IF COMPUTER PROGRAMS ARE TO SERVE AS USEFUL MODELS OF COGNITIVE BEHAVIOR, THEIR CREATORS MUST FACE THE NEED TO ESTABLISH AN INTERNAL ORGANIZATION FOR THEIR MODEL WHICH IMPLEMENTS THE HIGHER LEVEL COGNITIVE BEHAVIORS ASSOCIATED WITH THE HUMAN CAPACITY FOR SELF-DIRECTION, AUTOCRITICISM, AND ADAPTATION. PRESENT COMPUTER MODELS OF COGNITIVE BEHAVIOR HAVE EMBODIED THE ABOVE FUNCTIONS IN THE COMPUTER PROGRAMER RATHER THAN IN THE CAPABILITIES OF THE PROGRAM. THE CENTRAL FUNCTION SHOULD, HOWEVER, RESIDE IN A HIERARCHY OF CONTEXTER ROUTINES WHICH OPERATE PARALLEL TO AND CONTROL THE ROUTINES WHICH ACTUALLY PERFORM THE REQUISITE OPERATIONS. THE GREATEST PROBLEM IN THE DEVELOPMENT OF COMPUTER MODELS INVOLVING A CENTRAL FUNCTION IS THAT OF BEING ABLE TO DESCRIBE BEHAVIOR IN SUCH A WAY THAT THE CONTEXTER PROGRAMS CAN UTILIZE THIS DESCRIPTION IN PERFORMING THEIR FUNCTIONS AT ALL LEVELS WITHIN THE MODEL. ACCOMPLISHMENT OF THESE GOALS WOULD BE FACILITATED BY A MEMORY STRUCTURE WHICH ENABLES THE PROGRAM TO STORE AND RETRIEVE DATA UNDER ITS OWN CONTROL, AND BY A LANGUAGE WHICH SEPARATES THE MODEL OF COGNITIVE BEHAVIOR FROM THE MECHANICS OF THE UNDERLYING COMPUTER PROGRAMING LANGUAGE. THIS ARTICLE IS A REPRINT FROM "BEHAVIORAL SCIENCE," VOLUME 12, NUMBER 2, MARCH 1967. (HW)