Suggesting that natural science is unnatural and that religion, which traffics in the supernatural, is natural seems to turn things upside down. Sorting out these paradoxes, though, will offer insight about both enterprises.
The "naturalness" of cognition is the sense of the term in play here. Natural cognition is intuitive, fast, automatic, unconscious, and often difficult to articulate. It arises in a flash on the basis of scant evidence and without conscious reflection. Natural cognition is also domain-specific. We immediately detect which things around us have minds; we instantly read others' emotions on the basis of their facial expression, tone of voice, or bodily comportment; we spontaneously comprehend and produce syntactically complex utterances without planning or forethought, and we become careful straightaway about what we breathe and what we touch when we think that an uncontained contaminant is in the vicinity. Although culture tunes many of these maturationally natural systems, no one teaches us these responses. Babies learn to speak their local language, but not because someone instructs them. These natural cognitive dispositions seem firmly in place across cultures by the time children are six or seven years old, and the resulting folk physics, folk biology, and folk psychology constitute part of humans' standard mental equipment.
In contrast, cognitively unnatural thinking is slow, reflective, conscious, and, typically, articulate. Such reflective thinking is deliberate, conjectural, and easier to formulate linguistically. When thought closely resembles talking to ourselves, it exemplifies this measured, reflective, unnatural cognitive processing. Unnatural reflection is laborious and, frequently, the result of considerable instruction. Most of the knowledge that we acquire in school, for example, is the result of this slow form of conscious reflection - though some of it we practice so much that it can become "second" nature. With lots of experience in some domain, our knowledge can take on a sort of practiced naturalness that in most respects mimic those maturationally natural forms of cognition. There are plenty of domains in which most of us develop such expertise, such as reading and writing. With these distinctions in hand between practiced and maturational naturalness of cognition and between both of those kinds of natural cognition and the slow, conscious unnatural cognition, we can now turn to the comparison of science and religion.
Science is cognitively unnatural to the extent that it both depends upon such conscious reflective thought and does not depend upon the dictates of our maturationally natural cognitive systems. The sciences advance, usually sooner rather than later, representations that diverge drastically from the deliverances of our maturationally natural cognitive systems. Pursuing science reveals that the world is not as our folk beliefs would have it. The dominant theories in every mature science are radically counter-intuitive. The earth moves. Solid objects are mostly empty space. Biological kinds are not fixed. People can be unaware that they are blind. Contemporary science generates startling discoveries every week. They either upset our natural folk beliefs or they overthrow some prevailing scientific conception, whose predecessors overthrew such folk beliefs earlier.
The barriers to scientific understanding that render it cognitively unnatural do not stop with the radical counter-intuitiveness of scientific representations. The practiced naturalness of intuitions acquired through many years of dutiful scientific training does not deactivate maturationally natural cognitive systems. Recall that their processing is automatic. If some stimulus incorporates the cues sufficient to trigger one of these system's operation, its deliverances intrude and the empirical research indicates that those intrusions can swamp reflectively attained, scientific judgment under many circumstances.
Learning and doing science requires a couple of decades of formal education and obtaining competence with various forms of inference, including probabilistic and conditional reasoning, which constitute the most fundamental and pervasive types of scientific reasoning. Since humans are susceptible to all sorts of inferential fallacies, the cognitive processing science requires is often no less counter-intuitive than its representations are. Experimental participants routinely violate inferential norms. This is why mathematics, probability, and logic must be taught! Perhaps even more dismaying, research indicates that even experts can frequently go wrong when problems are shifted slightly in a way that is unfamiliar. Science, in short, is difficult to learn and difficult to do in no small measure because it is cognitively unnatural. It is the complex social arrangements that scientific institutions incorporate that compensate for individuals' cognitive weaknesses and limitations and increase the probabilities that the collective outcome in the long run surpasses individuals' efforts in the short run.
All of this stands in striking contrast to the cognitive naturalness of religion. Popular religion engages humans' maturationally natural cognitive dispositions, while in any given setting mixing in but one or two violations of the many default assumptions that come with those dispositions' activation. Contrary to initial appearances and in contrast to the radically counter-intuitive theories of science, the supernatural is reliably quite modestly counter-intuitive. In any particular story, gods and angels and demons and devils violate but one or two of the default assumptions of our cognitively natural conceptions of things. So, for example, angels may have wings and fly (or the ability to appear and disappear), but they are perfectly normal intentional agents with minds like other intentional agents and with bodies that, other than a couple of counter-intuitive properties, are exactly like human bodies. What this means is that, except for those counter-intuitive properties, everything that we assume about human beings can be assumed of them. They have preferences, desires, beliefs, and intentions that inform and direct what they say and do. The gods have kinship relations that give us clues about their loyalties. Although Jesus walked on water, we can also readily infer that he could be hungry, needed to sleep, and carried on conversations.
These ties between religious representations and our natural cognition make popular religion easy to acquire. Once children have the ability to detect agents and anticipate their mental states (theory of mind), language, an appreciation of environmental contaminants, and other natural cognition in place, they are fully prepared to understand popular religious representations and carry out creative inferences concerning them. Experimental research shows that adding one or two counter-intuitive properties makes such representations attention-grabbing and memorable.
Myths and stories are narratives, which means that they are driven by agents, their states of mind, and their actions. If the gods are angry at someone, even a seven year old knows what is likely to follow. The gods have social relations with one another and with us. Normal features of human social interactions - gift giving, for example -- characterize our social relations with the gods, though some transactions with the gods are even more heavily ritualized than human interactions are. Religions are filled with representations that cue other natural systems' operations. By behaving in ways that suggest a contaminant is present (such as decidedly deviating around some space when walking), religious practitioners cue humans' contamination detection systems. Thus, people automatically infer how to conduct themselves around sacred objects and spaces. Many religions include speaking in tongues. These utterances attract attention and, again automatically, initiate linguistic processing. We struggle to parse the utterances and we immediately infer that they are meaningful.
Not all religious thought is cognitively natural. It is perfectly possible to reflect about religious representations consciously and deliberately. Theologians do so routinely. My claim is only that the appeal and ease with which popular religious forms are acquired, retained, and transmitted turns overwhelmingly on deploying modestly counter-intuitive representations that engage our most natural forms of cognition. Science, by contrast, overthrows our folk conceptions of the world, advancing radically counter-intuitive representations and forms of critical assessment that do not come naturally to human minds.
Comparing the cognitive foundations of science and religion has many surprising consequences; here are two. First, science poses no serious threat to the persistence of religion, for religious representations will naturally erupt as normal variations on natural cognition and will possess a ready appeal. Second, it is science, its current cultural prestige notwithstanding, which is fragile, since its ideas and forms of thought are difficult to produce and learn. Such unnatural forms of thought require extensive cultural support, and without it, nothing would prevent the loss of science.