Definition
Plain language
An AI system that designs its own psychology experiments, recruits real people to take them, diagnoses why its theories were wrong, and rewrites them — with no human researcher in the loop.
As stated in the literature
The 'automated cognitive scientist' framework that closes the full theory-building loop: two advocate agents design discriminating experiments, real human data is collected online, theories are scored by whether they generate human-like behavior, and an arbiter diagnoses and revises the loser across cycles.
Why it matters: It points toward science that can build and test its own theories of human behavior at machine speed, removing the human researcher as the bottleneck in the discovery loop.
For example, it might design a survey about how people choose between products, recruit and pay online participants to take it, notice its theory predicted the wrong choices, and rewrite that theory before running the next round.
Heard on the show
“And this system — it's called AutoCog, the automated cognitive scientist — closes the full loop for the first time.”Episode 176 — An AI Designed Its Own Psychology Studies, Then Confirmed What It Found