An Investigation of Heuristics Used in the Construction of Conceptual Models

Abstract

Conventional Soft Systems Methodology (SSM) is a seven stage process for tackling real-world problems, most commonly those associated with organisations (Checkland, 1981). The methodology has changed over the years, two recent developments being the recognition that the seven stage logic-driven stream of enquiry is inadequate without also considering the problem situation as a culture (Checkland and Scholes, 1990), and the recognition that the conventional real world/systems thinking world dichotomy may not be helpful (Tsouvalis and Checkland, 1996). The modelling process at the heart of SSM, in which root definitions (RD) are formulated and conceptual models (CM) are derived from them, has also been subjected to greater scrutiny. Among the questions which have been asked are: whether an explicit RD and CM are both necessary (Watson, 1991); the meaning of the logical dependency arrows in CMs (Gregory, 1993); and the nature of the link between the RD and CM (Checkland and Tsouvalis, 1996; Cross and Turk, 1996). This paper is concerned with the last question, and argues that the link between RD and CM is not as mechanical a process as some maintain, and indeed that the way in which experienced SSM practitioners construct CMs from RDs is still poorly understood. Section 2 of the paper sets out some arguments that the RD-CM link is problematic, and the justification for the present research. Section 3 discusses an experimental program to discover the heuristics used by experienced SSM practitioners. Section 4 demonstrates the application of the approach to an SSM “expert” and presents some heuristics for deriving CMs from RDs derived from the analysis. Section 5 summarises the conclusions of the work and future directions.