Rationality does not guarantee results at the single human scale.
Making a decision that is statistically correct only works out in the long run, over a number of such decisions.
You can make a decision that was the correct decision given the information you had, and then it doesn't work out.
If the slaves have no money, but the slave owners are wealthy, does that mean that the slaves who try to use their time and energy to revolt don't care, because they have no money? And the wealthy owners, who free a slave every 20 years because they own so many one more or less makes little difference to them, they somehow care more?
No, that's a ridiculous theory.
Take any cognitive bias that is supported by previous experimental data. Replicate to confirm.
Subject students to various training regimens, with control group.
Test again for presence of cognitive bias, note any improvements.
Repeat, repeat again for other known cognitive biases.
Not perfect, but it should be enough to make some headway.
Also just subjecting a student to a battery of tests, ideally creative stuff potentially involving real life scenarios not just written tests, to look for all sorts of cognitive biases.
Should the student try to game the system by learning, well great!