• Weak Agnosticism: not knowing whether God exists or not, but remaining open to deciding later (presumably after thinking more on the subject or gaining more facts).
  • Strong Agnosticism: not knowing whether God exists or not, and deeming it impossible to ever know. No amount of future knowledge or evidence will change a strong agnostic's mind on this.
  • Pascal's wager: A non believer stands to lose more if God exists than a believer stands to lose if God doesn't exist. A rational person should behave as if God exists whether he has proof of it or not.

Now my reasoning is the following: A weak agnostic can assign likely hood to each outcome (theism vs atheism), since her position regarding the question can change over time. It is conceivable that at some point Pascal's criteria doesn't hold, since although she still thinks it is possible that God exists, the likely hood is so small that that she stands to loose more by believing than by not believing.

Implicit in the strong agnostic's point of view is that both outcomes are equally likely, since no information gain or loss on the question is ever possible. If both outcomes are equally likely, then Pascal's argument of loss vs gain stands, and a strong agnostic should live as if she believed in God.

Does this reasoning make sense? Should a strong agnostic follow Pascal's advice?

  • 1
    You could generate (probably innumerable) contradictory ideas which, according to Pascal's wager, you should believe in. For example, a monotheistic god and polytheistic gods, both promising infinite torture if not believed in exclusively. Then, you have to resort to old fashioned evidence-based reason to choose which is more likely true, which defeats the point of the wager. May 4, 2015 at 17:11
  • For the form of strong atheism stipulated in this question it is unclear to me that the strong atheist is required deem both outcomes as equally likely.
    – Dave
    Dec 18, 2015 at 18:47
  • It also strikes me that this definition of strong agnosticism is inconsistent with the definition of Pascal's god: there is at least one state of affairs, being rewarded after death by god, that would serve as "evidence [that] will change a strong agnostic's mind". Therefore, one cannot be a strong agnostic (as defined here) with respect to Pascal's god, since that god provides situations that provides evidence of its existence.
    – Dave
    Dec 18, 2015 at 18:59
  • Except here, I have not seen a definition of strong agnosticism that uses the language "impossible to ever know"; which I read as logical impossibility. One cannot be this kind of strong agnostic about bigfoot -- there exist logically consistent conditions (e.g. a population explosion of bigfoots [bigfeet?] causing them to be a nuisance, like deer are in some places, in Seattle) that would provide evidence to guide the agnostic's beliefs. My claim is that the definition of god used in this argument is potentially manifest in a similar way.
    – Dave
    Dec 18, 2015 at 21:06

6 Answers 6


The determining factor is not how strongly you are convinced that knowledge of God's existence is possible or impossible, but rather the extent to which you are convinced that belief in God would be beneficial in the case that God exists. It's all about the value you place on belief, not the probability you assign to that belief being correct.

In the classic version of Pascal's wager, it's assumed that there is no substantive cost to believing-if-wrong, and infinite gain to believing-if-right. If we can accept that as given, believing is the right choice no matter what probability we assign to the outcomes.

If you think that belief is always positive, even if wrong (for instance, social and mental benefits), the math becomes even simpler.

On the other hand, if you perceive a negative value to believing-if-wrong then the probabilities do come into play. To take it through to the extreme, if you feel believing is negative, even if correct, then the wager becomes a sure loss.

In summary, the probabilities become important only in the case where we can confidently assign a positive value to belief-if-correct and a negative value to belief-if-wrong. In all other cases the decision about valuating the outcomes dominates.

NOTE: I've elided the reverse cases of disbelief-if-right, disbelief-if-wrong because the logic is the same, mutatis mutandis.

  • "In summary, the probabilities become important only in the case where we can confidently assign a positive value to belief-if-correct and a negative value to belief-if-wrong." Not sure I follow, but isn't that an assumption of the wager to begin with? I haven't read the original in a long time, and I think there were several versions of it, but I had always thought there was some assumed cost to "believing-if-wrong" placed into the balance. If so, isn't "wait-and-see" a viable third option with a different probability? Dec 18, 2015 at 18:41

Short Answer: No, because the odds of winning on this wager are extremely low so no case (including Strong Agnosticism) justifies taking it.

Long Answer:

This wager only seems reasonable (to some) because of its implicit (unfounded) assumptions. Take these assumptions away and you'll see this wager is simply not worth playing. In particular this wager assumes that...

  • There's only a choice between believing in the Christian God and not. This ignores many religious beliefs including incompatible branches of Christianity. When you remove this assumption, your odds of winning are greatly reduced.
  • One needs only to assert a belief. This ignores the cost of adopting a religion. When you remove this assumption, you realize that you have much to lose by adopting a religion, such as the opportunity cost of living life as you see fit.
  • The choice is between heaven and hell. This ignores the range of reward/punishment existing in religions -- even within the spectrum of Christian belief. When you remove this assumption you'll see that the reward/punishment structure of the wager is not so grim.

Looking at the above, this may be better looked at as an economic or game theoretic problem. Try to calculate the utility or pay-off for this wager. I'm not saying you can, but the attempt to do so should build a greater appreciation for the grossly oversimplified nature of this wager.

  • I kind of place all of the monotheistic religions under one big tent (there is one deity and, do good, go to heaven - do bad, go to hell) and I was assuming Pascal was doing the same. Now that you mention it, I do realize that all sorts of beliefs can fall under the same wager. It's fun to note that the same wager can be used to justify Scientology. May 4, 2015 at 20:00
  • @AlexanderSKing What you describe sounds more like the Noachide than Christianity or Islam, although beliefs in those two religions vary. I've seen beliefs as extreme as Sunni Muslims believing that all non-Sunnis (even other Muslims) are damned. Good point on how the wager can support a variety of religious beliefs :)
    – R. Barzell
    May 4, 2015 at 20:08
  • This is a direct challenge to Pascal's wager, not an answer to the question of how the relative probabilities affect the scenario. May 5, 2015 at 15:25
  • @ChrisSunami with all due respect, it's an answer because it demonstrates that the wager should never be taken, which in turn implies that the wager should not be taken in the Strong Agnostic case. Still, it wouldn't hurt to adjust the answer to make its relevance a bit clearer.
    – R. Barzell
    May 5, 2015 at 15:33
  • 1
    @ChrisSunami I appreciate you taking the time to explain your reasons for the down-vote and your feedback has helped improved the answer. I much prefer this version -- thanks for your comments :)
    – R. Barzell
    May 5, 2015 at 17:37

In a word, No. Pascal's wager makes a number of assumptions which our hypothetical Strong Agnostic should question:

  1. Assumes the cost of feigning belief is negligible
  2. Assumes penalties for non-worship are worse than those for worshiping the "wrong" deity
  3. Assumes "god" doesn't (or can't) differentiate between those with true belief versus those who are just hedging their bets
  4. Assumes "god" does not reward intellectual honesty

As the end of the day, there is only one existence we can be certain of. The one we're living now. I'd recommend living it like it's the only one, because if you're right at least you didn't hold back due to some arbitrary rules. If there is another life after? Bonus!


A strong agnostic should live as if she believed in God

Which god? Any definition of Pascal's Wager goes (very roughly):

All else equal, for the lowest risk and greatest possible payout one should live as though an entity X possessing the set of properties S, which include that X has the power to assign humans "reward" and "punishment", exists.

Now consider an entity Y with the same power and the property that "Y will punish any human who lives as though X exists with extreme prejudice".

Both X and Y are equally likely to exist under our assumption of "all else equal" strong agnosticism. I would wager the payoff is not there.

The point being that for any belief which one might live by according to some formulation of Pascal's Wager, an opposing Wager may be formulated by which one absolutely should not follow the first. It just doesn't work.


Pascal's wager assumes there is a heaven and a hell, and that heaven is better. It also provides a very vague term "believe" to indicate how one gets to heaven.

It would seem better definition of these terms is required to even evaluate Pascal's wager. Likely an analysis of it could be made based on each different worldviews that exists. i.e. If Christianity is true, then x. If Buddhism is true, then y. Etc.

As another answer mentioned, the wager may require not only that you "believe", but that belief may require action - following the religion.

In reality, there are a LOT of people that have taken Pascal's wager - they don't necessarily believe in the core tenets of the religion, but due to culture, duty, fear of what really happens at death, etc., they do follow through with some, many, or all of the religion's dictates. Not everyone in this category realizes philosophically that this is the situation they are in...

All of that being said, what about the strong agnostic? I believe he needs to determine what he believes the "most likely" worldview is - or the most likely form of God. Is it the God of the Bible? A Deist God? Etc., and then evaluate the wager based on that.

What allows him to choose the "strongest"? Whatever he chooses as evidence. Logic, philosophy, the lives/reasoning of those around him, etc.


Just to be contrary, I do accept your reasoning, sort of. Though I agree with others about all the problems with Pascal's wager to begin with.

For the Strong Agnostic, the probability of G is exactly 50:50 and will never change. So the strong agnostic can at least respond to the wager, yay or nay, depending on how the relative costs-benefits are worked out.

For the Weak Agnostic, the probability of G changes over time, according to new information, so there is a third option... wait. However, there must be some initial assignment of probability and some actuarial expectation that the "lifetime" will afford significant new information. So the third option introduces a separate "life expectancy" probability intersecting with the initial assigned probability, even as the latter changes with new information...on the way to Damascus, say.

However, I do not think you can eliminate relative costs-benefits from the wager. Yet any assignment of values must somehow compare them within both a finite and an infinite span. The introduction of eternity, presumably attending upon an existent God, simply swamps any sort of probabilistic evaluation. Perhaps that was Pascal's true intuition, for he seemed to have a kind of rational horror of an eternal abyss.

  • Always good, and sometimes educational, to hear the reason for down-votes. Dec 18, 2015 at 18:29
  • 1
    Depending on how one sets up the model for "gods" and what fraction of that domain falls into the type of god stipulated in Pascal's wager, one could have very different priors (assuming a suitable principle of indifference can even be defined). Choosing 50:50 prior for a binary choice is pragmatically useful when you expect to use additional information to refine your belief, but is not often the most accurate representation of the prior, which in this case is crucial. In a nutshell the strong assertion that the degree of belief is absolutely 50:50 strikes me as wrong.
    – Dave
    Dec 18, 2015 at 18:41
  • No, I am just saying 50:50 for the "strong agnostic" as defined. The chances of any God (or just Afterlife) existing or not existing are exactly equal.... to the strong agnostic. She is a Kantian who claims this cannot ever be known, so odds will never change.I agree we can introduce all sorts of talk about kinds of gods or whatever, but that was not my point of agreement with the premise. I am simply agreeing that someone anticipating "further information" has reason to approach the probabilities differently. Dec 18, 2015 at 18:59

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .