Home » Posts tagged 'prior probability'

Tag Archives: prior probability

A Better Prior Probability for a Historical Jesus

In Richard Carrier’s book On the Historicity of Jesus (OHJ) he compares Jesus to a modified version of Lord Raglan’s myth-ritualist hero type. He uses this comparison, along with comparisons to other archetypes, to come up with a prior probability for his Bayesian analysis of the two hypotheses mythical Jesus vs historical Jesus. Here I want to examine what Raglan’s type really tells us and it how it functions in the analysis. Then I will argue that there is another type, and associated data set, that has a very similar logic to it but that Carrier has improperly excluded it from his analysis.

Bayesian Aside

Remember, a prior probability is intended to be a measure of our subjective confidence in a hypothesis before we start examining the evidence. In science this will often be driven by where our previous research has led us before we begin collecting data in an experiment. The nature of historical investigation is such that we already have the data, so picking what to single out as our prior probability is somewhat arbitrary.

Despite being subjective we are free (and even encouraged) to look to objective data to help us form our prior. It is still a subjective enterprise however, as we are not going to find a data set that shows how many times our hypothesis turned out to be correct (it either is or it isn’t), and we will have to use our judgment about what is a good proxy. For instance, if we are trying to decide whether a particular American man owns a car, there isn’t a data set on how many times this man was found to own a car vs how many times he wasn’t. He either owns one or he doesn’t. However, data from the US Department of Transportation on adult male car ownership would likely be a good proxy for what our confidence in the hypothesis should be, and data on female car ownership in France wouldn’t.

Raglan as Diagnostic Tool

Lord Raglan developed a set of 22 traits that he claims are common in different hero traditions across cultures. Not just any tradition however. In particular Raglan was looking for story elements that were strongly associated with rituals. As Neil Godfrey once explained:

Raglan has drawn these parallel motifs from what he terms “genuine mythology” — meaning “mythology connected with ritual”. That excludes mythical tales of the King Arthur sort. Raglan is interested in myths that appear to have been associated with ancient rituals as acted out in dramatic shows (e.g. the Dionysia, May Day rituals, Passion plays) and religious ceremonies.

Later in the article he adds:

Lord Raglan’s thesis is that these myths are the product of rituals. They originated as explanations (or even as dramatizations) of ritual ceremonies… The stuff of history (building cities or monuments, expanding the kingdom, etc) is missing. The most eventful moment in some such stories is the king’s inauguration of laws. Historically we know no one person was responsible for introducing complete sets of laws out of nowhere; we are confident that such stories are etiological tales.

While Raglan views these story elements as always mythical instead of historical, what the type is really diagnostic of is whether the hero of the story was celebrated in ritual (or perhaps is symbolic of those who are) or worshipped in religious ceremony. Of course, all sides of the debate already agree on the point that Jesus was worshipped as a god (or demigod). Perhaps some historicists would have to concede that the gospels were primarily written to promote said ritual worship of Jesus (or at least the parts that match the type) but most already concede that point as well. So it would appear that, as a diagnostic tool, the Raglan hero type doesn’t tell us anything we didn’t already know.

Raglan as Sampling Tool

Since the Raglan type is merely diagnostic of a fact already known in the agreed background it’s actual use in Carrier’s procedure is to sample from ancient hero stories. The idea is, if we can build a data set of characters using the archetype, then we can see how frequently historical characters fit the type. Or, in other words, we can see how often historical people are the subject of ritual worship.

Said another way, we are proposing that one of our agreed background facts (Jesus was worshipped as a god) has a different plausibility depending on which of our two hypotheses is correct. If Jesus is invented then it is no surprise he was worshipped because many invented beings are worshipped, but if he was historical, it is a less common occurrence.

Carrier confirms this is his conception of it in a blog post on his website.

Jesus meets even more markers for mythical persons than the Rank-Raglan type: he is, unlike most historical persons, a worshiped celestial savior deity (OHJ, pp. 96-108, 230), a dying-and-rising demigod (OHJ, pp. 168-73, 225-29), a revelatory space alien (137-41, 146, 197-206), a prophecy-fulfilling godman (OHJ, pp. 141-43, 230), an aetiological cult figure (OHJ, pp. 8-11, 159-63), and a counter-cultural hero (OHJ, pp. 222-25, 430-31; cf. Proving History, pp. 131-32).

So what I am measuring is how often historical persons get that heavily mythotyped (and indeed that quickly, which should be near impossible for a historical person: OHJ, pp. 248-52)

So using the Raglan archetype as a sampling tool we have discovered that invented persons are worshiped as gods more often than real persons. We might wonder if this isn’t just another thing we already knew.

At any rate scholars have an answer to this dilemma. The answer is that the closest followers of Jesus were motivated by grief following his death to have visions. I’m pretty sure Carrier agrees this is a fine proposal. I looked to see if I could find a direct quote and couldn’t ,but I’m pretty sure he has said that if Jesus was historical this is definitely what happened and that it is not at all improbable. For his part Carrier has contributed quite a lot of references to the fact that hallucinations and trance states, etc were quite common at the time and people in groups like the early Christians may have even selected for people prone to such things. That grief, especially over a death of someone close, is a strong trigger for such things is also well established.

So what effect should this have on us as we build our subjective prior? Should it cause us to adjust what we have learned from the Raglan data? I think probably so but I will leave it an open question while we examine another collection of data and an argument of a very similar nature.

Rare or Common?

Before I get to that other argument I can’t help but point out that there seems to be some tension between what I quoted from Carrier’s blog and what he says early in OHJ.

It’s quite common for historical persons to become surrounded by a vast
quantity of myth and legend, and very rapidly, too, especially when they
become the object of religious veneration.

OHJ p.18

Carrier makes the statement as a preface to talking about Haile Selassie, a real man, becoming elevated to status of a god. He then goes on to list off nine traits of Selassie that are remarkably similar to Jesus. He also quotes Edmund Standing’s argument that if all materials except the religious materials were lost we would incorrectly conclude Selassie was mythical. Carrier agrees but thinks the difference is that we have Paul where Jesus is a sky god and the Gospels that are so mytho-symbolic that no history can be recovered for them. Of course, much later in book, he ends up allowing someone could view Paul as favoring historicity. As far as the gospels, that isn’t actually a difference. That was Standings point. If only those remained we would think Selassie maybe didn’t exist.

The relevant issue here though is that the entire discussion there seems to be in conflict with what is being done with the Raglan archetype. If the difference between Selassie and Jesus is what we will find when we peel away the layers of the mytho-symbolic gospels, then the things that are supposed to be “peeled away” are precisely the things the Raglan archetype is measuring! So why are we using it? Aren’t we falling to the very trap Standing warned us of (which Carrier agreed with)?

So I’m left puzzled. Is a human worshipped as a god and mythologized quite common and rapid (per OHJ p 18) or rare and perhaps impossible (if it happens quickly) per the blog. Should we be basing our prior on the gospels, or is that the very data we need to peel away? Which view does the 33% prior probability represent? Or is it a subjective balancing of both?

The Not So Desperate Objection

The structure of the Raglan prior probability argument is that there is an element in our agreed background knowledge, that Jesus was worshipped as a god, that can be better explained by one of our two hypotheses, the mythicist one. In short “people don’t worship men, they worship invented beings,” because remember, that Jesus was the subject of ritual worship was all Raglan told us directly.

There is another element in our agreed background knowledge we should look at, that the first Christians thought Jesus was the messiah and that he was executed. As scholars have pointed out, this is also a case that can be better explained by one of our hypotheses, in this case historicity. All the extant conceptions of the messiah before Christianity envisioned a conquering hero. While it is not impossible for someone to buck that trend (as some have foolishly suggested), a historical trigger would surely make that more likely.

Now I want to be clear, I know Carrier made an argument against this view (which he labeled “A last desperate objection”). The Cargo cults arose in socio-political situations quite similar to Christianity so the situation was ripe for an innovation. I thought his treatment of the Cargo cults was quite excellent (in Element 29) and that section certainly cushions the blow and makes the extreme versions of the argument, “no Jew would invent a crucified messiah”, untenable. What it doesn’t do, however, is make the data disappear, and yet it has in Carrier’s analysis. After all, those other conceptions of the messiah arose in the same socio-political context as Christianity. We can only conclude that while it may be possible for such an invented innovation to arise the data show it is a low frequency occurrence at best. Why shouldn’t this low frequency affect our prior?

Compare this to the situation with the Raglan data. We have data that suggests worshipping invented people is more common than real people. We have an argument that explains why Jesus might have been an exception; grief based visions. Notice too that the same Cargo Cult argument also works to situate Jesus differently than other Raglan heroes. So why is a mere argument enough to dismiss the previous messiah conceptions but not the Raglan data?

It appears to me that this a double standard. Either a plausible argument is enough to ignore previously low rates of occurrence or it isn’t. So either both data sets should be ignored and a flat 1:1 prior adopted (and the relevant data considered to be already accounted for) or we should accept both of the low frequencies and acknowledge they point in different direction and probably just cancel each other out, again leaving a 1:1 prior.

This should not sound unreasonable. After all, either way something weird/unusual/rare seems to have happened. Either a group of Jews began worshipping a man, or they invented and worshipped a crucified messiah. But, under Carrier’s treatment, only one of the two hypotheses comes under scrutiny for it’s weirdness. The other one gets a pass and treated as perfectly common. The data says it shouldn’t be.