I've come across an article (in Russian), which describes a nonlinear two-component color vision theory made in 1975 by S. Remenko. The article heavily criticizes trichromatic theory as very imperfect. There's a whole page related to the criticism. Some of the citations there refer to Feynman's lectures vol. I ch. 35, where also some doubts are expressed. See e.g. the part by Feynman:
Using the three different kinds of color blindness, the three pigment response curves have finally been determined, and are shown in Fig. 35–8. Finally? Perhaps. There is a question as to whether the three-pigment idea is right, whether color blindness results from lack of one pigment, and even whether the color-mix data on color blindness are right. Different workers get different results. This field is still very much under development.
But the only English-language place I've found mentioning Remenko's two-component theory was the book "Computer Systems for Healthcare and Medicine" by Piotr Bilski and Francesca Guerriero, where the first name also looks suspiciously Russian.
So I'm somewhat skeptic about credibility of all this Remenko's theory and the criticism of the trichromatic theory. But having not much expertise in color vision theories, I thus prefer to ask the experts: is the theory by Remenko actually credible? Is it even known outside of Russian-speaking world? Has it been falsified? Is trichromatic theory really flawed in any significant way so that a replacement theory would be required (are Feynman's remarks out of date?)?