Defining Religion


            Most popular discussions comparing science and religion are crippled by the tendency to overlook differences in the uses of certain words, particularly the words “religion” and “belief”. This leads people to talk past each other without any chance of progress. I hope to show how attending to the different uses of these words can dissolve much needless disagreement.

            My thoughts today were inspired by Judith Shulevitz’s review of Daniel Dennett’s book “Breaking the Spell”. She criticizes Dennett’s ways of delimiting religion, which initially involves looking for “avowals of belief” in causation by supernatural entities. Dennett wants to rule out namby-pamby religion (or by his criterion, pseudo-religion) that tries to get by with talk of values, meaning or social utility. Literal belief in gods who act “in real time” is for him the essence of religion – so much so that even Christianity and Judaism (which are known to have some adherents who read their sacred texts metaphorically) seem barely to qualify.

Shulevitz argues that religion can’t be delimited by what people say they believe, because they might not really believe what they say; but neither can we use “real belief” as a criterion, because it can only be deduced from behavior, and we have no way of reliably distinguishing religious from non-religious behavior. She quotes Dennett himself as saying that we end up with no criteria beyond “behavior” (presumably excluding linguistic behavior), because religious statements are all beyond observation or experiment. Neither of them notices that this claim already conflicts with the demand that religious belief be about causality “in real time”: if religious belief were really as narrow and concrete a thing as Dennett demands, it would be susceptible to arguments based on observation and experiment. Indeed it is so susceptible, when it claims that the earth is 5000 years old, for example. If it sacrifices an ox to bring rain, and it doesn’t rain, then on Dennett’s interpretation of the meaning of the act, it has been experimentally disproven.

I was sympathetic to Shulevitz’s treatment, especially when she mentioned Wittgenstein. I happen to be a fan of his writings on religion, which place him squarely in the namby-pamby corner disdained by Dennett. Wittgenstein wrote about the same “primitive” religious behavior on which Dennett wants to focus, but he denied that it hinged on belief. He noted, for example, that rituals aimed at making the sun “return” are held on the winter solstice, not some earlier time that might be more convenient. Oxen are usually sacrificed to bring rain only at the beginning of the rainy season – that’s how the “meme” “survives”. And the very fact that one class of beliefs is tied to confirmation/disconfirmation, while another class is explicitly removed from it, was for Wittgenstein evidence that the word “belief” isn’t being used the same way in the two kinds of cases.

But in the end Shulevitz ignores what I thought was Wittgenstein’s most influential idea, something he called “family resemblance”. He argued that the meaning of a word need not depend on its picking out a single unique phenomenon. All the things called X may share some properties with some other things called X, without there being a single property of Xness that they all share. In my view the word “religion” is an exemplary candidate for such understanding. But Shulevitz eschews it. She is focused on the idea that religion must be well-defined before science can study it.

This leads her to a false dichotomy, based on a false assertion. She says that religion must either be defined in terms of belief (as Dennett has tried and failed to do) or experience; and she falsely claims that “philosophers” have chosen the latter. The result for her is that because “an experiential definition of religion renders it impervious to empirical observation”, religion and science remain walled off from each other, following a distinction made by Kant.

Most recent discussions of this topic would here rely on S.J. Gould and his two “magisteria”. But let’s stay with Kant. From Dennett’s perspective, Kant was another namby-pamby “believer in belief”. Kant taught that God is a necessary illusion. Not only does the idea of God motivate those who take it literally; those who can understand the purely rational basis of morality understand God as a metaphysical postulate that follows from the nature of morality. Far from being a matter of having certain experiences, it is a matter of pure understanding. It is “subjective” only if you think that we can never agree about morality, and that philosophy is a waste of time. But in any case, morality (and consequently religion) is separate from science, not because it deals in mere feelings, but because it is not developed through causal-explanatory means.

Without agreeing that morality is completely innate, you can still acknowledge that moral character isn’t pursued scientifically. Insofar as its pursuit involves trial and error, it isn’t based on determining the correspondence of propositions to facts, but on assessing and prioritizing competing strategies, and mutually adjusting strategies and goals. The wall between science and morality (and thus religion) is a wall of logic and grammar. It is like the “wall” between integers and real numbers. Trying to break it down is simply misguided.

So let’s return to the definition of religion. Although a scientific investigation would seem to have to start with a descriptive rather than a normative definition, Dennett’s investigation has more than descriptive aims. Consequently he seeks an essence so restricted that it ends up excluding not only Buddhists (always a problem for polemical atheists) but even Jews and Christians. Shulevitz opposes to this the “subjective” essence of Kierkegaardian religion; but Kierkegaard denied that any of the people calling themselves Christians were in fact Christians. Countless more examples could be assembled in which someone with a clear vision of what religion “really” is defines away 99% of what is ordinarily called religion. The problem is that, as in Wittgenstein’s example of games, there is simply no one property common to all the things that are in fact called religious.

For some people, religion really is that empty avowal that Dennett wants to downplay. The avowal may fit in with all kinds of concerns, from pleasing Mom to furthering a business deal or getting a date. It may be completely secondary to the lure of the food, the holidays, the stories, the music … or just belonging to a community. At a moment of distress the avowal (or the music behind it, the presence of friendly faces, etc.) might have psychological utility. It might then be avowed with a different feeling. There can be no doubt that many people take their religious language literally. But there is just as little doubt that the saints, mystics and religious philosophers agree that you shouldn’t take it literally. Using God’s “existence” as a touchstone is absurd when confronted by all those who declare Him beyond being and non-being. And it is just as absurd to define religion by the most extreme and rare kinds of experience, however heroic the characters (St.Teresa? Rabbi Akiba? Walt Whitman? Baba Ram Dass?) one posits as having had them.

Religion is not one thing. But all the things we call religion are connected in multiple ways to many other things we also call religion. Even within a religion there can be complete disagreement. The pacifistic communists of early Christianity would not have recognized the Christianity of Constantine or the Crusaders (not to mention George W. Bush). Therefore Shulevitz is right to point out that analyzing the survival of a “meme” itself is problematic, because there is no criterion of identity. What can be analyzed in terms of survival-value for human groups are things like dietary, agricultural and judicial practices – each of which may or may not continue to be associated with the religion or meme under analysis (while that meme may well shed and take on different sets of cultural baggage).

Neither beliefs nor experiences capture the essence of religion, because there is no essence. But a lot of trouble would be saved just by considering some of the ambiguities of the leading candidate for that essence, “belief”. For starters, please make a distinction between believing that something is the case and believing in a person, institution or ideal. The latter use of the word, besides indicating one’s preferences, has to do with things like loyalty, trust and resoluteness. It makes sense to try to take on these characteristics, or to actively make a commitment; therefore “faith”, or “belief-in…”, can itself be a virtue. But “belief-that…”, i.e. “belief” in the simple cognitive sense, isn’t a matter of will: I can’t make myself believe that Kerry is President, even if I may still believe in him. For many people, believing in God is tantamount to wanting to be a good person, wanting to be an upstanding member of the community, or wanting the world to be a better place. Unfortunately there is a mountain of rhetoric devoted to blurring the distinction between desire and cognition, insisting that belief is belief is belief.

            The bottom line is that, in this era of “culture war”, many non-religious people are spoiling for a fight, and have seized on the most superstitious assertions of some religious people as if they belonged to the essence of religion. Some religious people are so insulted by this misinterpretation that they deny that superstition has anything to do with religion. Both are plainly wrong. The person who coldly kills animals to make it rain and the tender soul who writes mystical poetry are both religious – even though they really have nothing in common. Ditto the person who protests against war and injustice in the name of religion, and the person so absorbed in a literalistic misreading of mythology that she is moved to reject science.

            But to deny religion an essence, in the sense of a common property, is not to deny that different kinds of religious understanding can be compared, even ranked, according to intellectually respectable principles. (They too will be normative rather than purely descriptive.) Maimonides gave an example of such a ranking when he compared literalistic interpretations of religious concepts to candy used to entice small children to learn. This example is instructive, in that it lets us see how different kinds or levels of religious understanding can be present in the same community, indeed the same individual at different times in her life. Nothing has really changed since the prophet Hosea wrote that God wants “mercy, not animal sacrifice; and the knowledge of God more than burnt offerings.”

Progress in religious education would then be inversely proportional to the degree of literalistic understanding, or superstition, yet remaining in a person or group. So it is easy to argue that religious progress aims (ultimately, normatively) at compatibility with science. It would be both politically prudent and scientifically useful if attempts to “explain” religious phenomena likewise aimed, not at “breaking the spell”, but at a perspicuous arrangement of heterogeneous religious phenomena, and an understanding of the conditions that favor the process of religious education.