Category: economics
Type: argument
Associated thinker(s): Blaise Pascal
Original publication or approximate period: Pascal, Pensées (published posthumously, 1670)
Difficulty: Intermediate
Themes: decision under uncertainty, expected value, belief
Related experiments: The St. Petersburg Paradox; Pascal’s Mugging
Pascal presents the choice as unavoidable: “God is, or He is not,” and reason cannot decide which way to incline. In the wager passage, he says that if you gain, “you gain all,” and if you lose, “you lose nothing”; he then describes an eternal life and happiness as the possible gain against a finite stake. This is Pascal’s theological payoff setup, not an independently established account of what God would reward.
The standard decision-theoretic reconstruction says that wagering for God has infinite expected utility if God’s existence has any positive, finite, non-infinitesimal probability and the other relevant outcomes have finite utility. Pascal’s conclusion depends on those assumptions and on treating infinite utility as usable in expected-utility comparisons; neither point is uncontroversial. Pascal also distinguishes the wager from a direct choice to believe: when his interlocutor says he cannot believe, Pascal recommends practices intended to lessen obstacles to faith and says they may lead to belief.
Can prudential reasons justify choosing a way of life or taking steps toward belief when the truth cannot be settled by reason alone?
The wager is often treated as an early explicit decision-theoretic argument. It raises questions about whether belief can be chosen or cultivated through action, whether prudential reasons can support belief, and whether infinite values can guide rational choice. Pascal’s text also turns to inquiry: in a nearby fragment he urges the reader to search for truth, so the wager should not be reduced to a proof or to advice to stop investigating.
Its structure reappears in modern debates about risks with tiny probabilities and enormous stakes.
The many-gods objection, raised by Denis Diderot and developed by later critics such as J. L. Mackie and Paul Saka, asks whether rival religious hypotheses generate competing wagers. Replies include George Schlesinger’s proposal to use probability to break ties between infinite expectations and Jeff Jordan’s appeal to restricting the relevant “genuine options”; these replies are disputed. Objections to infinite utility include whether infinite utilities are coherent and whether standard expected-utility reasoning can distinguish choices when several outcomes have infinite or undefined expectations. Alan Hájek and Anthony Duff discuss a mixed-strategy objection; defenders and later reformulations include Bradley Monton, John D. Bartha, and Liz Jackson and Andrew Rogers. The voluntarism objection notes that belief may not be directly under one’s control. Pascal’s response is practical: he recommends acting as if one believed and lessening passions that obstruct faith, not a guarantee that these actions will produce belief.
The wager is a prudential argument for wagering for God, not a proof that God exists; this does not mean that Pascal’s unfinished Pensées contains no other arguments or appeals to evidence. Nor does the wager settle which religion to choose: the many-gods objection challenges its payoff table, while proposed replies remain contested. Pascal’s advice to someone unable to believe is to take practices intended to cultivate faith; the text separately urges the reader to seek truth. These are distinguishable from an immediate, voluntary act of belief.
Versions use finite but very large rewards, or add the possibility of a god who punishes believers. These alter the decision structure and so alter the conclusion.
Pascal’s Pensées was consulted in the English text reproduced by Project Gutenberg; the relevant passage is in printed pp. 66–71, fragments 233–241 in that edition. The SEP identifies its own cited 1910 translation as W. F. Trotter’s, but I did not independently confirm that translator metadata on the Gutenberg transcription itself. The consulted text checks a translation, not a manuscript or a critical French edition. The Stanford Encyclopedia of Philosophy and Internet Encyclopedia of Philosophy entries were read in the relevant sections. The University of Dayton Libraries exhibit supplies publication metadata: the first edition appeared posthumously in 1670, after a preliminary printing in 1669.
Research check: 2026-11-01. The older research/expansion.md records the initial URL/topic check, not this passage-level audit. This is an original educational explanation, not a substitute for the primary text. The questions and modernized payoff descriptions are teaching framings, not claims about Pascal’s exact wording or established historical variations.