The second reason for the difficulty of achieving accurate learning is that nowadays organizations face an increasingly complex and fast changing environment. Even with sharp analytical ability and diligence, it is often difficult, if not impossible, to identify correctly the exact cause-effect relationships of the events concerned. In many cases, cause and effect do not follow a unidirectional and linear relationship-reciprocity and feedback loops are common. An additional obstacle of learning from experience is that samples of experience are usually small in size. "Historical events are observed, and inferences about historical processes are formed, but the paucity of historical events conspires against effective learning" (March, Sproull, & Tamuz, 1991, p. 1). Various imperfections of organizational learning have been summarized in Levinthal and March (1993).
Some people may suggest that in order to avoid the complications of incorrect learning, we can define learning as the acquisition of accurate knowledge only. The definition of Fiol and Lyles (1985) shown in Table I seems to move along this direction because "better knowledge" should mean that it is closer to reality than the previous knowledge held by an organization. However, empirically it would be hard, and sometimes impossible, to assess how far the knowledge learned by an individual or an organization corresponds to reality. It is advisable for descriptive researches on organizational learning to adopt a more neutral meaning of the term. The actual connection between learning and performance is an issue to be determined empirically instead of being assumed in the definition.
Despite all the previous reservations about its usefulness, learning usually, though not always, increases an organization's capacity to perform better. An organization which is quick to correct its errors and reacts fast to environmental changes should, on average, outperform one which seldom learns from past mistakes. In addition, lessons learned in the past, if properly stored in the organizational memory, are an important source of knowledge for members of the organization to draw upon. An empirical study on learning among joint venture sophisticated firms found that the accumulated experiences of these firms enabled them to manage their joint venture relationships more efficiently (Lyles, 1988). That is why prescriptions for building a learning organization have flourished in recent years. An evaluation of these writings is presented in the next section.