32. Predictive-Coding Display
- Status: Speculative
- Interface point: Perception and cognition
- Carrier: Any usable sensory, prosthetic, or symbolic channel
- Transducer: Display, speaker, haptic array, or future neural interface
- Encoding style: Cues that guide expectation and error correction
- Addressing method: Attention, salience, and mismatch
- Expected percept: A user-guided reconstruction rather than a rendered image
- Related modules: Perceptual Rendering, Memory Branch
Summary
A predictive-coding display uses the brain’s tendency to predict and fill in
missing information. Instead of sending everything, it sends strategically
chosen cues that help the user infer the rest.
Chain Map
scene
-> predicted structure
-> mismatch or important feature
-> cue
-> user reconstruction
Example
For navigation, the system may not need to display every chair leg and table
edge. It might cue:
open path center
chair left
person moving right
step down ahead
The user’s own perception and memory fill in a usable model of the space.
Expansion Ideas
- Prediction-first scene compression.
- Cue only what changed since the last moment.
- Encode surprise: new object, motion, obstacle, or mismatch.
- Let the user ask for detail when prediction is uncertain.
- Combine with memory of familiar rooms.
Research Questions
- Which cues best support reconstruction?
- Can too much prediction cause false confidence?
- How should the system show surprise or uncertainty?
- Can a user learn to interact with a predictive cue stream fluidly?