The science underneath
The Bayesian brain
In shortThe brain mixes what it already expects with what the senses say now, trusting whichever seems more reliable.
The idea
Thomas Bayes described how to combine what you already know with new information. The brain seems to do something similar: it combines expectation (what usually happens) with what the senses say now, giving more weight to what is more reliable. In a dark room we trust what we expect; in good light we trust what we see. Differences between people may lie in how they make this combination in certain situations, not in a general ‘fault’.
What to try
- When a child holds on to an interpretation despite what you tell them, ask what makes that expectation so reliable for them.
- Give new information reliably: the same person, the same words, several times.
- Information given once, in a tense moment, counts for little.
The common mistake
Saying a person ‘has strong priors’ or ‘weak priors’ in general. The combination changes with the problem, the level and the context.
Want to know more?
Start here
Short, plain pieces by the researchers and good writers.
Books
How the predictive brain shapes what we see, feel and do, with many everyday examples.
See on Amazon ›Amazon links are affiliate links: if you buy, Bounded Uncertainty receives a small commission at no cost to you.
The science behind itHow strong the evidence is, the model, videos and references.
How strong the evidence is
Promising Good studies, but little tested with children at home or school, or the step to daily life is our interpretation.
Optimal integration weights prediction and evidence by their precision (inverse variance). There is good behavioural evidence for near-optimal integration in perception and motor control (Wolpert; Knill and Pouget). Accounts of autism (Pellicano and Burr; Lawson, Rees and Friston; Van de Cruys et al.) propose differences in the precision of priors or errors, but the data are mixed: Sterner et al. (2026) found no greater weighting of sensory evidence with more autistic traits in a semantic task, and differences seem domain-specific. In assessment, manipulating prior and evidence independently and observing what happens when they conflict is more informative than assigning a global parameter.
More technical books
The technical textbook on active inference, for those who want the formal model.
See on Amazon ›Amazon links are affiliate links: if you buy, Bounded Uncertainty receives a small commission at no cost to you.
Videos
External videos, chosen because they agree with these ideas. Bounded Uncertainty has no link to the channels.
References
- Knill, D. C., & Pouget, A. (2004). The Bayesian brain: the role of uncertainty in neural coding and computation. Trends in Neurosciences, 27(12), 712–719.
- Pellicano, E., & Burr, D. (2012). When the world becomes ‘too real’: a Bayesian explanation of autistic perception. Trends in Cognitive Sciences, 16(10), 504–510.
- Sterner et al. (2026). Altered semantic prediction error processing with increasing schizotypal and autistic traits. Translational Psychiatry.
Found this useful? Buy Ricardo a coffee