Claims
Networks whose connections 'give way under stress' learn better-than-random solutions to constraint problems without selection or design
Documented What these labels mean →Plastic spring/network systems with occasional resets learn to find better-than-random solutions to constraint problems, with no natural selection or design ('natural induction').
Evidence
| Relation | Record | Note |
|---|---|---|
| documents | pmc.ncbi.nlm.nih.gov | Levin, Millidge, Tschantz, Watson, Natural Induction (Entropy 26:765, 2024). |
Sources
| Source | Relation | At |
|---|---|---|
| Cognition before selection (Watson, Giant's Shoulder) | asserts |
Ties
| To | Role | Years | Verdict | Source |
|---|---|---|---|---|
| Michael Levin | co-author; the GRN 'learning without weight change' work Watson credits | 2024 | Documented · occurred | pmc.ncbi.nlm.nih.gov |
Claim
Plastic spring/network systems with occasional resets learn to find better-than-random solutions to constraint problems, with no natural selection or design ('natural induction').