The modeling relation and models of complex systems expressed by non-integrable constraints were developed during ca. 1970–1987, when I worked most closely with Robert Rosen. I contrast the modeling relation within the organism itself as a necessary condition for life and evolution, as Rosen developed it in his fundamental work ‘Anticipatory Systems’, with the modeling relation within our brain as a necessary condition for understanding life, as Rosen developed it in ‘Life Itself’. Our approaches to the modeling relation were complementary. Rosen focused on the formal relational conditions necessary for life, and on the limitations that formal mathematical-symbol systems impose on our models. I focused on the physical conditions necessary for these abstract relations to be realized, and on the symbolic control in organisms that allows open-ended evolution. I contrast Rosen's views on physics and evolution in ‘Anticipatory Systems’ and later papers with his views in ‘Life Itself’, and I speculate on why they differ so greatly.