This is a bit of an aside before getting to the final instalment of this series on quantum theology next week. It is one more exploration of the question: What warrant do I have for any of this? What business is it for a scientist to talk about god?
There have been attempts over the years by both thesists and non-theists to "reconcile science and religion". These were all pre-Bayesian, before the nature of science was properly understood as "the discipline of publicly testing ideas by systematic observation, controlled experiment, and Bayesian inference". When these attempts were pursued by theists they could involved some fairly question begging notions about religion as well. None of them worked.
Before getting to that, I should repair a disservice I've done to scripturalists: I have not defined "religion", and I should probably at least take a crack at.
So: religion is a body of practice and/or belief performed and/or held as a matter of faith and tradition and involving the supernatural, spiritual, or transcendental. There are problems with this definition—the notion of “the supernatural” is incoherent, for example—but it’ll do for the moment.
Buddhists for example sometimes say they don't have a religion because if the facts change they will change their beliefs. They aren't scientists because the disciplines they use to update their beliefs in the face of new facts are a haphazard melange of pre-Bayesian ad hockery, but it would be fair to say that they aren't a religion either. The categories of science and religion are not jointly exhaustive. My question here is: are they mutually exclusive?
This definition of religion as something involving faith is a quite recent one, historically speaking. Ancient religions and hokey weapons didn't involve anything we would recognize as faith. Ancient peoples didn't understand probability at all, so they saw chance associations between rituals and outcomes, and then codified those into modes of propitiating, bribing, or otherwise appeasing or buying off the gods, who were considered to be more-or-less conscious natural forces that don't exist.
American historian Bret Deveraux's "A Collection of Unmitigated Pedantry" blog has a great discussion of how ancient religions worked. All you have to do is forget everything you know about probability and it makes perfect sense.
Creating knowledge is hard, and pre-modern people were not very good at it. We're better not because we're smarter, but because we're standing on the shoulders of giants. The critical juncture came when we understood enough to start publicly testing ideas by systematic observation, controlled experiment, and Bayesian inference (at least informally). The idea that an idea shouldn't be accepted until it has been subject to some kind of empirical test--or at least an empirically-justified theoretical test--is revolutionary. The name of the revolution is "science".
Over a thousand years prior to the scientific revolution, there was a religious revolution: monotheism swept the world by combining the universalizing tendency of the Roman pantheon--whom everyone was expected to do obeisance too--with the just-one-god belief of the Jews. This produced Christianity and Islam in all their variety, which are the models for what most Westerners think of as "religion" today: a body of belief and practice founded on faith and tradition, which is essentially the Jewish model. Different sects have a different balance between "belief" and "practice", but they all have at least some of both in the mix. Religions are things people believe and do because they have been brought up that way or bought into them as adults, and the ultimate answer to the question of "Why?" is "Because" rather the scientist's answer of, "We don't know yet, but here are some ideas as to how we could proceed to figure it out."
This definition of religion is, like all definitions, fuzzy around edges. Definitions are fuzzy because the world is continuous and only the boundary of our attention is infinitely crisp: we can tune it to whatever scale we please. I'm not going to worry to much about those edges. Questions like "Is this or that carefully crafted cult a religion?", or "Is a belief system with only one believer a religion?" bore me. There is nothing more sterile than arguing over whether or not some weird edge case is "really" some specific word. Just create a new word and move on, acknowledging that reality exists, no matter what you call it. Is this bit of the Earth’s surface “really” ocean or land? It’s a beach.
So that's religion on the one hand and science on the other, and here am I talking about both in the same breath. Is this some kind of attempt at reconciliation?
Nope.
I'm a Bayesian, and there is no way of reconciling Bayesianism with faith, because faith is an idea held with certainty, which is a plausibility of 1 ("certainly true") or 0 ("certainly false"). There is no way to reach that state via Bayesian updating, and given an idea held in error this way there no way to move away from it using Bayesian updating either. The easy way to see this is that Bayesian updating is done by multiplying your current level of belief by a ratio that depends on the quality of the evidence. No matter what the ratio, zero times anything is still zero.
Specifically, when a Bayesian is considering an idea, they look at the evidence and consider the ratio of how likely the evidence would be if the idea was true to how likely the evidence would be regardless. Things that would likely happen regardless are lousy evidence. Things that are unlikely to happen unless an idea is true are good evidence. None of this has to be strictly mathematical: Bayesianism is robust against informal application.
We reason like this all the time. If I wake up in the morning and there are a lot of tree branches on the ground the odds are there was a big wind in the night, because tree branches rarely fall down without a big wind. They do sometimes: I've heard a branch come crashing down on my neighbour's property on a still clear day. But mostly it's wind. It could be lots of other things, too, but all of those things are less likely than wind to begin with, so all else being equal they are less likely than wind when I look at the evidence of fallen branches. Ockham's razor is nothing but a statement about the distribution of prior probabilities.
On the other hand, if I wake up and see lots of fir-cones one the ground, it doesn't tell me much, because fir cones drop spontaneously. Since they are commonplace without wind, they don't provide much evidence of wind.
On the Bayesian view, knowledge is knowledge of plausibility: all we know is how plausible something is given the evidence, not whether it is "true" or “false”. In modern woo-meister terms, Bayesianism is non-dualistic.
Some things are so plausible--and the alternatives so implausible--we can treat them as "true", but those are special cases, and we shouldn't use special cases as our model of knowledge. Just think about how misleading it would be to let an incredibly weird special case like physics dominate one's view of what a "science" was. Platonic ideals are a bad way of looking at things: the interesting instances, the canonical instances, of a concept are things that barely make it into the category the concept identifies.
So "truth" is a bad name for a boring special case, not representative of knowledge in general. Likewise “falsity”.
For two thousand years philosophers sought "truth" or "certainty" because they thought mathematical propositions had it (they don't). Only when scientists started seeking plausibility rather then "truth" did we start accumulating knowledge in a big way.
And yes, Bayesianism itself is only very highly plausible, not "true". Show me some evidence that lowers its plausibility and I'll happily entertain it.
Attempts to reconcile science and religion are therefore doomed. Science is Bayesian, religion is not. If religion becomes Bayesian, it becomes science, because Bayesianism admits of no isolated bodies of fact: anything can be used to test any idea. Biologists use physics in tests of evolution. Computer scientists use biology to implement parallel algorithms, and so on. The world is continuous and deeply connected. So is our knowledge of it, if we make the effort (which isn't always worthwhile).
Theists of one kind or another have sometimes tried to escape this conundrum by invoking the notion of what Stephen Jay Gould called "the two magisteria", which is the idea that science deals with sciencey things and religion deals with religiony things.
Gould was almost as clever as he was nasty, but he wasn't clever enough to make this idea plausible. As biologist and anti-theist Richard Dawkins has said, as soon as religion makes any claim about the world, about things that exist, it is putting itself up against science. There can't be two magisteria because there is only one world.
Psychologically, though, people are often highly compartmentalized.
There are plenty of areas in our lives that don't overlap with each other much, and so we can comfortably think about them in their own more-or-less Bayesian compartments. I'm a sailor and an actor, for example, buy my knowledge of sailing and my knowledge of acting don't impact each other in any way I can think of. I may well believe something about sailing that implicitly contradicts something I believe about acting, and will likely never know it. Which is fine: there's no particular virtue in establishing formal consistency of belief between widely separated areas of knowledge, and despite looming fairly large in my life, sailing and acting are both niches. They aren't broad, sweeping categories but rather thin slivers, and therefore unlikely to collide.
But the domain of science is everything, which is why my personal blog has a banner that reads "No Question Left Unasked".
Religion, no matter how restricted its domain may be, necessarily falls into the domain of science if it makes any claim about the world at all. And for many religious people in the modern world, the domain of religion is also "everything", so overlap is inevitable, and conflict is extremely likely simply because science as I have defined it is the only way of knowing, and insofar as religion differs from science it will result in beliefs whose plausibilities are inconsistent with even the most informal of Bayesian updating from the evidence.
And I cannot be having with that.
There is only one world and only one way of knowing it, albeit a way so general that it encompasses all of what get called "ways of knowing" these days. I don't care if it's the traditional knowledge of my people, or the latest result out of someone's lab: if it is knowledge it has been arrived at by some kind of more-or-less Bayesian updating. Feyrabend’s “anything goes” holds true insofar as science is considered a method. If we instead understand it as a discipline, everything that goes is Bayesian updating of some kind, based on some priors, in response to some evidence.
The fact that Bayes' Theorem was only discovered two hundred years ago doesn't change this, any more than saying "If a boat floats it uses Archimedes' Principle" implies that prior to 200-something BCE boats didn't float. Archimedes and Bayes both just isolated one aspect of reality from the rest in the typical reductionist-humanist way that respects our nature as beings of limited attentional resources, and named it, or got it named after themselves. The facts named by the principle were and are being used by everyone regardless.
So I do not acknowledge any exclusive domain of religious knowledge.
The idea of "non-Bayesian knowledge" is as incoherent as the idea of "non-Archimedean buoyancy".
My knowledge of reality tells me the universe is a particular way, and that way lets me say something about the nature of god as philosophers and theologians have defined "god". It follows from this that I will say soemthing about the nature of god. I don't have a lot of choice in the matter: I'm either a scientist, or I'm not. I've encountered something in the world that conforms to the theoretical concept of "god". That the theoretical concept happens to come from "another domain" is irrelevant, because the domain of science is everything. There is no warrant for excluding scientific, Bayesian inquiry from anything, including religion.
Which leaves me where I am: a Bayesian talking about god.