Reality is Bayesian: the evolution of reality

How real is real

Our reality is constructed; quite simply, our brains build it.

Let’s start with some examples from sense perception.

  • We all live just a tiny bit in the past. Our brains wait for sight, sound, and touch to sync up before they deliver what we perceive. Typically, this takes .5 seconds.
  • The neural signal time difference from toes to brain is the slowest piece, and the taller you are, the greater the time lag. As a result, tall people live further in the past than short people!
  • Early television designers thought syncing sound and image would be difficult until they observed that our brains have an automatic 80-millisecond delay correction.
  • You can observe this by walking backward away from someone bouncing a basketball and noting when it goes out of sync.
  • If you’re watching a video that seems out of sync, then it’s really out of sync. (And, of course, that video consists of between 24 and 60 still frames per second that your brain is assembling into continuous motion.)
  • Here’s an experiment that demonstrates the impact of evolution on how our brain processes visual signals. Humans are “cusorial hunters.” We evolved to chase things down–like wolves do. You can see the result of that in our nervous system. Pick a spot a bit off in the distance. Now, move your head up and down as if running forward while keeping it in focus. Good? Then try that while moving your head from side to side. Quite different, right?
At ~16:00 Tall people live in the past; at ~22:00 syncing up sound and sight –

Furthermore, a lot of our moment-to-moment reality is prebuilt. We simply don’t have the brain capacity to look at everything anew each moment. We start with a model of sorts and correct it.

Here’s a great example from Merlin Sheldrake in an Entangled Life: How Fungi Make Our Worlds, Change Our Minds & Shape Our Futures.

A friend of mine, the philosopher and magician David Abram, used to be the house magician at Alice’s Restaurant in Massachusetts (made famous by the Arlo Guthrie song). Every night he passed around the tables; coins walked through his fingers, reappeared exactly where they shouldn’t, disappeared again, divided in two, vanished into nothing.

One evening, two customers returned to the restaurant shortly after leaving and pulled David aside, looking troubled. When they left the restaurant, they said, the sky had appeared shockingly blue and the clouds large and vivid. Had he put something in their drinks? As the weeks went by, it continued to happen—customers returned to say the traffic had seemed louder than it was before, the streetlights brighter, the patterns on the sidewalk more fascinating, the rain more refreshing. The magic tricks were changing the way people experienced the world.

David explained to me why he thought this happened. Our perceptions work in large part by expectation. It takes less cognitive effort to make sense of the world using preconceived images updated with a small amount of new sensory information than to constantly form entirely new perceptions from scratch. It is our preconceptions that create the blind spots in which magicians do their work.

By attrition, coin tricks loosen the grip of our expectations about the way hands and coins work. Eventually, they loosen the grip of our expectations on our perceptions more generally. On leaving the restaurant, the sky looked different because the diners saw the sky as it was there and then, rather than as they expected it to be. Tricked out of our expectations, we fall back on our senses. What’s astonishing is the gulf between what we expect to find and what we find when we actually look.

Bayesian reality

I’d like to dig a bit deeper into what Merlin terms “what we find when we actually look.” Even that is grounded in a prebuilt model.

My core metaphor here is a statistical method developed by Thomas Bayes in the 1760s. It works by starting with a model, noting differences, and using that to correct the model exactly as Merlin’s friend David describes above.

Here’s the math…

{\displaystyle P(A\mid B)={\frac {P(B\mid A)P(A)}{P(B)}}}

…which resolves to “take a starting model, make an observation, correct the model.”

I’m not saying our brains do the math. Instead, I claim that this is an excellent metaphor for what we actually are doing.

In fact, our base reality is itself a model created in the face of Darwinian selection that weeds out unworkable realities over time just like any other maladaptation.

First, evolution

Our perceptual system starts with a model generated by our species’ evolutionary history. Our package of senses and interpretive schema evolved around the environmental challenges of our species.

Banana slugs…

….have a perfectly workable reality. Four tentacles feel and smell. Two tiny black dots on the upper tentacles detect light and motion.


…are on the opposite end of the visual acuity spectrum and can see a rabbit from above two miles away. The acuity is built into the structure of their skulls. There eye sockets are angled 30 degrees from the midline of their face, which gives them a 340-degree visual field and, thus, excellent peripheral and binocular vision.

Our visual reality would be featureless grassland.

Photo – Create Commons by M.Kuhn at


…are an example of the evolutionary interactions that create different realities. Tigers are orange and black because gazelles see orange and black as the same colors as green and brown foliage.

How different a tiger appears to dichromats and trichromats. Fennell et al. 2019 

Each species lives in a “reality tunnel”

Image from Wikimedia, Bowermanlucas, Creative Commons 4.0 share with attribution

From this perspective, each species’s evolution is a trip down a particular reality tunnel where the possibilities of perception and cognition are built upon their immediate predecessors and have only a limited amount of flexibility. Our reality is a hypothesis shaped by evolution. It will be eliminated if it becomes maladaptive. Our inability to feel the impact of slow destructive change might, for example, be our undoing. It was futurists call the “long fuse, big boom” problem.

Culture and Learning

Every kind of ignorance in the world results from not realizing that our perceptions are gambles.
– Robert Anton Wilson


Stacked on our evolutionary rootstock is homo sapiens’ capacity for learning and our extensive cultures.

Piaget discovered that babies learn object permanence around 8 months old. Peekaboo is a big learning tool. Earlier in life, things that were moved out of sight were of questionable existence. Piaget considered this learning a major milestone in an infant’s development.


This all happens in a cultural context. Some of our perceptions are culturally specific.

Daniel Kahneman introduced the distinction between fast and slow thinking. Fast thinking is described as instinctive and emotional, but a close read shows that a lot of his fast thinking is simply learning that’s been taken deep enough that it’s automatic. In fact, that automaticity is the goal of a lot of training. Slow learning becomes “fast thinking.”

As a result, much of this fast thinking is learned and culturally specific.

For example, we don’t have to puzzle out the words on a stop or yield sign as we approach one. We see a stop sign well before the word is clear. But what does this sucker to the left mean?

Robert Anton Wilson, “American author, futurist, psychologist, and self-described agnostic mystic,” uses the term reality tunnel to describe the unconscious set of mental filters that frame up our worldview. Wilson was a confederate of Timothy Leary and might better be described as a psychedelic philosopher. Anyone taking what is now called a “heroic dose” of LSD or mushrooms (what in the olden days we called a “dose”:-) can pretty quickly see his point.

This can have some alarming implications


I’ll always love the false impression I had of you.
– favorite saying, Wendy Walsh

With our history of extreme sociality and cooperation, we necessarily build models of other people, complete with a story about why they’re doing what they do.

Even our models of people who are our close relationships are constantly refined. Starting with mirror neurons and building out to Freud’s projection and transference, the modeling of the folks around us is a case in point of how such modeling is essential and can go astray.


There is increasing evidence that this modeling is not only externally directed. Emotions—particularly the more subtle ones—seem to be a mix of signals from the body and a learned and occasionally culture-bound interpretive schema.

And you

All this is a bit disquieting.

My view: reality is a leash tethered often to unmoveable facts; in some cases, it’s a short leash and, in others, a long leash. Hitting the end of your leash can be amazing or unpleasant, enlightening or deadly. My brother Tim felt the whole world was less solid for months after the Volkswagen he was riding in fell through the ice.

Something appearing as a bolt from the blue can set us back for moments as we marvel at how amazing the clouds are–or for months as we recalibrate our reality if trust goes astray.

Thanks for reading.

PS, over on Medium, I’ve begun a series on Myth and Gender hosted by my friend Patsy in her feminist magazine, Fourth Wave. These should be the free links.

The Night Sea Journey in Campbell and Jung
Myth and Gender in Jungian Psychology

One thought on “Reality is Bayesian: the evolution of reality”

  1. Whoa. I’m still trying to sort this out. And my question is how much is our view of reality (and I say view because a lot of the examples are visual) create by other people — do we see what others see? Do we want to? Does it make us feel safe? I know you’ve written a lot about tribes and trust — but I want to know how much we’re influenced by those around us — and how much we influence them. P.S. Any article that quotes the philosophy of Wendy Walsh is worth a second read!

Leave a Reply

Your email address will not be published. Required fields are marked *

Creative Commons License
Except where otherwise noted, the content on this site is licensed under a Creative Commons Attribution 4.0 International License.
Useable (share, remix, etc.) but with attribution