The likelihood that an individual will perform consistently on different psychological dimensions (the “consistency across modalities” hypothesis) was examined by measuring self/nonself-segregation. 108 males aged 19 to 22 yr. were given a simple perceptual task (field dependence/independence) and a self-concept measure, the True-False Ratio dimension of Fitts' Tennessee Self-concept Scale. The results tend to support the hypothesis; however, most support derives from the extreme ends of the distribution. Data suggest the distribution of the True-False Ratio can be described as neither linear nor continuous but rather as discrete, indicating three distinct groups.