The paper contrasts the classical contributions in behavioral economics which investigated risk, time or social preferences independently of each others before the middle of the 2000s with the more recent contributions that jointly investigate these type of preferences by pair (e.g., how risk and time preference interact). It suggests that, while standard models could be used as normative benchmarks to judge some behavioral phenomena as non-rational in the classical contributions, this is more complicated in the recent contributions because interactions between types of preferences tend to decrease the normative force of standard models.