The CoRT Program

In the AI Story

The PMI tool is the program's foundational exercise. Before evaluating any idea, the thinker lists what is Plus about it, what is Minus, and — critically — what is simply Interesting. The Interesting category is the one conventional education omits. The conventional response to an idea is binary: right or wrong, good or bad. PMI adds a third option, and the Interesting category is where lateral movement lives — the observation that an idea is neither right nor wrong but leads somewhere unexpected, opens a direction not previously considered.

De Bono reported deploying PMI with children asked whether they should be paid to attend school. Before PMI training, responses were binary — mostly yes. After PMI training, the same children generated: 'If we were paid, older kids might bully younger ones for their money' (Minus). 'Schools might start competing for students the way businesses compete' (Interesting). 'Students might feel like employees and demand better conditions' (Interesting). Each Interesting response opened lines of thinking the binary framework excluded.

The program was deployed at national scale in several countries. Venezuela adopted CoRT as mandatory curriculum in the 1970s under the Ministry for the Development of Intelligence. Malaysia integrated it into secondary education. Singapore used it in civil service training. Teachers and administrators reported measurable improvements in student thinking, though formal controlled studies were sparse — a gap de Bono treated as less important than practitioner observation, a choice that made him unpopular with academic creativity researchers.

In the AI age, the CoRT framework has an unanticipated relevance. The educational system spent decades suppressing lateral thinking in favor of vertical correctness — precisely the capability that machines now perform better than any human. The curriculum optimized for the wrong thing. CoRT's tools — which explicitly train the generation of alternatives, consideration of irrelevant factors, perspective-shifting — are the skills the AI age demands of builders who want to produce output the machine cannot generate from its defaults.

Origin

De Bono founded the Cognitive Research Trust at Cambridge in 1969 and developed the curriculum through the early 1970s. The program has been deployed in schools and corporations in over forty countries, with particular scale in Venezuela (as part of the 1979 Ministry for the Development of Intelligence initiative), Malaysia, and Singapore. The empirical validation literature remains contested — practitioner reports are extensive, but peer-reviewed controlled studies are limited.

Key Ideas

Thinking as teachable skill. The program's foundational claim — thinking is a skill subject to deliberate practice, not a mysterious endowment.

PMI as foundational tool. Plus / Minus / Interesting — the Interesting category is where lateral movement lives.

CAF, APC, OPV as complements. Consider All Factors (scope), Alternatives/Possibilities/Choices (option generation), Other People's Views (perspective-shifting).

National-scale deployment. Venezuela, Malaysia, Singapore — programs taught at scale, with reported improvement observed by teachers if not always measured by researchers.

AI-age relevance. The curriculum explicitly trains the capabilities the machine does not possess — and which schools systematically suppressed in favor of the capabilities machines now outperform.

Debates & Critiques

The empirical record for CoRT is genuinely contested. Robert Sternberg's Handbook of Creativity noted that de Bono was 'more interested in the usefulness of developing ideas than proving the reliability or efficacy of his approach.' Defenders argue that practitioner evidence across decades of deployment and dozens of countries constitutes real validation even without peer-reviewed controlled studies; critics argue that without such studies the program remains a plausible intervention without rigorous proof. The AI age offers a new test: the same model directed by a CoRT-trained builder versus an untrained one should produce measurably different output if the framework works.