The conflicting perspectives of control theory (CT) and self-efficacy theory (SET) concerning positive discrepancy creation (PDC) were tested and are discussed. According to CT, discrepancies between past performance and future goals are continually reduced. This is contrary to SET's focus on setting future goals higher than past performance levels. Participants performed several trials in a multitask environment, during which they did as many or as few problems as they chose on each of 4 intellectual tasks. Results suggest that PDC is not uncommon in a multitask environment, even after extensive task experience. Regression decomposition techniques identified 2 types of PDC: goal driven and performance driven. Need for achievement, instrumentality, and expectancy predicted the 2 types of PDC with varying success across the 4 tasks. The 2 types of PDC reflect the different theoretical approaches and these 2 self-regulation theories.