Hello everyone,

I am reading Gabaix paper on sparse max, and am wondering about the nature of the 'infinite regress' problem in his endogenous inattention problem formulation? He proposes to use a Taylor approximation for step 1 when the attention decision at the beginning of agents' life is made. Does he do this because the rational inattention problem effectively becomes more difficult to solve for behavioral agents than under rational expectations (and is simply unrealistic); or because of analytical complications (which I fail to see)?

thanks in advance.