Single- and Double-Loop Learning

The distinction between single- and double-loop learning has entered the lexicon of writers in the fields of professional and organizational learning, reflective practice, and organizational and social change. The terms are most closely associated with two American scholars—Chris Argyris and Donald Schön. For them, learning is not the accumulation of knowledge but the detection and correction of errors. Errors are mismatches between the intended and the actual results of action, whether the action was taken by individuals, groups, or organizations. Learning involves inquiry into the cause of the mismatch and revision of the action to bring about the intended consequences. This entry explains single-loop and double-loop learning and discusses the challenges in the implementation of double-loop learning.

The distinction between single- and double-loop learning is illustrated in Figure 1. It portrays the relationships between the three components of a theory of action and how feedback triggers revision of one or more of its components. A theory of action constitutes actions, the governing variables that they satisfy, and the intended and unintended consequences of those actions.

Figure 1 Single- and Double-Loop Learning

Source: Argyris (1999, p. 68). Reprinted by permission.

Note: This figure portrays the relationships between the three components of a theory of action and how feedback triggers revision of one or more of its components. A theory of action comprises the actions, the governing variables that they satisfy, and the intended and unintended consequences of those actions. In single-loop learning, only actions are revised as a result of feedback; in double-loop learning, the governing variables are revised, which then causes revisions to the actions.

The distinction between single- and double-loop learning is illustrated with the following example. Imagine a supervisor who tells an employee that his or her performance is not up to standard. How the supervisor communicates this message (the action component of the supervisor’s theory of action) is determined by his or her desire to persuade the employee to the supervisor’s point of view and to do so with the minimum possible negative reaction. These two governing variables (persuade and protect) lead the supervisor to praise much of the employee’s work and then briefly mention the concerns about the employee’s performance. The consequence is no change in the employee’s behavior. This is an error because the consequence (no change) is contrary to what the supervisor intended (improved performance).

It may be possible for the supervisor to correct the error by finding smarter ways of persuading the [Page 755]employee of his or her point of view while still protecting the employee’s feelings. This would involve single-loop learning, because the change is restricted to the action strategies, and the governing variables (persuade and protect) remain unexamined. There is no consideration of the adequacy of the values that are guiding the communication, or of the behavioral world that is created by such values, even if the revised action strategies prove effective. Double-loop learning requires adjustment of the governing variables that specify what counts as effective action. In this example, this would require inquiry into and possible revision of the supervisor’s assumption that effective communication of negative feedback can be achieved by an appropriate balance of persuasion and protection.

...