IN TWO EXPERIMENTS, FEEDBACK WAS CONCEPTUALIZED AS A CYBERNETIC PARADIGM INVOLVING STUDENT-PROFESSOR INTERACTION, AND WAS USED VIA VIDEO TAPE TO IMPROVE TEACHING OF A FIVE MINUTE LESSON BY STUDENT TEACHERS. IN EXPERIMENT I, FOR THE 24 STUDENTS EQUALLY DIVIDED INTO FEEDBACK AND NO FEEDBACK GROUPS, NO SIGNIFICANT DIFFERENCES APPEARED IN 29 PERFORMANCE FACTORS AS RATED BY METHODS TEACHERS. THIS WAS EXPLAINED AS AN ERRONEOUS STATISTICAL COMPARISON THAT RESULTED IN A SLEEPER EFFECT, AND ALSO BY EXTREME VARIATIONS IN FEEDBACK CONTENT. IN EXPERIMENT II, FOR THE 19 STUDENTS DIVIDED INTO IMMEDIATE AND DELAYED FEEDBACK GROUPS IN FIVE TRIALS, EACH FOLLOWED BY SELF-EVALUATION, NO SIGNIFICANT DIFFERENCES APPEARED AMONG A HUNDRED COMPARISONS OF COVARIATES. THIS WAS INTERPRETED IN TERMS OF LOW INTERRATER CORRELATION, AND POOR CONCEPTUALIZATION. (LH)