There have been many applications of and successes with environmental enrichment for captive animals. The theoretical spine upon which much enrichment work hangs largely describes why enrichment should work. Yet, there remains no clear understanding of how enrichment should be applied to achieve the most beneficial results. This lack of understanding may stem in part from the assumptions that underlie the application of enrichment by practitioners. These assumptions are derived from an understanding that giving animals choice and control in their environment stimulates their motivation to perform behaviors that may indicate a heightened state of well-being. Learning theory provides a means to question the manner in which these constructs are routinely applied, and converting learning theory's findings to optimality predictions suggests a particularly vexing paradox—that motivation to perform appears to be maintained best when acquiring a payoff for expressing the behavior is uncertain. This effect occurs even when the actual value of the payoff is the same for all schedules of certainty of payoff acquisition. The paradox can be resolved by invoking rewards of an alternative type, such as cognitive rewards, or through an understanding of how the average payoff changes with changes in the probability of reward. This model, with measures of the average change of the payoff, suggests testable scenarios by which practitioners can measure the quality of environmental uncertainty in enrichment programs. Zoo Biol 28:609–622, 2009. © 2009 Wiley-Liss, Inc.