You can’t tickle yourself for the same reason you can’t see your own typos.
Read this quickly: PARIS IN THE THE SPRINGTIME. Most people see “Paris in the springtime” and never notice the second “the”. The letters reached your eyes. Something between your eyes and your awareness decided they were not worth showing you.
The common picture of perception is a camera: the world sends in data, and the brain assembles it into an image. Predictive processing, the framework Andy Clark lays out in Surfing Uncertainty, runs the arrow the other way. The brain is a combination of top-down prediction reconciling with bottom-up sensory information. It surfs uncertainty in a controlled hallucination to generate reality. If that is right, the missing “the” is not a glitch. It is how seeing works.
How the prediction runs
The brain maintains a hierarchical model of the world, and each level predicts what the level below it will report. When the prediction matches the input, nothing needs to happen: the prediction is the perception. When there is a mismatch, a prediction error travels upward and the model updates. So the bottom-up stream is not raw data. It is a correction signal that carries only the difference between what the model expected and what arrived.
Small errors never get that far. Minor perturbations are smoothed for coherence in the middle layers, and only an error large enough to trip the surprise alarm forces a re-prediction. Your model expected one “the”, the extra one was a small error, and it was smoothed away before it reached you.
How large an error has to be depends on how much the brain trusts it, its precision. In this framework, dopamine mostly carries something like this precision signal: less “this is good” than “pay attention, this error matters”. Turn the precision on errors up and the world never settles into a stable picture; turn it down and the model’s predictions override the senses. The framework uses exactly this dial to explain autism, where weak priors leave the world raw and a shirt tag keeps feeling new, and schizophrenia, where top-down models grow strong enough that sensory evidence cannot break through.
The same machinery predicts your own actions. You cannot tickle yourself because your model anticipates your own touch and attenuates it. Knowing what you meant to write does the same to your own typos, which is why a stranger spots them first.
When the model wins
The trouble begins when the prediction becomes more trusted than the signal. Then the brain stops seeing the extra “the” in places where it matters.
The brain would rather be right and miserable than wrong and happy. A nervous system trained on chaos expects chaos. A calm environment then generates massive prediction error, and the cheapest move is often to reject the data rather than update the model: sabotage the peace, manufacture a crisis, or choose surroundings that confirm the existing prediction. Trauma works this way. The model predicts threat in every ambiguous situation, and the precision on that prediction is so high that the bottom-up signal (this room is safe, this person is kind) cannot override it. The alarm fires constantly, not because the world is dangerous but because the model says it is. Trapped priors is the general name for this failure: a prior strong enough that every new encounter is processed as further confirmation.
The same mechanism works at the level of whole communities. In paradigm lock-in, the scientific paradigm is the top-down model, and contradictory evidence is smoothed away as noise until it piles up past the surprise threshold. Semmelweis’s handwashing data was prediction error that the prevailing model suppressed.
So predictive processing explains why smart people get stuck (strong priors resist update), why trauma persists despite safety (the model generates the danger), why expertise can become a trap (the prediction model compresses away the details that matter), and why beginners sometimes see what experts miss (weak priors let more bottom-up signal through).
Not a defect
It would be easy to conclude that the hallucination is the problem and the cure is to see raw reality. But the controlled hallucination is how an organism survives in a world with too much information. Without strong priors, every shirt tag, every repeated word and every ordinary room would demand fresh attention. The failure is not prediction itself. It is a hallucination that has become more real than reality.
That leaves a practical question rather than a metaphysical one. Contemplative practice can be read as training in catching the extra “the”: concentrating on experience closely enough, and with few enough preconceptions, to notice the gaps that priors are smoothing over. The brain does not show you the world. It shows you a dream of the world that is just accurate enough to keep you alive. The question is always: how much prediction error are you willing to tolerate before you update?
References:
- Andy Clark, Surfing Uncertainty: Prediction, Action, and the Embodied Mind
- Scott Alexander, Book Review: Surfing Uncertainty, Slate Star Codex