Is religious belief & behaviour an evolutionary adaptation? There are two main camps. One that holds religious belief and behaviour (hereinafter religion) as an evolutionary adaptation that helps show commitment to a group through costly rituals and thereby gain access to group membership and resources. Religion has also been postulated to increase the stability of large groups through sacralization of group norms. However, Pascal Boyer argues that there is no reason to think that all thoughts and practices we considered religious originated simultaneously and neither can one factor be singled out as fundamental to religion. His review paper provides a more plausible (and more accepted) model that frames religion as an incidental by-product of the development of certain cognitive functions of the brain that possibly had non-religious evolutionary purposes (2003).
According to Boyer, the joint activation of these mental systems renders notions of supernatural agency intuitively plausible. However, this incidental activation of these systems does not by itself result in the religious experience. The salience of religious concepts encourages the attribution of this state to religious concepts (Boyer 2003, Azari, et al., 2001). He reframes various concepts within religion in terms that show how they are enabled by regular non-religion specific cognitive processes (Boyer, 2003). He suggests a second step to this reduction which correlates these cognitive processes to patterns of neural activation. The evidence for which is scant in his review paper. Therefore I subsequently look at an original research paper (Azari, et al., 2001) that uses functional neuroimaging to investigate the neural correlates of religious experience to see if it corroborates Boyer’s model.
There are six concepts within religion that he reframes, each of which depends and builds upon the previous one. The first is regarding the naturalness of religious belief. This is posited as a by-product of an evolutionary adaptation that is geared to remember minimally counterintuitive concepts. A minimally counterintuitive concept is easily stored and recollected as it combines default expectations for a domain with special features that violate such expectations. It thus forms a memory optimum, that helps humans remember important details like which fruits are inedible/poisonous. This system is likely to also propagate and make salient belief in the supernatural which is an example of such an exception to the rule.
Second, religion entails belief in the presence of intentional agents that one does not encounter. Far from being irrational, the ability to attribute agency to intangible entities is a crucial feature of human cognition. We often run such off-line interactions to imagine past/future scenarios or counterfactuals with other agents. One crucial difference between these off-line interactions and ones with the supernatural is that the former is known to be imagined while the latter is taken to be quite real due to the salience of religious concepts in culture.
Third, supernatural beings are moral legislators. This belief is a result of our intuitive ability for empathy resulting from the ‘theory of mind’ – the ability to attribute mental states and intentions to oneself and others. A tweaked ‘theory of mind’ allows us to posit supernaturals as perfect-access intentional agents who know all our emotions, intentions and actions. This becomes useful as we often have moral intuitions which we cannot explain. In such situations, our moral frameworks become more cognizable when they are viewed as judgements of another agent.
Fourth, gods cause or prevent misfortune, including death. Humans have difficulty in estimating the probability of single events (like a flood). This explains why these events are seen as remarkable. The attribution of this event to an invisible intentional agent is explained through the ‘logic of social exchange’. Humans mistakenly attribute the notion of social exchange between agents, which we use in our regular interactions, to events of misfortune. Again, the salience of religious concepts makes attribution of agency to the supernatural god more likely.
Fifth, our dead ancestors look over us from the afterlife. The belief in the presence of ancestral spirits is said to come from a disjuncture between animacy-determination systems and social-intelligence systems. Death does not turn off regular social-intelligence systems, like facial recognition and ‘theory of mind’. This makes us prone to attribute supernatural agency to the dead. However, the functioning of systems involved in detecting animacy and contagion-avoidance enable us to recognise the dead body itself as non-intentional, dangerous and impure.
Sixth, performing certain rituals for a supernatural agent can help avoid misfortune. Rituals are proposed to be a natural consequence of the operation of hazard-avoidance and contagion-avoidance systems. These systems are automatically activated when encountering a decomposing body or an odd weather phenomenon for instance and help deal with the anxiety over the imminence of danger. The anxiety mitigating nature of such rituals is likened to symptoms of psychopathology like OCD and is explained through its neural activation pattern. For instance, neural systems dedicated to the detection and avoidance of invisible hazards are activated when performing rituals. By far, the ritualization of religion is provided with the only substantial neuroscientific explanation in the entire review paper.
Boyer provides a table summarising the two-step reduction of various religious concepts but it is unsubstantiated and inconclusive. It acts as a proof of concept more than anything else. He acknowledges the need for further research in neuroscience to see if these assumptions about cognitive processes hold up to scrutiny. Here, the original research by Azari et al. proves insightful. The objective of the study was to determine whether religious experience (RE) is a preconceptual, immediate affective event or an attributional cognitive phenomenon. Six self-reported religious and non-religious subjects were recruited for the experiment each. The participants performed reading and recitation tasks to operationalize either a religious state, a happy state or a neutral (control) state., All subjects rated (on a scale of 1-10) whether the target state had been achieved in each condition. PET scans were taken for each condition to derive the neural correlates of the religious experience.
The findings showed that the religious state and happy state were achieved and sustained only by the religious subject x religious-recite condition and non-religious subject x happy-recite condition respectively. Both groups achieved the neutral state. The PET scans revealed statistically significant differences in the activation patterns between the religious subject x religious-recite condition and others. There were certain additional activations for them in areas engaged in cognitive processes, while rCBF changes were not detected in the limbic system. The non-religious x happy-recite condition, however, did show significant amygdala activation, indicating that the religious experience is not a preconceptual, immediate affective event.
When religious individuals achieved and sustained the religious state, there was significant activation in the dorsolateral prefrontal, dorsomedial frontal and medial parietal cortex. In the human brain, the prefrontal cortex is the locus of schemas, including that of religion. The dorsolateral prefrontal cortex is important for memory retrieval (including those relevant for schemas) and the conscious monitoring of thought. The medial parietal cortex is involved in visual memory, the activation of which is possibly a result of recitation from memory. Pre-SMA is the part of the dorsomedial frontal cortex that was activated. It is involved in automatically controlling and sustaining the readiness to act based on the contents of working memory. It also receives strong inputs from the prefrontal cortex. The author argues that in the presence of religious cues (the recitation of the religious text), the activation of the pre-SMA aids retrieval of religious schemas and a conscious attribution of the experience to the schema. This suggests that religious experience is a cognitive process mediated by pre-established neural circuits. This inference seems to mimic Bayer’s claim that supernatural agency is intuitively plausible when religious concepts are salient due to the joint, coordinated activation of diverse mental systems that are present in non-religious contexts as well.
Consequently, religion does not seem to be an evolutionary adaptation but a by-product of other cognitive processes that could very well have been naturally selected for. However, this is an incomplete truth, as pointed out by Powell and Clarke (2012). The pluralist model of religion holds that although religion may have emerged incidentally, it is not merely a byproduct or spandrel. It’s proliferation and ubiquity in society could be a consequence of certain evolutionary advantages that one or more of its aspects provided. This kind of co-option of a trait into the evolutionary framework is called exaptation. Taken together, this is by far the most sensible answer to my question that I could find.
Atran, S., & Norenzayan, A. (2004). Religions evolutionary landscape: Counterintuition, commitment, compassion, communion. Behavioral and Brain Sciences, 27(6), 713–730. doi: 10.1017/s0140525x04000172
Azari, N. P., Nickel, J., Wunderlich, G., Niedeggen, M., Hefter, H., Tellmann, L., … Seitz, R. J. (2001). Neural correlates of religious experience. European Journal of Neuroscience, 13(8), 1649–1652. doi: 10.1046/j.0953-816x.2001.01527.x
Boyer, P. (2003). Religious thought and behaviour as by-products of brain function. Trends in Cognitive Sciences, 7(3), 119–124. doi: 10.1016/s1364-6613(03)00031-7
Elk, M. V., Rutjens, B. T., Pligt, J. V. D., & Harreveld, F. V. (2014). Priming of supernatural agent concepts and agency detection. Religion, Brain & Behavior, 6(1), 4–33. doi: 10.1080/2153599x.2014.933444
Fingelkurts, A. A., & Fingelkurts, A. A. (2009). Is our brain hardwired to produce God, or is our brain hardwired to perceive God? A systematic review on the role of the brain in mediating religious experience. Cognitive Processing, 10(4), 293–326. doi: 10.1007/s10339-009-0261-3
Powell, R., & Clarke, S. (2012). Religion as an Evolutionary Byproduct: A Critique of the Standard Model. The British Journal for the Philosophy of Science, 63(3), 457–486. doi: 10.1093/bjps/axr035
 A more refined explanation of this probability estimation problem can be found in other literature, that is based on a Hyperactive Agency Detection Device (HADD). The HADD is said to be an evolutionary adaptation that biases humans towards false positives over the risk of false negatives while detecting the presence of lurking predators (Elk, Rutjens, Pligt, & Harreveld, 2014). This makes us prone to detect an invisible intentional agent in an event simply because the cost of a false negative could be life-threatening in the future.
 Original research on agency detection does indeed show that religiosity was a moderating factor that inclined people to detect agency in certain ambiguous stimuli when it is preceded by a supernatural agent prime (Elk, Rutjens, Pligt, & Harreveld, 2014). This points towards the importance of internalised cultural norms in this process.
 For the religious state, all participants read the biblical verse Psalm 23. For the religious participants, Psalm 23 was essential for their personal religious experience as converts. For the happy state, all participants read a rhyme, and for the neutral state, all participants read instructions from the back of a calling card.
 The religious subjects were asked (and agreed) to self-induce and maintain a religious state throughout the duration of the experiment. This was reportedly because transient inducements of the religious state for just the religious-read and religious-recite conditions was seen as offensive and antithetical to the religious experience by the religious participants.
 This measure is self-assessed because there is no viable subject-independent measure of this criterion.
 They also took a self-reported PANAS test before and after each PET scan in order to evaluate the affective impact of the experience but it yielded no statistically significant results.