During the course of working on my book I have written a couple of outlines that bear witness to how the project has slowly changed. The latest outline is likely to be close to the final version as more than half the book is now written. Those who have seen the previous outlines will see that the basic arc of the book remains the same but many of the details have been shifted around. One important change is that the book is now unambiguously focussed on supernatural beliefs and practices in general, rather than just on superstitions, as previously. In effect, the outline is much truer to what my research has become about during the last couple of years. The change is also visible in the small change to the probable title – In a Mirror, Darkly: How the Supernatural Reflects Rationality.
Chapter 1 Introduction
The introduction spells out the aims of the book, sketches the argument contained within it and locates the issues discussed within the broader context of current research.
The aim of the book is to sketch out some aspects of a thoroughly naturalised picture of human reason, as well as of our religious and magical beliefs and practices and the inter-relationship between them. This particular way of framing the issues is motivated by the belief that, despite many misunderstandings concerning the nature of these phenomena and how they are interconnected, much can be learned by considering both in light of each other.
Being an exercise in naturalised philosophy, the book draws much of its evidence from empirical studies of cognition, especially from cognitive science of religion. The philosophical aspect of the endeavour may, perhaps, be seen most clearly in the effort to develop a coherent large scale picture on the basis of that empirical work, as well as in the way that the whole account ultimately comes to be tied to David Hume’s problem of induction. At the same time, however, the book can be just as well understood as working towards a scientific understanding of magic and religion – the line between the naturalist philosophy of any given science and theoretical work in that science being vague and, frankly, of no great import.
Chapter 2 Reason and Superstition
Starting from the view that reason should be thought of in terms of a perfect ideal, this chapter seeks to provide empirical and philosophical arguments that undermine this position and motivate the search for a naturalist account.
According to many Enlightenment thinkers reason had to be understood in opposition to ‘superstition’ – a broad category that included all those supernatural beliefs that the philosophe in question did not approve of. In its most overwrought form this view saw the two as locked in a Manichean struggle that would inevitably end with the elimination of all superstition and the triumph of reason. Reified in this way, reason was seen as a perfect ideal, quite separate from the fickle and weak human body. In effect, the Enlightenment view of reason was profoundly dualist and antinaturalist – both positions in tension with the overall tenor of Enlightenment thinking and generally rejected since that time.
Surprising, therefore, is the degree to which this ideal of perfect reason has survived till today – the inappropriateness of it being the most striking within the statements of those who would defend the broad legacy of the Enlightenment. Within more scholarly contexts it has most often come to take on the form of a normative value. Even here, however, the concept is highly problematic. While science must traffic in idealisations, not all idealisations are helpful: In the case of the ideal of perfect reason we are dealing very much with an idealisation that is profoundly misleading.
At least two lines of argument are available to showing this conclusion. The first line of argument goes back to the relationship between reason and the supernatural. One empirical prediction that followed from the Enlightenment conception of reason was that, as science progressed, supernatural beliefs should disappear. The truth has turned out to be somewhat more complex, with supernatural beliefs being far from gone in today’s world. This continued coexistence of reason and ‘superstition’ is hard to plausibly explain if reason is understood in terms of a perfect ideal. The second line of argument, I claim, was already presented by David Hume, whose so-called problem of induction I interpret as showing the inadequacy of the Enlightenment conception of reason. The hundreds of years of efforts to provide a solution to this problem have largely been caused by the failure to appreciate that the conclusions it leads to are naturalist rather than scepticist. What had been lacking is a positive account of what Hume called ‘habits’, i.e. a naturalist account of human reason. One way to develop this is by exploring the nature of human beliefs in the supernatural.
Chapter 3 Magic and Religion
The main aim within this chapter is to relate to each other the two categories of supernatural beliefs and practices that are often distinguished – religion and magic – as well as to explain how they differ from other categories of beliefs.
The claim is that both magical and religious beliefs are generally stabilised by cultural and cognitive factors, rather than by empirical factors that stabilise common sense beliefs and, especially, scientific beliefs. This is possible thanks to supernatural beliefs being protected against potential counterevidence by making assertion that are hard to verify, and by social (Durkheim) and technological factors that make the testing of those claims, respectively, unacceptable and impracticable. As a result I come to talk about magical and religious beliefs as superempirical, differentiating the superempirical from the ontological concept of the supernatural on one hand, and the concept of the unfalsifiable (which takes into account only the belief-contents) on the other.
The difference between magical and religious beliefs and practices is argued to consist in the different epistemic status of their intended effects. In the case of magic, the practices (such as rain dances) are intended to affect the mundane world in detectable ways while religious practices (including baptism) are aimed at achieving effects that are, themselves, superempirical. While this distinction cuts across real religions, entailing that they must be understood to be magico-religious complexes (Pyysiainen), it promises to earn its keep by drawing attention to a dilemma that superempirical beliefs face – while the religious beliefs are even less open to counterevidence, they also lack the immediate relevance to mundane existence of the magical beliefs. This dilemma serves as the leitmotif for discussion in the following chapters of the interaction between magic and religion.
Chapter 4 Magic as Cognitive Byproduct
The previous chapter characterised magic in terms of the contents of magical beliefs as well as the social attitudes to those beliefs that, together with the available scientific measures, render them superempirical. In this chapter the focus is upon the cognitive factors that determine the content of magical beliefs when the role of empirical factors is minimised, the aim being to explain how our cognitive system produces magical beliefs as a byproduct (Boyer).
Magical beliefs are argued to originate from the illusory causal connections that our cognitive systems generate in greatest numbers when under stress (Malinowski). As error theory explains (Haselton), while these beliefs are not functional in themselves, it is adaptive for us to generate many such illusory beliefs in order to avoid missing actual causal connections in our environment. Existing discussions of the hyperactive agent detection device (Guthrie, Barrett) and the contagion heuristic (Rozin) are related to this general conception as concrete examples of how cognitive mechanisms generate such illusory causal connections.
Illusory causal connections are not, themselves, magical beliefs proper as they lack that element of a superempirical explanation. Indeed, in so far as any explanations are provided by them, they are often mundane though false in nature. However, at times no mundane explanation appears available so that, if an explanation is sought, only a superempirical one will be psychologically satisfactory. The result is a typical magical belief for which the illusory causal connection is seen to provide evidence. Yet, most people do not form magical beliefs due to having experienced particular illusory causal connections. This necessitates that cultural learning be taken into account. Credibility enhancing displays (Henrich) are practices that are only rational if the relevant espoused beliefs are actually held, making it possible to judge when someone is expressing their true beliefs rather than trying to mislead. Magical (as well as religious) practices are examples of credibility enhancing displays since such practices only make sense if the relevant superempirical beliefs are held. This is particularly significant for religious beliefs since no illusory causal connections can provide psychologically satisfactory evidence for them as the purported effects of religious practices are superempirical. However, in the face of apparent counterevidence, it appears that magical beliefs often come to be reinterpreted as having superempirical effects (Zygmunt, Melton) and thereby become religious beliefs.
Chapter 5 Religion as Prosocial Exaptation
The difference between religion and magic has, thus far in the book, been drawn in terms of the epistemic status of the purported effects of religious vs. magical practices. While this works well with the cognitive byproduct explanation of religion it ignores the work which attempts to explain religion in terms of prosocial function (DS Wilson, Sosis). The aim in this chapter is to bring these two approaches together by developing a dual inheritance model of religion (Atran, Henrich, Norenzayan). In effect, this entails showing the connection between the purported superempirical effects of religious practices and their prosocial function.
Central to the chapter is the idea that religion should be understood as supernatural ideology, allowing the two different accounts of religion to complement each other. As David Sloan Wilson allows, his account of religion has nothing substantive to say about the presence of supernatural entities and forces within religious belief systems – the characteristic that the cognitive byproduct account focuses upon. In effect, Wilson provides an account of all belief systems whose function is prosocial, i.e. all ideologies, be they religious, nationalist or anything else. What Wilson fails to pay sufficient attention to is that the functionality of ideologies is not connected to their truth, unlike that of other beliefs (he goes some way in this direction when distinguishing practical realism from factual realism). The reason is that, while other beliefs have to reflect actual states of affairs in order to allow us to interact effectively with our environment, ideologies simply have to motivate people to act prosocially in order to avoid the free rider problem. In fact, in so far as they need to mislead people about what their own interests are, ideologies have to be false. Given their noncognitive function, it is actually useful for ideologies to be protected against counterevidence as this allows their function rather than their truth-value to determine whether they are stable within a given culture. At the same time, the noncognitive function of ideologies is parasitic upon the cognitive function of other beliefs as ideologies must generally be believed to be literally true in order to motivate behaviour. Religious beliefs are perfectly well placed to function as ideologies because they are superempirical and attractive cognitively thanks to the quirks of the cognitive system discussed in chapter 4. Magical beliefs are less suitable since their purported effects are not superempirical thereby making it possible that functional magical beliefs will come to be destabilised by counterevidence. In other words, the fact that their purported effects are superempirical is what allows religious beliefs to function well as ideologies.
At this point it becomes possible to provide an explanation for why real religions combine magical and religious beliefs. The religious beliefs are the element that allows the religions to have a prosocial function as they, primarily, serve to motivate prosocial behaviour. The magical beliefs, however, are essential to motivate belief in superempirical entities and forces, as can be seen in the relative lack of popularity of deist religions that eschew all magical elements. Even so, the most successful religions tend to have a somewhat ambiguous attitude to the magical elements contained within them for the reason that they are a potentially destabilising element, even though they are necessary. This becomes particularly significant in cultures that value rational criticism.
Chapter 6 The Reason Reflected
Having explored the nature of human belief in magic and religion and the connected practices it is possible to use what has been learned to reflect back upon our understanding of human reason and why religion has often been in conflict with reason.
The essential question to ask is why it is that human cognitive systems produce byproducts. The answer is hard to provide if one thinks of reason in terms of the perfect ideal. Kahneman’s dual-system account attempts to combine biased heuristics with normative rationality and, therefore, might be thought to be offering a solution to this problem, yet it does not provide a happy mix. It is necessary, instead, to accept the view that rationality is bounded (Simon, Gigerenzer, Wimsatt). Given the biased nature of all heuristics, cognitive byproducts become inevitable on this account. Yet, this in a sense only rephrases the original question as: why is human reasoning bounded? The question becomes even more pressing when it is recognised that the traits of human cognitive heuristics that produce cognitive byproducts are shared by all evolutionary adaptations, be they genetic or cultural – a result that should hardly be surprising given the ubiquity of byproducts in evolutionary processes.
To provide an answer it is necessary to go back to Hume and his so-called problem of induction. Given that it was claimed that no solution to the problem exists, and that a naturalist account of reason only seeks to show how we manage to live with it, the so-called problem should, instead, be understood as a fundamental epistemic limit which shapes human cognition. In effect, it is argued that human rationality is bounded because it must make do in the face of the limit that was identified by Hume. Furthermore, it is not just cognition but evolution, also, that faces this limit and is moulded by it. Once this is recognised, it can be seen that the traits evolution and cognition share (Wimsatt), such as their fallibility, are precisely what could be expected of processes that develop in the face of Hume’s limit.
A potential problem for this view is the existence of explicitly held norms of rationality that to a large degree focus on the truth of beliefs and that seem to reach beyond the purely pragmatic normativity evolution seems to admit (Stich, Haack). Yet, even this is readily explained in terms of the developed account. The necessary piece of the puzzle is provided by the argumentative account of reasoning (Mercier & Sperber) which shows that these norms function to allow humans to communicate effectively as they provide another means of judging whether an interlocutor is trustworthy. In short, truth as an explicit norm of reasoning is adaptive when cultural learning plays a significant role.
In the end, it is possible to return to the question of the conflict between reason and religion. At its core lie two adaptations. One is the abovementioned existence of explicit norms of rationality such as truth. The other is the noncognitive function of ideology. Both concern the social sphere but pull in opposite directions. Even if we consider religions from a purely pragmatic point of view, not all need to be adaptive in our current environment nor need they necessarily have effects that we, as individuals, should find desirable. To know if they do, it is necessary for us to examine them. Such examination, however, undermines their superempirical status, the strength of people’s belief in them and, ultimately, their functionality.