TWO STUDIES CONCERNED WITH THE REPRESENTATIONAL ROLE OF DEMONSTRATIONS IN TEACHING CONCEPTS AND PRINCIPLES OF SCIENCE ARE REPORTED. THE TWO STUDIES WERE PERFORMED WITH DIMENSIONS OF VISUAL TV PRESENTATIONS VARIED IN THE CONTEXT OF PROGRAMED LESSONS. STUDY ONE COMPARES THE EFFECTIVENESS OF REALISTIC (LIVE) WITH NONREALISTIC (ANIMATED) DEMONSTRATIONS UNDER INTUITIVE AND NONINTUITIVE CONDITIONS. TIME-TO-COMPLETE SELF-PLACED VERBAL PROGRAMS, ERRORS ON PROGRAMS AND ACHIEVEMENT TESTS, AND INTEREST AND CREDIBILITY MEASURES YIELDED COMPARATIVE DATA. THE DATA SUGGEST THAT NONINTUITIVE OUTCOMES MAY HAVE LESS CAPACITY FOR CONFIRMING STUDENT PREDICTIONS OF EXPERIMENTAL OUTCOMES THAN INTUITIVE OUTCOMES. DIFFERENCES BETWEEN INTUITIVE AND NONINTUITIVE DEMONSTRATIONS WERE SIGNIFICANTLY AFFECTED BY THE MODE OF PRESENTATION. RESULTS SUGGEST REALISTIC AND UNREALISTIC PRESENTATIONS MAY HAVE THE CAPACITY TO REINFORCE ATTENDING BEHAVIORS. THE CAPACITY APPEARS TO DEPEND ON THE DEGREE OF INTUITIVENESS OF THE PHENOMENA DEMONSTRATED. STUDY TWO COMPARED THE EFFECTS OF DIRECT AND INDIRECT EXAMPLES. THE STUDY WAS CONCERNED WITH THREE PROPERTIES OF THE EXAMPLES--CONVERGENCE OR DIVERGENCE OF STRUCTURAL AND FUNCTIONAL CHARACTERISTICS, THE DEGREE OF INTERNAL SIMILARITY AMONG EXAMPLES WITHIN A SERIES OF EXAMPLES ILLUSTRATING A CONCEPT, AND THE AVAILABILITY OF MEDIATING VERBAL RESPONSES. FEW STATISTICALLY SIGNIFICANT DIFFERENCES WERE OBTAINED. (WN)