I was very interested to read the query by Mohamed Saleh with its
Lakatosian view of the field and the subsequent contributions. (Sorry to
join this conversation late but I have been away)
I have been thinking about this question along with a colleague in the
Dept of Philosophy, Logic and Scientific Method here at LSE and have also
had a useful exchange with Greg Scholl on the matter. Theres a Working
paper in press at the moment that deal with the question and as of one week
ago I submitted a version to System Dynamics Review.
I was looking at things from a social theory perspective and was
considering the systems theory of Talcott Parsons and the criticisms made of
this Grand Theory approach by C. Wright Mills. The issue that I was dealing
with was that system dynamics is criticized as deterministic. Heres my
twopennyworth (Ive made bold the sections most closely relevant to the
Lakatos question):
Crude appeal to grand theory?
The word determinism is also used to describe the position that cause and
effect are related by laws which exist outside of human subjectivity and
which together form a timeless grand theory. Relevant in this context is
Maxwells principle of determinism, that "The same causes will always
produce the same effect" and that this is so because the laws underlying the
principle describe the operation of invariant relations.
The pursuit of grand theory via causal laws is based on the Humean
tradition that science implies explanation and that explanation involves
relating phenomena to be explained to other phenomena via such causal laws.
The value of the approach is that a set of antecedent conditions may be
combined with relevant laws to make a prediction about an event. Even in
natural science the logical status of grand theory is not a closed issue
since the exact formulation of causal laws is complex and the philosophical
weaknesses in the different formulations of causality would apply to any
field seeking to employ the concept. However, a point made very clearly by
Popper is that determinism via the principle of causality is quite different
from determinism as prophecy and the latter is alive and healthy as a
central concept in natural science.
This form of the determinism criticism is therefore that system dynamics
posits a grand theory about the existence of objective causal laws in social
systems. It has been observed that an assumption of the field is that,
"well-defined laws govern behaviour" (1, p.33) and that, "Forrester talks
about fundamental laws of nature and the social sciences" (2, p.13).
System dynamics has a ranges of theories which appear deterministic in
this sense. For example, models are said to be causal theories and Forrester
writes of a "general systems theory" (3, p.135) and lays out Principles of
Systems (4). There is also the grand claim that underlying models is, "a set
of principles..., incomplete as they may be, that I believe do represent the
actual nature of physical and social reality" (5, p.250).
On the nature of the theories in system dynamics System dynamics offers
theories at four different levels. At the highest level is the claim that
the time evolutionary behaviour of social systems is explainable in terms of
feedback loops and state variables. This very crude statement would need to
be elaborated to establish the plausibility of the claim.
(A slightly less crude statement of the position might be:
Social systems frequently behave in ways which are counter to the intuition
of actors implementing policies aimed at influencing their behaviour.
Because actors who are part of social systems collect information about
their environment and, in light of that information, take action to
influence aspects of the state of the system, the aggregate behaviour over
time of such systems can useful be explained using the concepts of feedback
loops and stocks. The feedback loops are a plausible representation of the
collect/influence/collect cycle of activities and the stocks are a plausible
representation of the system state variables. Models based on these concepts
offer a representation of the causal links between the variables which
allows the rigorous deduction of the consequences of those links. Such
models are therefore theories of the structural source of particular
aggregate behaviours and can be used to make deductions about the mode of
behaviour that will flow from implementing a given influencing policy.)
The important point here is the nature of this claim. This is not a grand
content theory. There are no specific variables or conceptual categories.
For example, it is not suggested that socialisation or class
stratification are meaningful concepts which are necessary to explain the
evolution of all societies. In contrast, this grand claim of system dynamics
is a structural theory - it makes a grand methodological claim about how
certain types of phenomena might be explained.
Moving down to the next theory, we find the principle of how the concepts
of feedback loop and stock should be used to construct models (4). This is
not a grand theory. This is a representation theory, or scheme, proposing a
way of implementing the above grand methodological theory.
A more specific theory is that, unassisted, humans cannot infer the
behaviour of systems represented in the above fashion in a way which is
logically consistent; computers are needed to deduce the behaviour. This is
an empirical claim. As such it is well supported by data (6). Models
therefore allow the application of logic to reveal essentially tautological
but, in practical terms, hidden results (7).
Finally, we come to the idea that each model is a theory. These are
clearly minor content theories, albeit valuable ones. A model is a
concatenation of causal laws offering a plausible representation and hence
explanation for a behaviour mode.
With this clarification it is clear that none of the above theories
proposes an invariant causal law and that there is no crude grand (content)
theory. The only universal law/theory on offer is a grand methodological, or
structural theory, associated with a representation scheme. System dynamics
offers a new structure for thinking about causality but it does not specify
the content of that structure. The laws are structural, or representational,
they have no specific content. This position can still be criticised but it
does not attract the determinism-related criticisms attached to grand theory
in the sense of Parsons and Mills.
Position clarified, but Case open
A suitable note on which to end is that the grand structural theory of
system dynamics, along with the theories of logical deduction and
representation stand or fall not as separate elements to be criticised but
as a coherent research program (8). They should be considered in terms of
the plausibility and coherence of its grand structural claim, the
reasonableness of its representative scheme and its empirical success in
providing explanations for a range of novel phenomena. Viewed in these
Lakatosian terms, the various theories of system dynamics form the hard
core of the field and the healthy flow of diverse applications of the
approach indicates that the research program is a progressive one. Whilst
the issues raised here are still open and the clarified position will be
objectionable to some, system dynamicists are happy to justify their
position in terms of their ability successfully to illuminate the phenomena
to which the field applies. It is on those terms that any of the theories
and laws of system dynamics should be judged.
References
1 Flood, RL and Jackson, MC (1991). Creative Problem Solving: Total
Systems Intervention. Wiley: Chichester.
2 Bloomfield, B (1982). Cosmology, Knowledge and Social Structure: The
case of Forrester and system dynamics. Journal of Applied Systems Analysis
9: 3-15.
3 Forrester, JW (1968) Industrial Dynamics - After the first decade.
In: Collected Papers of Jay W. Forrester (1975 collection). Wright-Allen
Press: Cambridge, MA, pp 133-150.
4 Forrester, JW (1968). Principles of Systems. MIT Press: Cambridge,
MA.
5 Forrester, JW (1994). System dynamics, Systems thinking, and Soft
OR. System Dynamics Review 10: 245-256.
6 Sterman, JD (1994). Learning in and about complex systems. System
Dynamics Review 10: 291-330.
7 Simon, HA (1969) Understanding the Natural and the Artificial
Worlds. In: The Sciences of the Artificial. MIT Press: London, pp 3-29.
8 Lakatos, I (1974) Falsification and the Methodology of Scientific
Research Programmes. In: Lakatos, I and Musgrave, A (eds). Criticism and the
Growth of Knowledge. CUP: Cambridge, pp 91-196.
Regards,
David
____________________________________________________________________________
________________
Dr. David C. Lane
Operational Research Department
London School of Economics and Political Science
Houghton Street Tel.
(44) (0) 171-955-7336
London WC2A 2AE Fax. (44)
(0) 171-955-6885
United Kingdom
e-mail:
d.c.lane@lse.ac.uk