I believe the readers of this post will be interested with the gentle introduction I provide on utility theory and partial moments:
https://www.linkedin.com/pulse/behavioral-finance-partial-moments-fred-…

Further, a recent paper we have on behavioral finance and utility theory:
https://www.scribd.com/document/230067430/Behavioral-finance-in-financi…

A quick note on Markowitz... Harry has been behavioral since 1959, as nearly 1/4 of his book was devoted to utility theory. He has long argued that variance is a "good enough" estimate to the underlying quadratic utility function he assumes for the investor. Research over the following decades demonstrated that individuals exhibit markedly different utility functions than the quadratic.

Harry also spends a significant amount of time extolling the benefits of semi-variance over variance (and geometric means over means). However, due to computational limitations at the time, variance was used.

The quadratic utility function and the use of expected values as beliefs of future performances are indeed poor assumptions. Statistics that can capture all utility functions (partial moments) and more representative estimates of future performances address these MPT shortcomings.

CAPM and EMH essentially assumed utility theory away, and, well, we all know how that worked out!