A set of conditions for a consistent recovery of a subliminal stimulus

3 experimental groups of 10 Ss, each with their corresponding controls, were given mathematical problems in a tachistoscope as a supraliminal stimulus. At a level established by a pretest group, the experimental groups received subliminal answers to the problems. Group 1 attempted to solve the problems, Group 2 to guess at the answers, and Group 3 to select their answers from dual possibilities on a given list. A significant tendency was found in Groups 1 and 2 to repeat various subliminally projected digits in their answers, without the answers directly affecting their computational processes. Group 3 selected the projected answers significantly over the correct answers. The need for a positive relationship between supra- and subliminal stimuli as well as the relevancy of the task to the Ss' present activity was also observed.