In this study of errorless learning a procedure is tested which allows the subject himself to adjust the speed at which a supplementary stimulus aid is withdrawn. A standard match-to-sample apparatus with 1 sample window above and 4 matching (response) windows below was used. Tokens were delivered for correct responses and the correct matching windo could receive extra illumination to provide the supplementary stimulus aid. Four children between the ages of five and eight from a lower income neighborhood served as subjects. The research involved 3 phases: (1) a preliminary measurement of task performance without the extra-dimensional cue, (2) a set of trials in which the subject could produce the supplementary aid, and (3) a set of trials in which the subject could produce the extra cue, but only at the cost of a subsequent loss of tokens. It was hoped that in this final phase a self-programmed fading out would occur. However, the subjects never phased out the supplementary stimulus aid, possibly because of the time lag between task aid and token loss. (MH)