Natural Explanations of Religious Belief

The previous article closed with a question it did not answer. If a belief has a comforting cause, and the inference from a belief's origin to its falsity is a recognised mistake, what is the point of explaining where religion comes from?

There is a point, and locating it precisely is most of the work. The explanations on offer have improved enormously in the last thirty years, and what they establish is narrower than their popular use suggests and considerably more than nothing.

The discipline

The reason it is a mistake is that causes and truth-values are independent. Someone who believes there is a snake in the grass because they are terrified of snakes may be right about the snake. Kekulé's account of arriving at the ring structure of benzene involves a daydream, and the structure is what it is. A belief acquired for terrible reasons can be true, and a belief acquired for excellent reasons can be false.

Now the qualification that the rest of this article depends on, and it is the reason the fallacy is not the end of the discussion.

An account of a belief's origin cannot show the belief false. It can bear on whether the believer is justified in holding it. Those are different properties, and confusing them is what generates the impression that any appeal to causes is fallacious.

An illustration makes the difference clear. Suppose you find that a friend's confident view about a company was implanted by a drug rather than by evidence. You have learned nothing about the company. You have learned a great deal about whether your friend's confidence is worth anything, and if you were relying on their confidence as your reason, you have lost your reason.

So the useful question about the material below is never whether it shows religious belief false, because it does not and its practitioners generally say so. It is whether it removes a support that the belief was resting on.

The nineteenth-century accounts

Ludwig Feuerbach set out the version everything else descends from, in Das Wesen des Christentums (1841), translated by Marian Evans under the name George Eliot as The Essence of Christianity in 1854. Feuerbach's claim is that the predicates ascribed to God are human predicates, taken to their limit and placed outside their owner. Knowledge, power, and love are ours; the idea of God is the idea of those raised to infinity and made into a separate person, and worship is therefore the species contemplating its own nature and not recognising it.

Feuerbach did not present this as a demolition, which is worth knowing since he is usually cited as one. He took himself to be preserving the content of religion while relocating it, so that what had been said about God was retained as true about humanity.

Two developments followed and moved the mechanism to different places.

Freud, in The Future of an Illusion (1927), located it in the individual psyche and in the residue of childhood dependence on a protecting father, calling religious beliefs illusions in a technical sense, meaning beliefs whose adoption is driven by wish rather than by evidence, without prejudging whether they are also false.

Durkheim, in Les formes élémentaires de la vie religieuse (1912), located it in society. What is worshipped, on his account, is the collective itself, which really does exceed each individual, really does impose obligations, and really does outlast them, and the sacred is what society has set apart.

These accounts are important historically and they have a common weakness. They were arrived at by reflection rather than by investigation, they specify mechanisms nobody had tested, and in the form their authors gave them they are hard to falsify, since any religious phenomenon can be redescribed as projection after the fact.

Where they did make a prediction, the record is mixed rather than favourable. Each implies that religion should recede as education spreads, insecurity falls, and social forms change. In much of western Europe that has happened. Globally it has not, and the association between religiosity and insecurity holds well enough to keep the general idea alive while falling well short of what these theories claimed.

The empirical successor

The central move is the refusal to treat religion as a thing needing its own explanation. On this approach there is no religion module and no religious instinct. There are ordinary capacities, developed for detecting agents, attributing mental states, inferring purposes, and remembering and transmitting information, and religious concepts are what those capacities produce when they operate in the way they normally do.

Stewart Guthrie's Faces in the Clouds (1993) set out the anthropomorphism version: we take the human case as a default model and read humanness into things that do not have it, because that model is the one we know best and the one that matters most.

Pascal Boyer's Religion Explained (2001) contributed the transmission side. Concepts that violate one or two of our expectations about the kind of thing they are, while behaving normally in every other respect, are remembered and passed on much better than concepts that violate nothing and better than concepts that violate everything. A statue that hears prayers is memorable. A statue that does nothing is forgettable, and a statue that exists on Tuesdays, has no location, and is made of grief is not transmissible at all. Religious concepts cluster tightly in the narrow band that is optimally memorable, and this is a claim about what survives cultural transmission rather than about what anybody chose to believe.

Related findings concern children: a readiness to attribute purposes to natural objects, and intuitions about minds and bodies that make the idea of a mind persisting without a body easy to entertain. Whether these amount to what has been called an intuitive dualism, and how much they explain, is disputed.

Justin Barrett named the device and developed it, later preferring "hypersensitive" to "hyperactive" to cover a wider range of triggers, in work including Why Would Anyone Believe in God? (2004).

The argument for the bias is about asymmetric costs. Treating the wind in the grass as a predator wastes a little energy. Treating a predator as the wind ends the lineage. Under that asymmetry, selection favours a system that fires readily and is wrong most of the time, and the proposal is that a system tuned that way will also produce agents where there are none: in the storm, in the illness, in the run of misfortune.

Honesty about the state of the field is required here, because the popular presentations are more confident than the researchers. The cognitive science of religion is roughly three decades old. Agency detection in particular has faced empirical challenge, with experimental work questioning whether threatening conditions produce the predicted over-detection and whether the device does the explanatory work assigned to it. The by-product framework is well motivated and parts of it are well supported; the specific mechanisms are live hypotheses rather than results, and anyone using them in an argument should say so.

What follows

The structure avoids the genetic fallacy by targeting justification rather than truth, as the first section set out.

Sharon Street gave the model case for morality in "A Darwinian Dilemma for Realist Theories of Value", Philosophical Studies 127 (2006), pp. 109-166, arguing that if our evaluative attitudes were shaped by selection, and selection tracks reproductive advantage rather than moral truth, then the realist owes an account of why the two should have coincided, and has no good one available.

Applied here, the argument runs: religious belief is produced by mechanisms whose operation is fully explained without reference to any god; a mechanism that would produce the belief whether or not the belief were true provides no evidence for it; so belief formed that way is unsupported.

Two conditions have to hold for this to work, and both are contestable.

The mechanism has to be insensitive to the truth of what it produces. This is the strong step. A mechanism can be caused, evolved, and reliable all at once: vision is a product of selection, has a complete natural explanation, and tracks the world, which is why the account of its origin does not debunk what we see.

And there has to be no independent route to the belief. A debunking explanation of how someone first came to a view does not touch the view if they now hold it for reasons that stand on their own. This is the limit that matters most for this subject, since the eleven preceding articles consisted entirely of such reasons. The debunker's answer is that the arguments were built to support a conclusion already held, which is the objection from the theodicies article reappearing, and it is answered the same way: by examining the arguments.

The replies

Barrett's own response is the one to take seriously, partly because he is a practitioner rather than a critic of the field. If God intended people to be capable of recognising divine action, then equipping them with a strong disposition to detect agency is a reasonable way to have done it, and finding the mechanism no more debunks its output than finding the visual system debunks sight. A cause and a reason are not competitors.

The limit of this reply should be stated with the reply. It is available only to someone who already holds that there is a God who designed the mechanism. It defends theism against the debunking argument; it does not offer any support for theism, and using the discovery of the mechanism as evidence for design would be helping oneself to the conclusion.

The second reply is that the argument generalises past where its users want to stop. The same structure applies to moral beliefs, which Street pressed deliberately, and can be extended to mathematical beliefs and to beliefs about the external world, none of which selection shaped for the purpose of getting things right. A debunker who wants to keep those and lose religion owes an account of the difference. There is a good answer available, which is that we have independent checks on the deliverances of vision and arithmetic that we do not have on reports about gods, and that answer is the interesting reply rather than a dismissal of the generalisation.

The third reply holds that religious belief can be justified without being based on arguments or on the deliverances of a mechanism, which changes the question rather than answering it. That is a substantial position with a substantial literature, and the advanced tier takes it up.

Next

That completes the intermediate tier. The arguments for and against have been set out, the internal tensions examined, and the natural history of the belief considered.

The advanced tier takes up what has been repeatedly deferred: the accounts of God's nature that the classical tradition developed to keep the attributes consistent, the positions that abandon parts of that tradition instead, the question whether religious belief needs arguments at all, and the treatments of evil and divine hiddenness that go beyond anything attempted here.