Now suppose I take out a marble at random and show it to you, before putting it back in the jar. If that marble was red, would you feel confident in saying that there were 85 red marbles in the jar? What about if I pulled out another red, and another red? And what about if I then pulled out a blue marble?
This is the game that Michelle Lim (Washington University in St. Louis) and colleagues from two Universities in Melbourne, Australia, played with different groups of people: 25 people with full-blown psychosis, 29 Hare Krishna devotees from the International Society of Krishna Consciousness temple in metropolitan Melbourne, and a control group comprising 23 Christians and 40 nonreligious individuals.
What they found was that, while the control group asked on average for 7 marbles before they were confident enough to say whether the jar was filled with mostly red or mostly blue marbles, the psychotic patients asked for only two.
The Hare Krishna devotees were halfway between the two. They asked for around 4 marbles before making their minds up.
Lim and colleagues ran several similar tests. One in which there were 60 marbles of one colour and 40 of the other. And two tests in which they gave their subjects words to describe an individual (purportedly from a survey of that individuals acquaintances), and asked them to say if the individual was mostly liked or disliked. These words were either positive or negative, with a similar distribution to the red/blue marble task.
All of these tests gave the same basic results. Psychotics required little evidence before coming to a conclusion, while Hare Krishna devotees required more, but less than the control group.
In other words, the Hare Krishna devotees were prone to jump to conclusions.
Lim believes that this is evidence that psychosis runs on a continuum, from psychotic to normal, and that members of new religious movements (like the Hare Krishna devotees studied) lie somewhere on that scale - not psychotic, but with some characteristics that are similar.
The main differences between psychotic individuals and members of new religious movements, she suggests (based on detailed analysis of the results) are that psychotic individuals find it especially hard to weigh evidence related to emotions (the survey task in this study), they have more delusional distress, and have more severe hallucinations and delusions.
But she concludes that the tendency to jump to conclusions based on limited evidence is a real contributor to both having and maintaining delusions. And that probably explains why these people are attracted to new religious movements.
Lim MH, Gleeson JF, & Jackson HJ (2012). The Jumping-to-Conclusions Bias in New Religious Movements. The Journal of nervous and mental disease, 200 (10), 868-75 PMID: 22996398
This article by Tom Rees was first published on Epiphenom. It is licensed under Creative Commons.
Thanks for an interesting article. I recall reading elsewhere that Hare Krishna devotees and New Age believers are more prone to magical thinking than the conventionally religious. These results seem to fit in with.
ReplyDeleteHow many marbles did the Christians ask for (distinct from the "nonreligious" people)? What kind of Christians were they?
ReplyDeleteInteresting to have some more facts, I think, before I'd draw a conclusion that applies only to new religions.
They don't present the data for the Christians and non-religious separately, but they do state that "no significant differences were found between the Christian and nonreligious participants for demographics, JTC responses, and social desirability and IQ scores. Therefore, these groups were analyzed as one control group."
ReplyDeleteIf the jar contains 85 red and 15 blue marbles, the probability that the first marble pulled from it will be blue is 0.15, and the probability that the first two marbles will be blue is 0.0225. The probability that the first four will all be blue is only 0.0005.
ReplyDeleteMaybe the psychotics aren't jumping to unwarranted conclusions after all...
What's the probability that, of the first four marbles, three will be blue and one red? It's low, but not as low as 0.0005. But yes, maybe the problem is that the 'controls' are too cautious!
ReplyDeleteWhat is interesting, though, is that when the scenario changed - to one where it was 60 red and 40 blue (or vice versa) the hare Krishna devotees still needed only 4 marbles to make up their minds. Whereas the controls needed 10 (up from 7).
So it seems as though the number of marbles is not really sensitive to the nature of the problem, for these people at least.
Was their some sort of cost to giving the wrong answer? As blue marbles are drawn, one's confidence that it's a predominantly blue batch ought increase stepwise, not suddenly shift to certainty after a certain number of marbles, so, well, how sure were respondents when they said they were sure?
ReplyDeleteAs far as I can tell from the paper, they weren't incentivised in any way. They just pulled out a marble and asked which jar it was from. And kept going if the subject didn't offer a firm answer.
ReplyDeleteAgree with the others -- it looks like the controls were being too cautious and the psychotics were being reasonable. From a Bayesian perspective, if we know that the jar the marbles are being drawn from is initially equally likely to be either
ReplyDeleteA: 85 red and 15 blue marbles
or
B: 15 red and 85 blue marbles,
then the probability that the marbles are being drawn from jar A is (85/15)^(number of red marbles drawn - number of blue marbles drawn).
If you've drawn two more red than blue marbles, then you're 97% sure it's jar A.
(Sorry, what I said was the "probability" was really the odds ratio, and if the marbles aren't replaced between draws then it's only an approximation.)
ReplyDelete