Under the definition used by Levitt and March (1988), learning is recognized once it leads to a revision of organizational routines on which behavior is based. The routines themselves encode and perpetuate lessons learned in the past. On the other hand, definitions of Fiol and Lyles (1985) and Swieringa and Wierdsma (1992) stress observable behavioral change.
The three conditions usually associated with definitions of organizational learning and their relationships are illustrated in Fig. 1, which is a parsimonious way of classifying many of the definitions. The three circles are not sets of elements. Each circle denotes that a certain condition has been fulfilled. If two or more conditions are fulfilled at the same time, it is represented by the overlap of the respective circles. For example, areas 1 and 4 depict that the change in potential behavior is realized, resulting in actual behavioral change. Most definitions of organizational learning can be represented by either areas 1 and 2 or areas 1 and 3 depending on whether change in potential or actual behavior is demanded. Some special modes of learning can also be denoted in the diagram. "Forced learning" and "imitation" found in Chinese and Hungarian joint ventures (Child & Markoczy, 1993), for instance, are depicted by area 4 assuming that the change in behavior is supported by revising the related organizational rules and practices. But in both cases, there is no change in the learner's cognition; that is, the learner does not understand the rationale behind the new rules and practices.
Prescriptive writings tend to adopt definitions which incorporate actual behavioral change more because their target audience, mostly practitioners, are action-oriented. As for descriptive studies, a crucial issue is to determine whether learning has occurred in an organization. Difficulties concerning the recognition of learning may arise if definitions that require actual behavioral change are used. The first problem is that there may be a long time lag between the time when knowledge is acquired and when it is put into practice. Furthermore, sometimes it is not easy to establish the linkage between actual behavioral change and what has been learned. An organizational action can be a product of a complex decision-making process. Existing knowledge within an organization is just one of the factors influencing the decision. Hence it is not advisable for descriptive studies to incorporate actual behavioral change in their definitions.
Recently, in an attempt to expand the framework for investigating organizational learning, Nicolini and Meznar (1995) interpret organizational learning "as a social construction which transforms acquired cognition into accountable abstract knowledge" (p. 727). They claim that their conceptualization, which is very much cognitive in perspective, is superior to the restrictive traditional interpretation of what constitutes organizational learning. Since they have neither provided empirical evidence nor illustrated how their constructs can be ope-rationalized in empirical research, their claim is yet to be validated.