History [of finite automata]
Simple finite automata have implicitly been used in electromechanical machines for over a century. A formal version of them appeared in 1943 in McCulloch–Pitts neural network models. (An earlier analog had appeared in Markov chains.) Intensive work on them in the 1950s (sometimes under the name sequential machines) established many basic properties, including interpretation as regular languages and equivalence to regular expressions. Connections to formal power series and to substitution systems (see page 891) were studied in the 1960s. And with the development of the Unix operating system in the 1970s regular expressions began to be widely used in practical computing in lexical analysis (lex) and text searching (ed and grep). Regular languages also arose in dynamical systems theory in the early 1970s under the name of sofic systems.