Tuesday, June 12, 2012

A response.

Science declined to publish my technical comment (oh well, no surprises there, maybe someday), but it seems a shame to let it go to waste, so here it is.  This is a response to "Analytic Thinking Promotes Religious Disbelief."

Gervais and Norenzayan (1) contended that their research demonstrated a causal relation between induction of analytic thinking and decreased endorsement of belief in God or supernatural agents.  Although their findings are not inconsistent with this conclusion, they failed to demonstrate causation; plausible alternative explanations for the results exist.  In four experiments, participants were exposed to a visual stimulus or completed a task that had previously been shown to increase analytic thinking and then were asked to rate their level of religious belief.  In all cases, the participants in the experimental condition rated themselves as less believing than those in the control condition.  However, because analytic thinking and religious belief were never measured in the same study, it cannot be determined that the manipulation reduced religious belief through the mechanism of increasing analytic thinking.  Instead, the manipulation could have increased analytic thinking and decreased endorsement of religious belief through separate pathways.  I contend that the latter is a more parsimonious explanation for the results of Gervais and Norenzayan’s studies.

For example, in one experiment, participants viewed “artwork depicting a reflective thinking pose” or control artwork.  Pilot data showed that those who saw “thinking” art were more able to override intuitive responses to a syllogistic reasoning task with analytic responses, and in the main study, those who saw thinking art reported less certain beliefs in God.  It is, of course, possible that viewing a reflective thinking pose primed more analytic thinking, which subsequently caused participants to be more questioning of their religious beliefs.  However, it may be that the increase in analytic thinking and decrease in reported religiosity were the results of separate priming pathways: in the case of religious belief, presentation of the thinking art may have operated by activating participants’ associational networks around intellectual pursuits and academia, which are associated with greater secularity, thus causing participants to revise estimates of their belief in the supernatural downward.  Gervais and Norenzayan noted in the supplemental material that only five participants reported noticing a relation between the artwork and the religious belief measures and that these individuals were excluded from analyses; however, such associations need not be conscious to have an effect, as demonstrated by myriad priming experiments (2, 3).  Similar, separate priming processes may have taken place in all of Gervais and Norenzayan’s studies, even those with subtler manipulations. 

This alternative explanation for the results, that the experimental manipulations caused increased analytic thinking and caused decreased reports of religious belief through separate pathways, seems more parsimonious than the authors’ explanation that the manipulations caused increased analytic thinking, which then caused a decrease in belief in the supernatural.  The latter explanation requires that, after being exposed to the prime and seeing the question about their religiosity, participants consciously reevaluated the validity of their intuitions about teleology, mind-body dualism, and so forth, prior to responding to that question.  It seems implausible that such complex cognitions were engaged during such a short period of time and that those cognitions led to conclusions that resulted in less certainty in beliefs.  After all, the question of whether the mind and body are separate has been debated extensively within the highly analytic tradition of philosophy for centuries, and many philosophers have not immediately concluded that dualism is incompatible with logic (4). 

I do not dispute that analytic thinking may have some impact on religious belief.  However, I think it probable that those who are high in trait analytic thinking are less likely to hold permanent beliefs in the supernatural, whereas Gervais and Norenzayan’s work attempts to demonstrate a relation between a transient state of heightened analytic thinking and a likely transient fluctuation in reporting of religiosity.  It is the former conclusion that is consistent with past research cited by the authors showing correlations between cognitive style and religiosity (5), as well as with the author’s first, correlational, study (1). 

However, it remains possible that Gervais and Norenzayan are correct in their interpretation of their results.  However, in order to demonstrate more convincingly that experimental manipulations can induce decreased religious beliefs through increased analytic thinking, future research should administer measures of both analytic thinking and religious belief to the same participants and should test for statistical mediation.  Perhaps an even simpler and more persuasive strategy would be to interview participants after the experiment about what thought processes took place prior to responding to measures of religiosity: after all, if System 2 (i.e., analytic) processing is engaged, participants should be able to report on re-analyzing their beliefs, as System 2 processing is typically characterized as consciously accessible (6). 

Regardless of the outcome of future research, Gervais and Norenzayan are to be commended for their contribution to the literature.  They demonstrate that certain interventions can have an impact on reported religious belief, whether that be through activation of System 1 associative networks, demand characteristics, or System 2 processing.  More importantly, they raise the question of whether simple manipulations can alter thought processes in a way that affects an individual’s core beliefs, a question that has potential implications and applications beyond identifying causes of religious disbelief. 

1.      W. M. Gervais, A. Norenzayan, Analytic thinking promotes religious disbelief. Science 336, 493-496 (2012).
2.      P. M. Gollwitzer, P. Sheeran, R. Tr├Âtschel, T. L. Webb, Self-regulation of priming effects on behavior. Psychol. Sci. 22, 901-907 (2011).
3.      A. Dijksterhuis, H. Aarts, Goals, attention, and (un)conscious. Annu. Rev. Psychol. 61, 467-490 (2010).
4.      R. Warner, T. Szubka, Eds., The Mind-Body Problem: A Guide to the Current Debate (Wiley-Blackwell, Cambridge, MA, 1994).
5.      A. Shenhav, D. G. Rand, J. D. Greene, Divine intuition: cognitive style influences belief in God. J. Exp. Psychol.: General (2011), doi: 10.1037/a0025391.
6.      J. S. B. T. Evans, In two minds: dual-process accounts of reasoning. Trends Cogn. Sci. 7, 454-459 (2003).


  1. I had just got around to reading the original paper and saw your comment on the article's web page. Pity Science didn't publish your response! Issues with correlation vs. causality are so pervasive alas. I also blogged about the original article at pmokeefe.blogspot.com/2012/07/religion-and-analytic-thinking.html - nothing very substantial, just trying to entice people into reading the original contributions.
    Cheers, Patrick

    1. Thanks for the kind words! I imagine Science got a lot of responses to this article, and they're understandably very protective of their space. I'll check out your blog, and I've followed you back on Twitter.