The science underneath
Predictive coding
In shortThe brain passes on mostly what it did not expect, which is why surprises stand out.
The idea
A proposal about how the brain makes its predictions: higher levels send predictions down, and lower levels send up only what was not predicted (the prediction error). Instead of processing everything that comes in, the brain works mostly with differences. That is why what is predictable passes almost at no cost, and what is unexpected catches attention and forces adjustment.
What to try
- For a child who reacts strongly to small changes, say what changes and what stays the same.
- When something new goes well, repeat it enough for it to become expected.
- Notice what the child picks up that others miss: it may be a prediction error that matters to them.
The common mistake
Thinking prediction error is always bad and the goal is to remove it. It is what makes learning possible; the problem is when it is too large, too frequent, or does not lead to updating.
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.
In Rao and Ballard’s model, error units and representation units pass messages in a hierarchy; extra-classical receptive field effects arise from suppressing what is predicted. Predictive coding is an implementation scheme compatible with free energy minimisation, but there are alternative models and the neural evidence (for example mismatch negativity, repetition suppression) allows other interpretations. For assessment the useful question is empirical: what happens when predictability and signal quality are manipulated independently, and what changes afterwards (Xue et al., 2026, show that large errors after strong predictions recruit control and attention and are associated with better later learning).
Videos
External videos, chosen because they agree with these ideas. Bounded Uncertainty has no link to the channels.
References
- Rao, R. P. N., & Ballard, D. H. (1999). Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature Neuroscience, 2(1), 79–87.
- Xue, Jokhani, Norcia & Wagner (2026). Strong mnemonic prediction errors increase cognitive control, attention, and arousal. eLife (reviewed preprint).
Found this useful? Buy Ricardo a coffee