A very pointed observation from a New Scientist interview article

But the idea that emotions are hardwired and universal underlies many things...

Very much so. The example that really gets me is the training of autistic children to recognise the stereotyped expressions stipulated by the classical view. This training is supposed to improve children's social functioning. But nothing changes for these kids because these facial expressions don't generalise outside the lab.

Professor Lisa Feldman Barrett interviewed by Shannon Fischer for New Scientist

March 11th 2017 Issue 3116 p. 40-43