Maybe I'm not understanding what you mean here but there definitely is an area under the ROC curve of a person (or radiologist or whatever you are interested in). You can push around people's decisions by communicating to them different base rates or costs/rewards of errors/correct responses and can also trace out ROC curves by asking them to give confidence ratings (where each rating corresponds to a point on the ROC curve). These are frequently studied in psychophysics and in medical decision making.

You *can* do this, but it's such a forced exercise that we learn bizarre things from it. We know that experts are not rational utility maximizers. They make decisions in far more complex and valuable ways. Treating people like they are deficient computer algorithms and making arguments about the results doesn't lead us to better people.

This thing has so many names. It is called hypothesis testing in detection theory, and it is called contingency table in game theory, or just Pascal's wager. But in its essence, isn't it just one step stochastic dynamic programming?

Maybe I'm not understanding what you mean here but there definitely is an area under the ROC curve of a person (or radiologist or whatever you are interested in). You can push around people's decisions by communicating to them different base rates or costs/rewards of errors/correct responses and can also trace out ROC curves by asking them to give confidence ratings (where each rating corresponds to a point on the ROC curve). These are frequently studied in psychophysics and in medical decision making.

You *can* do this, but it's such a forced exercise that we learn bizarre things from it. We know that experts are not rational utility maximizers. They make decisions in far more complex and valuable ways. Treating people like they are deficient computer algorithms and making arguments about the results doesn't lead us to better people.

This thing has so many names. It is called hypothesis testing in detection theory, and it is called contingency table in game theory, or just Pascal's wager. But in its essence, isn't it just one step stochastic dynamic programming?