Editing Organizational learning (section)

=== Theoretical models ===
Attempts to explain variance of rates in organizational learning across different organizations have been explored in theoretical models. Namely the theoretical models conceived by John F. Muth, Bernardo Huberman, and Christina Fang.
* The Muth model (1986) was the first to represent the learning curve in a log-linear form and focused on cost effectiveness in organization processes. This model looks at the relationship between unit cost and experience, stating that cost reductions are realized through independent random sampling, or randomized searches, from a space of technological, managerial, or behavioral alternatives. This model did not aim to explore variation across firms, but solely looked at improvements in production with experience within a single firm.<ref>{{Cite journal|last = Muth|first = John F.|date = August 1986|journal = Management Science|doi = 10.1287/mnsc.32.8.948|volume=32|issue = 8|pages=948–962|title = Search Theory and the Manufacturing Progress Function}}</ref>
* The Huberman model (2001) filled that void and aimed at explaining the variation missing from Muth's model and focuses on finding increasingly shorter and more efficient paths from end to end of an assembly process. This model is visualized best in a connected graph with nodes that represent stages in a process and links that represent the connecting routines. By way of this model, learning can occur through two mechanisms that shorten the route from the initial stage to the final stage. The first is by some shortcut that can be identified by looking at the nodes and mapping and discovering new routines, the ideal goal being able to eliminate certain touch points and find shorter paths from the initial to final node. The second mechanism involves improving the routines: the organization can work to select the most efficacious link between two nodes such that, if an issue ever arises, members of an organization know exactly whom to approach, saving them a considerable amount of time.<ref name=":03" /> 
*The Fang model (2011) shares a major goal with the Huberman model: to gradually decrease the steps towards the final stage. However, this model takes more of a "credit assignment" approach in which credit is assigned to successive states as an organization gains more experience, and then learning occurs by way of credit propagation. This implies that as an organization gains more experience with the task, it is better able to develop increasingly accurate mental models that initially identify the values of states closer to the goal and then those of states farther from the goal. This then leads to a reduced number of steps to reach the organization's final goal and can thus improve overall performance.<ref name=":03" />