Changing Beliefs Rationally: Some Puzzles

  • Dorothy Edgington
Part of the Philosophical Studies Series book series (PSSP, volume 52)


The puzzles I shall discuss belong to the subject names (by Richard Jeffrey’) “Probability Kinematics,” that is, the question of how probability judgments should change in the light of new information. But I start with a bit of the pre-history of this subject as I see it.


Conditional Probability Belief Revision Inductive Logic Labour Party Probability Judgment 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    There is less agreement aey, The Logic of Decision, (McGraw-Hill, 1965) chapter 11.Google Scholar
  2. 2.
    I have in mind the work of R. Carnap, especially The Logical Foundations of Probability. Google Scholar
  3. 3.
    B. van Fraassen, “A Demonstration of the Jeffrey Conditionalization Rule,” Erkenntnis, 24, 1986,pp. 17–24.Google Scholar
  4. 4.
    B. van Fraassen, op. cit., pp. 22–23.Google Scholar
  5. 5.
    Peter Williams, “Bayesian conditionalisation and the Principle of Minimum Information,” British Journal for the Philosophy of Science, 1980.Google Scholar
  6. 6.
    Bas C. van Fraassen, “A Problem for Relative Information Minimizers in Probability Kinematics,” British Journal for the Philosophy of Science, 1981.Google Scholar
  7. 7.
    Op. cit. Google Scholar
  8. 8.
    R.I.G. Hughes and B.C. van Fraassen, “Symmetry Arguments in Probability Kinematics” in P. Kitcher and P. Asquith (eds), PSA 1984,vol 2 (East Lansing Michigan); and Bas C. van Fraassen, R.I.G. Hughes,and Gilbert Harman A promble for Relative Information Minimizers,Continued (unpublished).Google Scholar
  9. 9.
    There is less agreement about how to explain this fact. David Lewis gives a Gricean explanation in “Probabilities of Conditionals and Conditional Probabilities,” Philosophical Review 85,1976,pp. 297–315,reprinted in his Philosophical Papers Volume II (Oxford, 1986). He later abandoned this in favour of Frank Jackson’s explanation in “On Assertion and Indicative Conditionals,” Philosophical Review 88 (1979), pp. 565–589, and “Conditionals and Possibilia,” Proceedings of the Aristotelian Society 81 (1981) pp. 125–137. See also Ernest Adams, The Logic of Conditionals, (Reidel, 1975); and my “Do Conditionals Have Truth Conditions?,” Critica, 1986, pp. 3–39, forthcoming in Frank Jackson (ed.) Conditionals, (Oxford: Oxford Readings in Philosophy, 1990).Google Scholar
  10. 10.
    This puzzle is a (depoliticised) version of one put by John Skorupski to Crispin Wright as a problem for an intuitionist-style conditional; and also, I heard, by David Lewis to Frank Jackson as a problem for Jackson’s claim that conditionals are assertible only if they are “fit for modus ponens.” See Frank Jackson, op. cit.. I have since discovered by solution to it is to be found, in essence, in Robert Stalnaker, Inquiry, (MIT Press, 1984),p. 105. See also David Lewis, Philosophical Papers void’, op. cit, pp. 154–156.Google Scholar
  11. 11.
    I believe, following Ernest Adams. op. cit., that conditional judgments express high conditional probabilities. But you do not need to agree with me about that. Here, the puzzle is about high conditional probabilities, and conditionalisation.Google Scholar
  12. 12.
    My treatment of this example shows that it is not a counterexample to Jackson’s claim: It is just that one can never get into the position to use modus ponens (one would if one could!), for one can never get the information that the antecedent is true without getting further relevant information. Whether the intuitionist-style conditional can be similarly saved is part of the general question of how to adapt logical constants designed for mathematics to empirical discourse. It is not my problem. See Crispin Wright, Realism, Meaning and Truth,(Blackwell, 1988), Appendix.Google Scholar
  13. 13.
    I am grateful to my ex-student Ruth Weintraub both for introducing me to this puzzle, and for being the first to suggest that A’s probability of being hanged should not change in response to this information.Google Scholar
  14. 14.
    According to this theorem of the probability calculusGoogle Scholar
  15. 15.
    Alex Edgington, my son, suggests the following (which is somewhat above my head). (It is framed in terms of a second question-and-answer equivalent to my 2’. “Given that we’re in R, how likely is it that we’re in R27” “0”): “another way of showing that the amount of information we get depends on how we ask the question is as follows: suppose Judy Benjamin assigns some initial probability density to the speaker’s degrees of belief (which are themselves probabilities) in the four possibilities. From this density function, which is over four real variables, we can derive density functions corresponding to variables defined in terms of those four—in this case we are interested in ”R2“ and ”R2/(R,+R2).“ When we evaluate these derived densities at the answer received (0 in both cases), we get a measure of how plausible she would have thought that answer, and hence (inversely) of how much information it gives her. The point is that value is different for different variables (”R2“ and ”R2/R1+R2)“), even if the answers tell her the same fact about R2. For example, suppose Judy assigns the following distribution to the speaker’s beliefs: R2 is distributed uniformly on [0,1] and R, uniformly on [0,1-R2]. It is then easy to show that the p.d.f.of ”R2“ evaluated at 0 is 1, whereas the p.d.f. of ”R2/(R,+R2)“ evaluated at 0 is infinite (so she gets more information in the first case).”Google Scholar
  16. P(A&B) = P(A)P(B/A) = P(B)P(A/B). So P(B/A) = P(B) iff P(A/B) =P(A).Google Scholar
  17. 17.
    “Probabilities of Conditionals and Conditional Probabilities,” op. cit. Google Scholar
  18. 18.
    See my “Do Conditionals Have Truth Values?,” op. cit. Google Scholar
  19. 19.
    See Simon Blackburn, in Charles Travis (ed.) Meaning and Truth, (Blackwell, 1986).Google Scholar
  20. 20a.
    Bas van Fraassen, “Probabilities of Conditionals” in W.L. Harper and C.A. Hooker (eds) Foundations of Probability Theory,StatisticalInference,and Statistical Theories of Science, Volume 1 (Reidel, 1976)Google Scholar
  21. 20b.
    Brian Ellis, Rational Belief Systems (Blackwell, 1979), pp. 74–80;Google Scholar
  22. 20c.
    Brian Ellis “Two Theories of Indicative Conditionals,” Australasian Journal of Philosophy, 1984.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1992

Authors and Affiliations

  • Dorothy Edgington
    • 1
  1. 1.Birkbeck CollegeUniversity of LondonEngland

Personalised recommendations