People are very bad at “games”. What explains this? Well, they aren’t computers. But if people aren’t computers, what are they? How can they best be modeled? What is lost by modeling them as purely rational? Is this loss systematic? What could be gained by a more nuanced, realistic model of human behavior?
First, people work with imperfect information. A lot of work has been done on this, and methods have been created to deal with this. I do not feel like I need to say anything more in regards to this topic.
Second, people are not perfect at processing this information. They are bad at probability, and in general calculating expectations.
Third, people often have other goals that are not correctly being modeled (i.e. punishing low offers in the accept/reject both lose game).
Fourth, processing time is limited and exerting effort inflicts a cost on the economic agent.
I see people making choices in a kind of fog of imperfect information using their imperfect means of evaluation and taking into consideration what could be gained by spending more time on the problem and getting closer to the rational choice. That is why people spend more time deciding what house to purchase rather than which brand of cookies is the best. I feel like there is some analogue to probability theory and sample size, as in a larger purchase is similar to a larger sample size. I also think things we know such as risk aversion and prospect theory can fit right into this theory. I am not sure how exactly yet.
Though I have been discussing these difficulties in terms of individuals, it seems as if a similar thing could be applied to businesses and institutional actors. In fact, the processing time and cost might be more easily quantifiable and modelable.
To what extent are these imperfections in individual agents either smoothed out or amplified in the aggregate? Are the deviations from rationality systematic or random? Do these variations cancel each other out in some way, or magnified?