≥f P. Since this holds automatically, we must show Q ≥f P whenever P is true. Take any v ∈ Q. Since P is true, @ ∈ P. But @ \({ \succcurlyeq }\) v by Weak Centering.
Subcase 2
P false, Q true. This case is trivial since it is not one in which Q ≥f P can hold.
Subcase 3
P, Q false. Suppose Q ≥f P in S. Then Q ⊆ P. So P↓ ⊆ Q↓; and hence P ≥f Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) by lemma 2(ii).□
Lemma 4
(Completeness) For any world-space S = (W, @):
if P ≥t,f,v Q in every linear ranked space \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) then P ≥t,f,v Q in S.
Proof
As before, it suffices to prove the result for the alethic values t and f.
Case A
Suppose that P ≱t Q in a world-space S = (W, @). Again, we distinguish sub-cases according to the truth-values of P and Q.
Subcase A1
P is false or Q true. Then P ⊆ Q. So for some w0, w0 ∈ P but w0 ∉ Q. Since P is false or Q true, w0 ≠ @. Choose \({ \succcurlyeq }\) so that w0 ≺ w and w ≈ @ whenever w ≠ w0. Thus the worlds are partitioned into two blocks, with w0 the most distant world and all others the near worlds, as depicted below:
Then for no v ∈ Q is w0 \({ \succcurlyeq }\) v; and consequently, P ≱t Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) as so defined.
Subcase A2
P true, Q false. Then not P ⊆ Q+. So for some w0, w0 ∈ P but w0 ∉ Q+. But since @ ∈ Q+, w0 ≠ @. We may then proceed in the same manner as Case (1).
Case B
Suppose that Q ≱f P in the world-space S = (W, @). We distinguish three sub-cases:
Subcase B1
P true. This case cannot arise.
Subcase B2
P false, Q true. Since never Q ≥f P, we must show that Q ≥f P always fails in some ranked space \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\). Since Q is true, @ ∈ Q and, since P is false, @ ∉ P. Choose \({ \succcurlyeq }\) so that w ≺ @ whenever w ≠ @ and w ≈v whenever w, v ≠ @. Thus the worlds are again partitioned into two blocks, but this time with all worlds other than @ the distant worlds and @ the near world:
Then for no w ∈ P is w \({ \succcurlyeq }\) @ and consequently Q ≱f P in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) as so defined.
Subcase B3
P, Q false. Then Q ⊆ P and so, for some world w0, w0 ∈ Q but w0 ∉ P. Choose \({ \succcurlyeq }\) so that w0 ≈ @, w ≺ @ whenever w ≠ w0, @ and w ≈ v whenever w, v ≠ w0, @, as depicted below:
Then for no w ∈ P is w \({ \succcurlyeq }\) w0 and, consequently, Q ≱f P in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) as so defined.□
Theorem 5
(Soundness and completeness) For any propositions P and Q in a world space S:
-
P ≥t/f/v Q in S iff P ≥t/f/v Q in every \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Proof
Note that the proof only uses linearly ranked spaces with two blocks. If we insist upon Strong Centering, then the number goes up to three. Indeed, we may provide an exact characterization of the classes of Hilpinen rankings which give rise in this way to the Popperian ordering. Let X be a class of linear rankings on a world-space S. We say that such a class X is t/f/v-adequate if P ≥t/f/v Q in S whenever P ≥t/f/v Q in every \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) for which \({ \succcurlyeq }\) ∈ X (by Soundness, the converses will also hold). Given a linear ranking \({ \succcurlyeq }\) on S, say that the world w is worst if w ≺v for any world v ≠ w, i.e. w is most distant from actuality, and say that w is best if w \(\succ\) v for any non-actual world v ≠ w, i.e. w is closer to actuality than any non-actual world.
Theorem 6
Let X be a class of linear rankings on a given world space S . Then:
-
1.
X is t-adequate iff each non-actual world is worst in some ranking \({ \succcurlyeq }\) of X.
-
2.
X is f-adequate iff each world is best in some ranking \({ \succcurlyeq }\) of X.
-
3.
X is v-adequate iff each non-actual world is worst in some ranking \({ \succcurlyeq }\) of X and each world is best in some ranking \({ \succcurlyeq }\) of X.
Proof
(i) Suppose each non-actual world is worst in some ranking \({ \succcurlyeq }\) of X. We may then follow the method of proof in Subcases (A1) and (A2) of the proof of theorem 4 to show that P ≥t Q fails in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Suppose that some non-actual world w0 is worst in none of the rankings \({ \succcurlyeq }\) of X. Let P = {w0} and Q = W − {w0}. Then P ≥t Q in every \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) for which \({ \succcurlyeq }\) ∈ X. However, P ≥t Q fails in S.
(ii) Suppose any world is best in some ranking of \({ \succcurlyeq }\) of X. To take care of subcase (B2), we let that world be @ and, to take care of (B3), we let it be w0.
Suppose that some world w is best in none of the rankings \({ \succcurlyeq }\) of X. We distinguish two cases. (a) w = @. Let P = {@} and Q = W − {@}. Then P ≥f Q in every \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) for which \({ \succcurlyeq }\) ∈ X while P ≥f Q fails in S. (b) w ≠ @. Let P = {w} and Q = W − {w, @}. Then again P ≥f Q in every \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) for which \({ \succcurlyeq }\) ∈ X while P ≥f Q fails in S.
(iii) From (i) and (ii) since X is v-adequate iff it is t-adequate and f-adequate.□
Corollary 7
For no linear rank on a world space S with at least three worlds is P ≥v Q in S identical to P ≥v Q in \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\).
Proof
We let the X in Theorem 6 be {\({ \succcurlyeq }\)}. The result then follows from the fact that no non-actual world in S can be both worst and best with respect to \({ \succcurlyeq }\).□
Let the discrete ordering \({ \succcurlyeq }\) on a world space be {(w, w): w ∈ W} ∪ {(@, w): w ∈ W} and let the discrete ranked space \(\varvec{S}^{{\mathbf{ \succcurlyeq }}}\) be the one in which \({ \succcurlyeq }\) is the discrete ordering on S.
Theorem 8
The Popper relations ≥t/f/v are the same as the Hilpinen relations ≥t/f/v over a discrete ranked space.
Proof
It suffices to show that in a discrete ranked space,
-
P↑ = P = Pt when P is true
-
P↓ = ⊤ = Pf when P is true
-
P↑ = P+ = Pt when P is false
-
P↓ = P = Pf when P is false.
But these facts are readily verified.□
Rights and permissions
About this article
Cite this article
Fine, K. Some Remarks on Popper’s Qualitative Criterion of Verisimilitude. Erkenn 87, 213–236 (2022). https://doi.org/10.1007/s10670-019-00192-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10670-019-00192-5