Let’s try build a picture to model theism, atheism and agnosticism using the probabilities approach from the previous post. For that we need some statement of theism. Of course this devolves into the problem of definitions, but let me go around it with the following statement which will do for now:
G: there is at least one god. For “god”, take any coherent/non-vacuous definition by any prominent adherent of (and authority in) Christianity, Islam, Buddhism, Sikhism, Juche, Judaism, Bahai, Shinto, Cao Dai, Zoroastrianism. So another phrasing for G: at least one coherent/non vacuous definition of god within Christianity…Zoroastrianism corresponds to an entity that exists.
I’m bypassing a lot of definitional nonsense here. I’m not a lawyer so I’m expecting the above to be read charitably. You probably know what I mean even if I didn’t say it. By vacuous I’m excluding things like “God is love” or Karen Armstrong’s definitions of a god and so on. Deistic-like belief in some abstract “god of the philosophers” or anything similar should be covered by G since Christianity includes prominent clergy who in essence talk of the Christian god in that way. If all definitions you’ve seen were incoherent/vacuous and you expect any others you haven’t encountered to be the same, well then your probability of G being true will be low, that’s all. And so on.
Now, to list at the relevant definitions that we’re working on. These are phrased by me but obviously gleaned from elsewhere.
- atheist v1: there are no gods
- atheist v2: does not believe there are gods
- agnostic v1: whether or not there are gods is unknown
- agnostic v2: whether or not there are gods is unknowable
- agnostic v3: whether or not there are gods is uncertain
- theist: at least one god exists
So let’s say you’ve placed a dot that represents your assessment that G is true somewhere on the scale below (last post that argued for the necesity of such a scale). FYI, my current estimate for G is 10% so I’ve also placed an X showing where I sit:
I think the assessment of whether you believe G (or not) is based on personal opinion and does not follow from the dot’s placement automatically. How strong must your confidence be before you believe something? This is different for different people. Two people might both put G at 75% but to one of them this qualifies in “believing in” G and to the other it doesn’t. Of course it would be very credulous to believe say everything which has more than 60% chance of being true (in your estimate). On the other hand, 90% might be too high. You probably believe things which to you are less certain than that. But that’s the whole point — our idea of “believing in” is hard to map onto a more accurate probability scale because it’s intrinsically hard to map a two valued simplification onto a more sophisticated model. In the end, the very idea that “I believe” something is a shortcut.
Since I’m writing the blog post I’ll set the number at 75%. So 75%-100% makes you a theist. Interestingly, by symmetry, 0%-25% makes you an atheist v1. This is because the whole scale is symmetric. If you’re only 25% certain of G then it’s just a mathematical fact that you’re also 75% certain of ~G (ie. the nonexistence of gods) which combined with my arbitrary 75% threshold puts you in the “there are no gods” category. So I guess whether I like it or not, I’ve just placed myself into the strong atheist category based on my own model.
What about atheist v2? Well again because in my model to believe in something you have to be 75% sure, a v2 atheist would include EVERYONE from 0% all the way to 75%! This is unintuitive but follows completely from the definitions — so once again I blame the fuzzy nature of our concept of “believing in” something. Or at the very least, what it says is that if you require a reasonably high threshold of evidence to believe something (eg. 75%) then you should consider yourself an atheist if your own estimate for the existence of gods is below that very threshold.
Now, what to do about the 3 versions of agnostics? I will conflate v1 and v3 as speaking of very similar things, but that “certain” speaks of higher probabilities than “known”. Again this reflects our intuitive usage (“I know this but I’m not certain”). But “known” also sounds more sure than simply “believe”. So if I put “believe” at 75%, I will put “know” at 85% and “certain” at 90%. In this case, an agnostic v1 is everyone whose dot on the above scale is from 15%-85%. An agnostic v3 is an even wider band: 10%-90%. Note that this model highlights that gnosticism/agnosticism is orthogonal to theism/atheism. Which is at least a point in the favour of those who claim these things are orthogonal.
Now what to do about agnostic v2? Saying something is unknowable could mean saying that you can’t even estimate a probability for G, or that the probability for G is undefined. But I think that in the type of approach I’m describing, unless G is meaningless it might not even be possible for there to be an undefined probability. Please comment if you know something about this point because I don’t. But in my ignorant opinion, I think for any statement you would simply start at the 50-50 mark and then use external evidence to update your estimate, ie. nudge it to the left or right. So in the absence of any decent evidence you would remain at 50%. If this is true then you would only be an agnostic v2 if your estimate for G is 50%.
There is another interpretation: that the universe is such that we will never be able to get into the “know” bands of the scale. In my case the “know” bands are 0%-15% and 85%-100%. Of course if you say that you are making a claim that’s separate from G. However you are also telling us that your estimate is in the 15%-85% range. Under this analysis, an agnostic v2 is anyone whose estimate is in this range AND who makes the extra claim that no evidence exists in the universe to get to either side of this band.
This whole thing might leave you unsatisfied but I think that’s the point. Folk terms like “believe” aren’t used very consistently by our brains. If we attempt to make their usage precise and consistent, we get unusual results, like my above system classifying you as an atheist (v2) if you are 70% sure that Islam is TRUE. But I think this is as close as we can get to formalising this. If you see anything wrong with the model, have at it!