Do ideas matter in religion?
(Immigration and Religion, pt. 2 of 4)

Peter A. Taylor
July, 2016

If we want to carry on an intelligent conversation about ideas and religion, I think we have to talk in terms of probability and statistics. Let's get out our polyhedral dice.

If I were writing a Dungeons and Dragons-like fantasy role-playing game (FRPG), and wanted to incorporate religion, I would be tempted to adopt Richard Dawkins' model of religion as a memetic infectious disease (and/or symbiont). An infected non-player character (NPC) tries to proselytize a player character. I would treat this as an attack in which the attacker's charisma is compared against the defender's wisdom, and the defender rolls the dice. There may also be modifiers for the characters' races.

If the "attack" is successful, there are likely to be both good and bad effects. In fact, the defending character may choose not to defend at all. But once infected, the player may have to act as if there is someone else living in his head, occasionally trying to take over, and getting him to do things he doesn't want to do. To some extent, the D&D alignment system already kind of works this way. I knew a guy in college who was always getting into trouble with his Dungeon Master (DM) because his alignment was supposed to be lawful-good, but he kept wanting to make unprovoked attacks on NPCs that he identified as evil. The DM would punish him for acting contrary to his alignment. The god or prophet or book author living inside my head may want me to tithe, or be celibate, or kill apostates, or let myself or my kinfolk go unavenged, or tell the truth against interest, or act confused about the difference between friends and kin vs. strangers. Different religions will challenge my desired behavior in different ways, to different extents, and with different probabilities. But most of the time, the character is able to "make his saving throw" and resist these attempts, if he finds it in his interest to resist. Some people (higher wisdom?) are better able to resist than others.

So yes, ideas matter in religion, but the various religions will produce different behavioral probability distributions in their devotees, which will be hard to distinguish from the effects of race, intelligence, wisdom ("emotional IQ"?), peer effects, and probably a lot of other variables that I haven't thought of.

My Episcopalian minister friend says that people are constantly arguing about how to interpret scriptures (e.g. particularism vs. universalism), but that the Bible and the Koran give their devotees different probabilities of winning for different positions. Christianity has more support for separation of church and state ("render unto Caesar") than Islam does; a Christian might decide to force his religion on others and a Muslim might decide not to, given the opportunity, but a Christian who argues with other Christians in favor of separation of church and state has a higher probability of winning the argument than a Muslim arguing that position with other Muslims. Christians and Buddhists who get caught up in holier-than-thou competitions have relatively high likelihoods of becoming pacifists (e.g. Quakers), where Muslims in similar competitions have a higher likelihood of becoming jihadis. Also, the Bible is a more heterogeneous book than the Koran, so the Christian who wants to cherry-pick the Bible to support a pre-determined position has more to work with than the Muslim (Christians find it easier to make their saving throws).

Someone like Larry Iannaccone would probably argue that, once you understand the peer effects in religion, a lot of the apparent irrationality (possession by a hostile power living in your head) disappears, at least in certain religions. As John Wesley put it, "Christianity is essentially a social religion and that to turn it into a solitary religion is indeed to destroy it." I may tithe not because I am irrational, but because my peers reward me for this, perhaps with higher social status. Much of the economics of religion literature views "strict" or "sectarian" churches as "clubs" that exist largely to produce certain kinds of collective goods. It's going to be hard to talk intelligently about collective action without asking whether a peer group exists that is large enough and powerful enough to be capable of performing it. It's also going to be hard to talk about an individual's incentives to follow a set of rules or obey an authority figure within a religion without asking about the peer group that is going to be providing many of these incentives. The size of an opposing peer group should matter, too, because they also provide incentives.

From a game designer's perspective, consideration of peer effects may merely move the conflict from an internal struggle with "the man in the breast" to an external struggle with one's peers or leaders. But the easiest way to incorporate peer effects in a game might be to say that the effective strength of the god living in my head is a strong function of the strength of my peer group, and the likelihood of me being able to make my saving throw against a possession attempt will vary accordingly. The presence of an apostate nearby should also have a strong effect on the saving throw. Yes, ideas matter, but the strength of the peer group matters too, and so does the nature of the people who work the ideas.

This is not to suggest that all religions are basically the same. One weak deity in my FRPG may want me to be a pacifist and another weak deity may want me to set the world on fire. The fact that neither has very strong influence on my behavior doesn't mean that they influence me in the same direction.

Bibles and theologians are like cloth and tailors. The tailor is more important than the cloth, but the cloth still makes a difference.

The above discussion treats religion as a hostile takeover, an infection rather than a symbiont. A more conventional way to handle religion (e.g. in D&D) is to treat it as a powerful but expensive tool that one shops around for and chooses. Hopefully, one chooses one's purpose first, and then chooses the appropriate tool. If you want to go to war, you buy a sword. If you want to farm, you buy a plowshare.

You may not have much of a choice. If you want to get along with your neighbors, you may have to join the neighborhood church (or temple or mosque). Sometimes you have to use the tool you have at hand rather than the tool you want. You may have to adapt it, and some religions may be more adaptable than others. So instead of thinking of a religion as a sword or plowshare, we can think of religion as a metal blank that comes from the mill in some shape that's intermediate between a sword and a plowshare. You may have to take it to your forge in the back yard and pound it into the shape you want. Some religions may come in a shape that's pretty near to being a finished plowshare, and you may have to work on it a lot to make a sword. Other religions may come in a shape that's pretty near to being a finished sword, and you may have to work on it a lot to make a plowshare. If this is the case, the question arises: To what extent does the tool you have in front of you (and the peer group that comes with it) influence what purpose you decide to pursue?

The above discussion also supposes that there is a natural default non-religious state. Depending on how you define "religion", this may not be true. You can define "religion" narrowly, so that your belief system doesn't count, but you still have a belief system and peer group of some kind. Your peers are still going to want you to do something against your interests, and you still need to make your saving throw.

Now let's think about non-player characters. Is an NPC likely to engage in predatory behavior? Like, maybe, strangling other members of the caravan, or voting for predatory politicians? I propose a statistical model for NPC behavior as a function of religion:

The probability (P) of predatory behavior on the part of an NPC is a function of an index of suspicion (Z), that could be any real number. The function might be, say, a cumulative normal distribution with mean μ and standard deviation σ.

P = normcdf( ( Z - μ ) / σ )

The index of suspicion might be the sum of some baseline value, A₀, and a bunch of other functions of variables that have nothing directly to do with religion, A₁(x₁) ... A_n(x_n). If someone is bigger than you are, for example, he is a lot more likely to try to overpower you than he is if he's smaller than you.

Z = A₀ + A₁(x₁) + ... + A_n(x_n)

Now let's throw in a function, B, of the NPC's religion.

Z = A₀ + A₁(x₁) + ... + A_n(x_n) + B(r)

If your fellow traveller is a genuine Quaker, he's not nearly as likely to strangle you as he is if he's really a Thuggee, only pretending to be a Quaker.

It might be fun to break this down further. Religions like Christianity have written rules and unwritten rules. The written rules (i.e. the Bible) are basically the same for Quakers and Presbyterians, but the unwritten rules (i.e. which parts do you emphasize and how do you interpret them) are different. Quakers are famous for being pacifists, or for having been pacifists in the past. Presbyterians are famous for fighting in the English Civil War and beheading Charles I. Different groups of Muslims venerate the same Koran, but not necessarily the same Hadiths. I'm also not sure to what degree I can associate religion with culture. Theologically, I may be a crazed space alien cultist, but culturally, I probably have more in common with Donald Sensing's United Methodists. It's hard to say to what degree the behavior of the Tsarnaev brothers (Boston marathon bombing) is the result of Islam and to what degree it is the result of Chechen culture. Then there is politics. The economics of religion literature (e.g. Iannaccone) tends to attribute religious violence to the political environment. The beauty of the First Amendment in this view is that it tends to prevent religious groups from fighting over government favors, by promising that there will never be much to fight over. So we could distinguish between different groups within the same nominal religion by treating a particular religion as a function of its written (w) and unwritten (u) rules, its cultural context (c), and its political environment (v).

Z = A₀ + A₁(x₁) + ... + A_n(x_n) + B(w,u,c,v)

But for purposes of playing Dungeons & Dragons, it's easier to just treat Quakers and old-time Scottish Presbyterians as having two different religions, represented by the variable name "r".

However, I do still want to consider peer effects in religion. If that fellow traveller in your caravan is a Thuggee, the peer effects are important because the Thuggees won't strangle you until they have enough numbers relative to the other caravan members to win any fights easily. So let's add a term for religious peer effects, p(r).

Z = A₀ + A₁(x₁) + ... + A_n(x_n) + B(r,p(r))

If I were a player rather than a game master, I might use a similar model to estimate the likelihood of being attacked by a stranger, assuming I could figure out what religion he really was. But if I were making an estimate regarding someone I knew, I would throw in an update term, U, to account for the fact that I knew more about what he was like than just the baseline A₀ value for NPCs in general.

Z = A₀ + A₁(x₁) + ... + A_n(x_n) + B(r,p(r)) + U

The application of Bayes' formula in order to do the update is left as an exercise for the reader.

Comments?