Abstract
Although behavior analysis has been criticized for failure to account for response novelty, many common behavior-analytic concepts and processes (e.g., selectionism, the operant, reinforcement, and stimulus control) assume variability both in the environment and in behavior. The importance of the relation between variability and novelty, particularly for verbal behavior, is discussed, and concepts used to account for novel behavior are examined. Experimental findings also are reviewed that suggest that variability in behavior can come under discriminative control, and these findings are applied to describe novel instances of behavior that may arise during problem solving. We conclude that variations provided and selected by the terms of the three-term contingency are powerful means for understanding novel behavior.