Atheism is Dead and Karl Friston has killed it

The computational physiology of belief refactors religious disputes

Arguments about the existence of God are not merely a waste of time. These arguments are rendered meaningless because the predictive processing framework of whole brain functionality renders them mute.

Differences in worldviews are entirely explicable in terms of different beliefs about the causal structure of things that make us feel good. This means that atheism and religion are not opposites, so much as orthogonal dimensions in the space of possible beliefs.

The real thing being debated is not “does God exist”, but rather, “why do good things happen?” More precisely, “what is the causal root of positive valence?” We cannot help but have beliefs about “what is the cause of Good” because “inferring the causal structure of valence” is how our cortices are wired.

All of us are meat-robots that try to go away from bad things, towards good things.

The thing we disagreed on for most of history was, “What does good mean?”

We took a brief state-sponsored digression for most of the 20th century because of the idea that Science was a way of out ignorance and superstition. We came to Hume’s fork in the road, and said, “we ought not to spend too much time on beliefs about what ought to be, it is better to develop consensus about what is,” and so we did.

The results were impressive: penicillin, microscopes, modern medicine, the green revolution in agriculture, and a global decline in childhood mortality! Hooray! We also got partially hydrogenated soybean oil, suicide epidemics, and shrinking generations portending future demographic chaos. Boo!

Now that our ability to measure, to test, to sense, has increased dramatically in precision, we find ourselves looking at our own source code: the causal structure of our own belief systems.

We have conceptual belief networks, and these networks compute a function that an economist or some philosophers would call ‘a utility function’, a theologian with a background in topology would call ‘the Gradient which points towards God’, and some ancient philosophers would call “The Good” or “The Mandate of Heaven” Or ‘The Dao” or “The Logos” or “The great Spirit” or “El Shaddai.”

Science has taken us far along the ‘is’ path of Humes’ fork that we realize that the ‘is’ fork dead ends back into the ought fork. Except it doesn’t tell us which way to go! No, you can’t drive an ‘is’ from an ‘ought’, any more than you can derive an egg and some grain from plate of sunny side up over rye. All reasoning, observation, and direction of attention is motivated by some value system. The question is - and always has been - well, what does ‘good’ mean?

Fukuyama was partially right: it wasn’t boredom that brought us back to the path of Thymic striving. It was an acceleration in our capacity to manifest our desires by means of technology. Some worry that the singularity is summoning a demon. Others imagine we are at the end times. Trantrikas argue that any desire is a reflection of our desire for God. What if all that has happened is, after getting really good at understanding how to predict and structure of the tiniest, least meaningful bits of reality, we have arrived at the same place our ancestors always knew: a place of tremendous conflict over values?

The only question - and boy, it’s a deusy - is, well, what exactly does “Good” mean?

Remember Hume’s fork: you can’t get an ‘ought’ from an ‘is’. Science can only tell us about what is. Science now tells us, quite clearly, that there IS, in your brain, a representation of what ought to be. There’s no way around it. You either work with that thing consciously, trying to improve your definition of good, or you don’t.

Any attempt to define “good” constitutes a religion. Any conceptualization of good is what I would call a God¹.

“In the day-to-day trenches of adult life, there is actually no such thing as atheism. There is no such thing as not worshiping. Everybody worships. The only choice we get is what to worship.”

―David Foster Wallace

We cannot choose not to worship, any more than we can chose to have a brain with no conceptual roots for valence.

Because we have no choice but to navigate reality using a predictive model of valence, we all have some religion. We all worship some concept of “the cause[s] of Good.”

The only differences between our religions are as follows:

• How many gods you have (i.e. are you using a single conceptual root with positive valence to predict all good things, or does your concept network contain multiple conceptual roots with valence, which have not yet been stitched together by a single conceptual node?). I’ll bet good bitcoins that the formation of the prefrontal cortex relates to the construction of a single conceptual node that caps the others, i.e. predicts them. Monotheism would then be like the cultural equivalent of a prefrontal cortex, sort of how a stove is like an external stomach.

• Whether that god is the root of your predictive network (i.e. is your god also omnipotent? Or, perhaps subject to some other entity, a blind idiot demiurge called “Physics”?)

• What your relationship with God is (i.e. do you consciously think about what ‘good’ means, and possibly tinker with your internal representation of it? Or do you just do what it tells you you must, at all times, without question, a ball rolling down a hill?)

If you have a brain, it means you use a predictive processing network of concepts to cause your body to navigate a valence manifold in the direction of increasing expected valence. You move towards things you love, away from things you fear. When the predicted valence manifold is stable, you move stably. If you do not have a conceptual model of good which has converged with your experienced reality, you’ll be full of hesitance, doubt, and anxiety over arbitrary timescales.

The differences between people of various religions can be characterized in terms of the shape of that predictive processing network. Someone who believes² that God is the union of all good things and the source of causality has a root node on their causal network with maximum valence. ³ Someone who believes “good” is an evolved concept knitting together different drives has multiple nodes in their causal network which generate positive valence predictions, underneath predictive nodes such as ‘evolution’ and ‘physics’, all with zero (or possibly negative!) valences. This kind of person is, effectively, a polytheist.

So, how may gods do you have? How well do they serve you? How computationally complex is that valence gradient you are attempting to navigate? Has your conceptual model of ‘good’ converged and stabilized? Are you aware of its own existence? Do you look directly at it, or simply go about life under the control of some part of you that you are only dimly aware of?

If there were a conceptual strategy for defining “good” which gave you a smooth continuous valence gradient that always sloped positive, with a belief that causality arcs naturally towards the Good, so long as you merely make the effort to turn in the direction of that conceptual root - would you accept it?

Would you reject most conceptualizations of “good” as being unscientific, and instead use your own, hand-rolled code for that important portion of your computational physiology?

Or would you take what the state priests are selling?

1. One might object that, “thinking something is good isn’t’ the same thing as personifying it.”

This is a good objection! A GREAT objection! So i’ll punt and handle it later!

2. This would only be true if a person actually believed these things, rather than believing that they ought to believe them. I think a lot of what we call ‘spiritual growth’ can be understood in terms of making your predictive network correspond to the ones articulated in various holy books. Lots of people say they believe in an omnipotent, loving God. But how may people act as if this is true?

3. These people are probably extremely rare through history, because WOW that would be an effective way to go through the world.