Random polygons were used as stimuli in a two-choice multiple discrimination learning paradigm designed to test individual learning ability. A nonverbal test was developed in order to avoid the cultural and racial bias which, according to the author, is inherent in verbal tests. Information processing rate (IPR) was used as the measure of learning ability. Variables in the test design were: racial group (white or nonwhite), pacing mode (self-paced or machine-paced), and stimulus similarity. Subjects were 121 white and 39 nonwhite male Navy recruits. Over ten trials, a learning effect was demonstrated, with overall internal test reliability of .84. White performance was superior to nonwhite only in the machine-paced mode. Significant correlation between IPR and Navy General Classification Test scores was seen for the entire group, but was present only in the white subgroup when the sample was divided by race. Stimulus similarity did not prove to be a significant factor. It was concluded that a reliable, culture free test of general learning ability was practicable, although its validity with respect to actual job performance has yet to be established. (Author/BW)