Back to observatory
Information Theory & Media Ecology

Predictive Processing & Constructed Experience

Level: intermediateModel #54
Description

We don't experience the world as it is, but as we predict it to be. The brain constructs conscious experience from scratch using layered predictions built from past experiences, then updates these predictions when sensory input contradicts expectations. What feels like direct perception is actually sophisticated prediction that happens too quickly to notice.

Applications
Recognize that your immediate experience is prediction, not perception. When you feel certain about what someone "obviously meant" or what's "clearly happening," pause and notice you're experiencing your brain's best guess, not unfiltered reality. Meditation traditions developed techniques to deconstruct automatic predictive processing, allowing temporary experience of consciousness without predictive filtering. Practice catching yourself in predictive mode by noticing when expectations don't match incoming data.
When learning new fields, expect your early experiences to be heavily filtered by inaccurate predictions. Experts literally see different things than novices when examining the same data—not metaphorically, but through different predictive processing. Speed up learning by explicitly updating your predictive models: after each exposure to new domain, reflect on what surprised you and what confirmed predictions, then adjust internal models accordingly.
In communication and conflict, remember that different people genuinely experience different realities from the same inputs. They're not being difficult or irrational—their brains generate different predictions based on different prior experiences. Bridge this gap by making predictions explicit: "I expected X because Y" helps others understand why your experience differs from theirs.
Design systems and environments to provide rapid feedback that challenges inaccurate predictions. Predictive models improve through prediction errors—the gap between expected and actual. Environments that hide consequences or delay feedback allow inaccurate models to persist unchallenged. Real-time dashboards, immediate testing, and fast iteration cycles accelerate prediction updating.
Referenced in the brief

Backlinks to brief references will populate as this model is used.

Source material
Loading sources…