Programming languages do not have a why. The turing machine is an arbitrary model of computation, as is the lambda calculus, as is horn clauses. The 'why' is deciding which one is most convenient to solve a particular problem. Given that very good computer scientists come up with very different means of solving the same problem, it's clear there's no universal agreement upon which paradigm is better and that there's a lot of subjective reasoning as to why particular models are better.