This perspective combines elements of systems thinking (an essential element of cybernetics) with the subjective epistemology typical of organization development. It tries to model reality in a subjective way by emphasizing the mental models people use. It is a regulation approach, because it stresses the opportunities for shared mental models.
An important representative of this perspective is (again) Peter Senge, who states: " ...What we carry in our heads are assumptions. These mental pictures of how the world
works have a significant influence on how we perceive problems and opportunities, identify courses of action, and make choices" (Senge, 1990b, p. 12).
Organizational learning results in these mental models. This is often difficult, because many of our assumptions are tacit and hence difficult to test. The elicitation of mental models is therefore an essential step in learning. A good model in Senge's view not only explains our reactions to events, but explains it by understanding some deeper lying systemic structure. Senge has developed so-called system archetypes, some general ways in which systems are supposed to behave, that simplify the model elicitation and the detection of systemic structure. One of these archetypes is 'Shifting the Burden', explained as follows:
" ...A short term 'solution' is used to correct a problem, with seemingly happy immediate results. As this correction is used more and more, fundamental long- term corrective measures are used less. Over time, the mechanisms of the fundamental solution may atrophy or become disabled, leading to even greater reliance on the symptomatic solution. Classic examples: using corporate
human resource staff to solve local personnel problems, thereby keeping managers from developing their own interpersonal skills" (Senge, 1990b, p. 7). See figure 4.6.
In 'The Fifth Discipline' Senge mentions 9 additional archetypes. I do however hesitate in using them, because they can lead to too much pre-conceptualization. This could lead to people trying to see their problems in terms of a chosen archetype, and lacking the creativity to make a model that might better suit their situation. Hence, I developed a software tool that supports the creation of models by means of Critical Success Factors, called CSFmatrix. An essential assumption of CSFmatrix for organizational learning is that models for organizational learning are used not as individualistic models, but as shared knowledge. This implies several things:
1. Specific pieces of the model's puzzle are brought together by participants in the model development process, for example by selecting specific topics. This can be done by naming them in the group and a facilitator writing them down on a whiteboard. One can also use hexagons and attach them on a (magnetic) board (cf. Morecroft and Van der Heijden, 1992).
2. The participants must agree on a clear understanding of the meanings (semantics) used in the expressions mentioned under number 1. Participants must find a common vocabulary in which they can understand each other. This also can lead to a considerable reduction of topics.
3. Participants also need to agree on the importance and validity of each other's statements. To have an effective third step, people must pay enough attention to the first two. The result should be a list of not more than five or seven critical success factors (cf. Rockart, 1979).
4. The participants now connect the major issues with each other, in terms of possible negative or positive causal relationships. They can do this in a group discussion to reach a shared vision. The result is a filled-in n*n matrix.
5. CSFmatrix now can automatically generate a system dynamics model (with reinforcing and balancing relationships). The facilitator presents this model to the group which then discusses the results, and possibly comes up with some modifications. It is my experience that participants, especially in this phase, get a strong 'Aha Erlebnis' (suddenly seeing the light). Often they want to go back to adjust the list of critical success factors and relations. See figure 4.7.
Although this approach stresses the finding of shared mental models, it can be used for analyzing conflictual situations as well. For instance, participants can learn to understand the basic assumptions of the conflicts they are in. Lee, Courtney and O'Keefe (1992) and Acar and Heintz (1992) also developed computer-based tools that support the analysis of incompatibility between models. Both publications
have found methods to solve conflicts when the models are complementary. When the models are not complementary, a synthesis is required to construct a joint model. A suggestion for an information system that could cope with model incompatibility is given by Hedberg and Jönsson (1978), but hardly elaborated yet.
4.6.2 Limitations of Soft Systems for Organizational Learning
The soft systems perspective does not exclude the possibility of objectivism. For instance, a group can find sources of ineffectiveness and the means for a more effective control of an organization. The results of this analysis can be used as input for the design of a MICS. In principle the soft systems perspective only provides a tool that can be used for modeling objective (cybernetics) or subjective (OD) reality. The technique when used in the case of order must lead to one model that relates all important issues. In the case of the conflict view it leads to several subjective models. For further reading in the soft systems perspective we recommend: Rosenhead et al. (1989) and the special issue in the European Journal of Operational Research on modelling for learning (Morecroft and Sterman eds., 1992). Both publications use a subjective epistemology, but differ in their approach. The first places emphasis on conceptualizing how we talk about problems. The second stresses the importance of describing major variables and their interrelations in systems dynamics models. In my opinion conceptualizing preceeds modeling, however, several interations between modelling and conceptualizing will improve both. Soft systems modelling does not always have to be followed by system dynamics modelling. Many other techniques can be used to further define vaguely defined problems (for instance PAM, Stamper et al., 1988; Kolkman, 1993) When reality is clearly defined, soft systems lose their value, and hard modeling techniques are more relevant (Wijnhoven, 1992a). Also when major conflicts about problems exists, soft modelling will not solve them. It then depends on how the actors in the political arena want to regulate their conflict. Precise data about possible outcomes are more valuable then and hard models are also required (Coleman, 1972; Teich, 1991).
4.7 The Scientific Management Perspective of Organizational Learning