If you were hanging around ScienceBlogs yesterday, you probably came across this post at Pharyngula. In it, Dr. Myers links to an article on a study by Bushman et al.1 purporting to show that people are more aggressive after reading passages from the Bible in which God sanctions violence than after reading passages that don’t involve sanctioning violence. In the study, two sets of participants, one from Brigham Young University, and the other from Vrije Universiteit in Amsterdam, were told that they would be participating in two seprate studies. After being asked a few questions about their religious beliefs (their religious affiliation, whether they believe in God, and whether they believe in the Bible), they completed the first “study,” in which they were told was on Middle Eastern literature. This first task involved reading a passage from the Bible (Book of Judges, Chapters 19-21), which involves a man taking revenge for the rape and murder of his wife. In one condition, participants were told that the passage came from the Bible, and in a second condition, they were told that it came from a scroll discovered in the Middle East in 1984. In each of these conditions, half of the participants read a version of the story that contained this verse:
The Lord commanded Israel to take arms against their brothers and chasten them before the Lord.
The other half read the story without that verse. It doesn’t appear, from the paper, that the participants had to do anything but read the passage.
After reading the passages, participants were told that they were to participate in an unrelated study in which they would compete against other participants in a reaction game. In the game, two participants (one actually a confederate) were told that they would be competing to see who could press a button faster on 25 consecutive trials. The slower participant would be punished with a noise in their ears. Participants were then asked to set the decibel level of the noise that their competitor would hear if he or she was slower. Presumably, more aggressive participants would choose higher decibel levels. The prediction, then, is that participants who were told that the story they read was from the Bible, and who read the the version with the verse in which God sanctions violence, would choose higher decibel levels than participants who were told that the passage was from a recently discovered scroll, or read it without the violence-sanctioning verse.
Bushman et al. then compared the number of trials (out of 25) in which participants in the different conditions chose the highest decibel levels. As predicted, in both sets of participants (BYU and Vrije Universiteit), participants who were told that the passage was from the Bible chose the highest decibel levels on more trials than those who were told that it was from a recently discovered scroll (though this difference only approached statistical significance for the Vrije Universiteit participants). Participants in both sets who read the version with the verse about God sanctioning the violence in the story also chose the highest decibel levels more than those who read the version without that verse. In the BYU sample, virtually all of the participants believed in God in the Bible, but in the Vrije Universiteit sample, only 50% indicated that they believed in God, and only 27% said they believed in the Bible. So in the Vrije Universiteit sample, Bushman et al. also looked at differences in aggression (decibel levels) as a function of belief. Overall, believers chose the highest decibel levels more often than non-believers, but this was largely due to the greater influence of the Bible passage on believers than non-believers. That is, while both believers and non-believers were more aggressive after reading the passage containing the violence-sanctioning verse, and told it was from the Bible, than after reading the passage and being told it was from another source, this difference was greater for believers than non-believers.
What conclusion should we draw from this study? Well, Bushman et al. write:
This work also supports theories proposed by religious terrorism scholars who hypothesize that exposure to violent scriptures may induce extremists to engage in aggressive actions (e.g., Juergensmeyer, 2003). Of note here is that we obtained evidence supporting this hypothesis in samples of university students who were, in our estimation, not typical of the terrorists who blow up civilians. Even among our participants who were not religiously devout, exposure to God-sanctioned violence increased subsequent aggression. That the effect was found in such a sample may attest to the insidious power of exposure to literary violence.
Except, well, I don’t know. For one, in no condition in either of the samples did participants tend to choose the highest decibel levels on even half the trials (the highest mean, for any condition in either sample, was 8.7 trials out of 25), and the increase in aggression from the non-sanctioned to violence-sanctioned conditions ranged from around 2 trials to 4 trials out of 25. So it’s not like reading the passage suddenly made people crazed terrorists. It may of course be true that repeated exposure to violent religious passages, and only the violent ones, can lead to sustained and practically significant changes in aggression levels, but that’s not what this study shows. Instead, it shows that reading that God sanctions retributive violence makes people a little more aggressive temporarily.
Interestingly, another recent study (which, I might add, you won’t find discussed on Pharyngula) presents evidence that religion may have the opposite effect as well. That is, religion may increase prosocial behavior. One of the long-standing theories of the origin and purpose of religion is that, in large groups that are not held together by kinship relationships and reciprocal altruism necessitated by close physical proximity, religion serves to promote prosocial behavior and discourage cheating. In a recent version of this theory, Ara Norenzayan and his colleagues have hypothesized that religions achieve this function by activating our hair-triggered agency detectors, which thereby activate “reputational concerns.” That is, when we feel like someone’s watching, we don’t want to cheat, because it will harm our reputation in the community. Religions almost always include supernatural agents that are omnipresent, and can see what you’re doing even if you can’t see them. In other words, we’re being watched, and that activates our reputational concerns, causing us to act in more prosocial ways.
To test this hypothesis, Shariff and Norenzayan2 conducted two studies. In the first, they first had participants complete a scrambled sentence task. This task, which previous research has shown can implicitly prime concepts (i.e., prime concepts without participants being aware of them), involves having participants read scrambled sentences with one extra word, like “felt she eradicate spirit the,” and asking them to put the words in the correct order, leaving out the extra word. Thus, from “felt she eradicate spirit the,” participants should get, “she felt the spirit.” Each participant unscrambled ten sentences. For half of the participants, five of the sentences contained words that prime God concepts (“spirit,” “divine,” “God,” “sacred,” and “prophet”) while the other half unscrambled sentences that didn’t prime any single concept. After the priming task, participants were told that they were going to participate in an unrelated experiment involving the “Dictator Game.” The instructions for the game were as follows:
You have been chosen as the giver in this economic decision making task. You will find 10 one dollar coins. Your role is to take and keep as many of these coins as you would like, knowing that however many you leave, if any, will be given to the receiver participant to keep. (p. 6)
Since the amount of money that the receiver gets is entirely up to the giver, and since the participants were always the givers, Shariff and Norenzayan could measure the effect of priming God concepts on prosocial behavior by comparing the difference in the amount of money participants gave depending on whether they’d unscrambled the sentences with God-related words or only neutral sentences. As their hypothesis predicted, participants primed with God-related words gave more on average ($4.22) than those who unscrambled the neutral sentences ($1.84). Furthermore, more than half (64%) of the participants in the God-prime condition gave at least half of the money to the receive, while only 12% of the participants in the neutral condition gave half or more. Interestingly, they found that activating God-concepts increased the amount of money participants gave to the receivers regardless of whether they were theists or atheists.
The second study involved the same tasks (the scrambled sentence task and the Dictator Game), but used a more diverse group of participants (larger age and income ranges), and included a third scrambled sentence task condition. In addition to the neutral and God-concept conditions, they also included a condition in which five of the sentences contained words that primed secular moral concepts (“civic,” “jury,” “court,” “police,” and “contract”). As in the first study, participants gave more, and were more likely to give half or more of their money, in the God-concept condition than in the neutral condition. The same was also true in the secular moral prime condition. Participants who unscrambled the sentences with secular words gave more than those in the neutral condition. The difference between the God-prime and secular-prime conditions was not significant. In this experiment, atheists in the God-prime and neutral condition gave the same amount of money (very little; friggin’ selfish atheists), in contrast to the first study. Shariff and Norenzayan believe that this difference between the two studies was the result of stricter criteria for determining whether participants were classified as atheists in the second study.
So, activating God concepts causes people to behave in a more prosocial fashion. While this isn’t direct evidence for the agency-detection theory of Norenzayan and his colleagues, it is suggestive. The fact that secular moral concepts had the same effect on prosocial behavior could be a result of similar processes, with the government or its agents (e.g., the police) replacing supernatural agents.
Perhaps the message to take home from these two sets of studies (the Bushman et al. and Shariff and Norenzayan studies) taken together is that religion, like any other social institution, can cause good and bad behavior, depending on the context and the ways in which it is used. Overall, religion and similar secular institutions may serve to promote prosocial behavior, but when individuals focus on certain parts of a religion’s or government’s message, aggression and violence can also result. As is usually the case in the social sciences, the role that religion plays in society and the individual psyche is complex and messy.
1Bushman, B. J., Ridge, R. D., Das, E., Key, C. W., & Busath, G. M. (In Press). When God sanctions killing; Effect of scriptural violence on aggression. Psychological Science.
2Shariff, A., & Norenzayan, A. (In Press). God is watching you: Supernatural agent concepts increase prosocial behavior in an anonymous economic game. Psychological Science.
“The Eyes Are Watching You” photo from here.