Several studies have been reported in recent years concerning the effect of non-informative punishment for error, and non-informative reward for correct response, upon learning. Working independently, three workers almost simultaneously reported studies in this field. Eisenson (3), in a multiple-choice situation with human subjects, found that non-informative reward, while less effective than informative reward in stimulating the repetition of correct responses, still had a decided positive effect. In the same experiment he found informative and non-informative punishment for errors to be about equally effective. Bunch (2) reported data from a stylus maze experiment which indicated that the effectiveness of shock for error is only partly attributable to its informative value. That effectiveness over and above its informative value she called the 'punishment' value. Gilbert (4), also working with a stylus maze problem, drew the cautious conclusion that "electric shock may, under certain conditions at least, have a facilitating effect upon learning apart from its guidance value, and that there is evidence (although less positive) that learning motivated in this way is followed by better retention than that which follows less highly motivated learning." He later confirmed this finding in a punchboard maze situation (5).During the progress of a study reported in a recent paper (i), the writer and Gilbert ran a group of 24 subjects who learned the stylus maze employed in that experiment without shock for error. This additional group was intended to throw further light upon the problems of non-informative shock, was not in any way concerned with the main study of shock specificity, and was therefore not discussed in the report on that experiment. The main experimental group received shock in connection with certain of the culs-de-sac, and no shock in others. The control group was run on the hypothesis that a comparison between the rate of elimination of non-shock errors in the main group and the rate of elimination of the same errors in the non-shock group would be equivalent to a comparison between a non-informative shock group and a non-shock group, since the main group was receiving shocks which gave no information as to non-shock errors. If it were found that non-shock errors in the main group were eliminated more rapidly than the same errors in the control group, it would constitute additional evidence as to the effectiveness of non-informative shock in learning. The following is a report on this study.