Missionloops · research seat (open) · judgment & rational thinking
Smart is not the same as rational.
A research seat for the psychometrician of rational thinking: the person who builds the instruments that measure whether someone reasons well, apart from how intelligent they are.
What this seat is about
Being intelligent and reasoning well are different things, and the difference is measurable. Across the standard reasoning tasks, the correlation with intelligence runs from moderate down to zero, and the most consequential biases sit at zero. Two of them, the slant toward your own side and the pull of how a choice is framed, do not shrink with intelligence or with age. The work of this seat is to measure the part of good reasoning an intelligence test misses, and to find out whether deliberate practice can move it.
What the literature already shows
- Rationality is separable from intelligence, and the gap is measured. The correlation is far from unity, so an intelligence test cannot stand in for a test of rational thinking, and a high-scoring person can still reason badly.
- Myside bias is the bias intelligence does not touch. Evaluating evidence in a way slanted toward your own prior opinion is essentially unrelated to intelligence; in some studies the correlation is zero, and across many policy positions the higher-intelligence group is no less biased.
- But myside bias becomes correctable when something outside calls for the override. Tell a person to set their own belief aside and argue as if they did not hold it, and intelligence starts to help, because now there is a specific move to carry out. Left to themselves, people do not make it, and lower myside bias tracks years of education, with intelligence adding nothing.
This seat's bet follows from that last finding. The guide is the outside call: the second person who asks the operator to argue the case as if they believed the opposite, on a problem that is the operator's own. Whether the guide reliably produces the override is not assumed. It is the first study below.
The open questions
The open question is whether real-stakes practice moves myside bias. The bet is that the driver is an incentive: when the problem is the operator's own and the cost of being wrong lands on them, reality sets the standard rather than the side. Either the slant is effectively fixed, and trained operators show it as much as matched high-ability controls, or real stakes and a guide-supplied override move them toward the accurate answer.
Separating the gain from intelligence. The hard measurement problem is showing that judgment improved, with any intelligence gain measured separately, on instruments that can be given twice. A performance test and a self-rating of the same capacity also routinely diverge. The platform will hold both for the same operator: a record of how the work went, scored blind to the self-report, and the operator's own self-assessment. Calibration between them belongs to the metacognition seat, and the scoring meets the statistics and measurement seat.
Both questions have a defence case: cognitive-warfare resistance and cognitive sovereignty.
Why the platform is the instrument
The lab can elicit the slant toward one's own side on real opinions, and does. What it cannot supply is a consequence: nothing lands on the reasoner who keeps the slanted answer. The platform is designed to supply what the lab cannot: the operator's own identity-loaded problems under genuine stakes, a second observer who can call for the override, and a controlled way to deny or corrupt the tool at test.
The studies this seat can run
Take these as a starting menu: recast any toward your own published line, or bring a question we did not anticipate.
- Real-Stakes Practice and Identity-Protective Cognition Whether real stakes move the slant toward one's own side, the bias intelligence does not touch.
- Trained Judgment Under AI Denial and Compromise Whether judgment holds when the AI is removed or turned against the operator, and whether a second observer catches what the operator cannot.
The seat, and the terms
The seat is open, and the first step is a conversation. The terms are a free option: tell us the question you would bring, and commit only if the work is funded.
We are looking for the researcher whose work is the measurement of rational thinking: the instruments that show whether someone reasons well apart from how intelligent they are, and the biases intelligence does not touch. The seat is scoped to assessment. We are not asking you to adopt a theory of judgment. The platform is being built as the substrate the measurement needs, instrumented from the first operator so judgment can be tested before and after, against a criterion set in advance.
scott (at) missionloops (dot) ca
See also: All research seats · Defence · Member · Investor