Relatively simple machines can learn to
recognize happy and unhappy: human
facial expressions, voices, body language
 

Use this as machine reinforcment values
 

Positively reinforced when we are happy
Negatively reinforced when we are unhappy
 

Simulation predicts future human happiness