he institute's three subjects are the way people behave, the way machines behave, and what happens between them. Each page below takes one research question at a time: the question, the classic finding with its source, what held up and what did not, what it looks like in machines, and what is still open. Citations are to the original papers; every DOI was checked before it went up. Where we summarize an author's argument it is marked as paraphrase. A fourth shelf, the nervous system, holds the biology the other three keep borrowing from.

human behavior

	Signal detection — how a yes/no answer splits into seeing and deciding
	Prediction error — learning from the difference between what you expected and what happened
	Replication — which findings come back when you run the study again
	Bounded rationality — deciding well with limited time, memory and information

AI behavior

	Sycophancy — models that tell you what you seem to want to hear
	Calibration — when confidence matches how often you are right
	Shortcut learning — right answers for the wrong reasons
	Interpretability — what a trained network is actually using, and how anyone could tell
	Hallucination and confabulation — fluent, confident, and not true
	Specification gaming — the letter of the objective, not its point

human–AI interaction

	Automation bias — trusting the machine too much, and too little
	Anthropomorphism — why we see minds in machines, and what follows
	Complementarity — when a person plus a model beats either one
	Algorithm aversion — why people pick the worse forecaster when it is human
	Mental models of machines — the theory in your head about the system in front of you
	How people respond to machine-made art — the same picture, liked less once it is called a machine's; what the label is doing

the nervous system

	The nervous system — the front page of this shelf
	The engram — where a memory is, if it is anywhere
	Predictive processing — the brain as a guessing machine
	Connectomics — the wiring diagram, and what it does not tell you
	Brains and models compared — when a network predicts neurons, and what that shows

how these pages are written

Plainly, we hope, but without skipping the hard part. Terms link to the glossary the first time they appear. Named people link to the people pages. Each page ends with a reading list built from the works it actually cites, and the annotated bibliographies say a little more about each.