I'm not saying that the users
know that free options have lower cost-benefit tradeoffs, but in reality, experiments show that they do, specifically the on I cited.
But you raise an interesting point. It could well be that the "mental cost of even having to do a cost/benefit analysis" is too high for many people to consider paid options. That reenforces the proposal that Apple should wean users off of low (even negative) utility free options and train them to make such evaluations efficiently, the way many of us who are used to paying for software can. Again, all of this predicated on the thesis that benefit - cost - mental cost (which improves as we learn how to evaluate software) of many paid apps is higher than the average benefit of free apps.
As for my notion that people's actual behavior doesn't conform to a model of rationality, perhaps we should agree to disagree. There is plenty of research to suggest that people are bad at making calculations and require intervention.