The science underneath
Active inference: acting in order to know
In shortWe act in order to find things out: looking, touching, asking and testing are ways of reducing doubt.
The idea
In active inference, acting is a way of perceiving. We move our eyes to where we expect to find information, touch to confirm, ask to reduce doubt. A child who tests limits, repeats a question or only does what they already know is also choosing actions that give information or predictability. The choice between exploring and staying with the familiar depends on how uncertain they are and how much they trust what they already know.
What to try
- When a child repeats questions, ask yourself what doubt they are trying to reduce.
- Offer small ways to explore where the outcome is predictable (watch first, try a little, leave if they want).
- Notice what the action gives them: new information, or confirmation that everything is as expected.
The common mistake
Seeing repetition or seeking control as stubbornness, without asking what uncertainty the action reduces.
Want to know more?
Start here
Short, plain pieces by the researchers and good writers.
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.
Policies are selected by expected free energy, which sums an epistemic term (expected information gain about hidden states and parameters) and a pragmatic term (agreement with prior preferences). The precision of beliefs about policies works as confidence in action and is linked to dopamine in several models. In formulation this lets us separate beliefs about the world, preferences, confidence in one’s own action and the value of exploring. Repetitive behaviour may reflect habitual policies with a strong prior, low epistemic value because exploring is expected not to inform, or narrow preferences; each hypothesis predicts different responses when the predictability of the outcome is manipulated.
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
- Friston, K., FitzGerald, T., Rigoli, F., Schwartenbeck, P., & Pezzulo, G. (2017). Active inference: a process theory. Neural Computation, 29(1), 1–49.
- Parr, T., Pezzulo, G., & Friston, K. J. (2022). Active Inference. MIT Press.
Found this useful? Buy Ricardo a coffee