[dogfishbar]

I spent a lot of time on this. See M-LISP: a representation-independent dialect of LISP with reduction semantics, TOPLAS, 1992, the relevant bit is in section 2.

It's true that J. McCarthy had only a passing familiarity with LC. M-expression LISP, as it was originally conceived, was all about first-order recursion schemes over S-expressions. But due to a very simple error in the base case of an inductive definition, LISP 1.0 "featured" or "supported" higher-order functions, ala LC.