Predictions and papers
Where the mathematics meets experiment
A theory of observer limits should say something testable about real observers. The program makes three predictions that come from the threshold theorem. Each is stated in advance, so that data can confirm or refute it.
PREDICTION 1A threshold in perception
If awareness depends on integrating patterns up to some depth, then for stimuli built from Walsh patterns, the point where people stop detecting structure should sit at a predictable fraction of the stimulus size, and the transition should get narrower as stimuli grow, roughly like .
Status: a preregistered protocol is written. No results yet.
PREDICTION 2Neural population codes
In recordings from populations of neurons responding to Walsh-structured stimuli, the share of information carried by high-order patterns should show a compressed transition near depth one half, with the same width scaling.
Status: a manuscript describes the analysis. No results yet.
PREDICTION 3Learning in neural networks
When an artificial network suddenly learns a rule after a long plateau, the abruptness of that jump should scale with network width the way the threshold theorem predicts for order-limited access.
Status: a manuscript describes the prediction. No results yet.
Papers and archive
The program's papers, proofs and data are being prepared for this page. Each will be listed with its status, and every computational result will link to the script that produced it.