Early Precursors
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
136 137 138 139 140 141 142 143 144 145 146
By early precursors, I refer to the streams of thought that are closely related to causal
mapping, although they may not have been the sources of the original formulation of the
tool. Although it is almost impossible to sketch all possible precursors, at least two
distinct streams of thought have close affinity to the original causal mapping technique:
structure of arguments, and industrial dynamics.
Structure of arguments. The idea that arguments can be represented has been well
established in the philosophy of science for a quite a long time. A specific example was
provided by Toulmin (1958): His analysis scheme is complex as it embraced a broader set
of foci than the ones we find in contemporary causal mapping. The scheme responded
directly to the need for a methodology that systematically probes the content, logic, and
reasonableness of an argument, irrespective of the discipline or context and intent of the
argument. The Toulmin framework was intended to achieve three purposes:
1) to enable the elements (and thus the structure) of any “argument” to be captured
and delineated;
2) to allow any individual (whether the argument purveyor, opponent, or interested
third party) to assess the quality of reasoning at the heart of the argument; and
3) to facilitate the comparison and assessment of two or more arguments, that is, to
identify differences among arguments and to determine what these differences
mean to both the reasoning and the outcomes.
The Toulmin framework consists of a number of elements: data, warrants, backing for
each warrant, conclusion, and qualifiers including the conditions for rebuttal. The core
of any argument is always woven around data to conclusions via warrants: some set of
data allows a claim or conclusion (that is, an inference) to be drawn because a warrant
enables a connection to be made between data and conclusions. Data can be simple,
descriptive facts, historical statements, or projections about the future such as descriptions
of the current conditions in the economy, its historic performance along multiple
indicators, or judgments about the direction of emerging economic change. The overall
intent of the framework is to test and establish the merit of, or justification for, the claim
or conclusion.
From the 1950’s onward, these ideas began to migrate into management circles. For
example, the Toulmin method was employed by management scholars such as Mason &
Mitroff (1981) to facilitate strategic planning in organizations. More recently, this method
has been used to deconstruct entire theories (Narayanan & Fahey, 2005). For empirically
oriented scholars desirous of tracking social phenomena, this method was too complex
and fuzzy since it required a large number of researcher-imposed judgments. For this
reason, Toulmin was used mostly in the deconstruction of a theory or in interventions,
not in empirical works. Nonetheless, Toulmin vividly underscored the idea that arguments
can be examined as social facts.
Industrial dynamics. In many ways, the field of industrial dynamics that originated with
Jay Forrester at MIT incorporated many of the central features of causal maps. Industrial
dynamics aimed to describe the dynamics of a system (a firm, an industry, a city, a region,
and even the world) with the aid of a mathematical representation of the system as nodes
and flows. Using the power of computers, Forrester and his colleagues wanted to examine
the behavior of the system under study. Forrester argued:
“As industrial societies emerged, systems began to dominate life as they
manifested themselves in economic cycles, political turmoil, recurring financial
panics, fluctuating employment, and unstable prices. But these social systems
became so complex and their behavior so confusing that no general theory
seemed possible. A search for orderly structure, for cause-effect relationships
(emphasis mine), and for a theory to explain system behavior gave way to a belief
in random, irrational causes.” (1968, pp.1-2)
The feedback and related principles developed in electrical engineering formed a basis
on which to formulate a set of partial differential equations to capture system dynamics.
A central facet of the system dynamics model is the incorporation of two-way causality
that, in social sciences, was relatively less prevalent until the advent of the systems
dynamic principles.
Unlike the Toulmin analysis, which was highly qualitative, system dynamics was highly
quantitative, and hence, did not widely diffuse into organization sciences. The sheer
mathematical sophistication required for its effective use, and the attendant information
requirements, made it ill-suited for much of the social science work. Nonetheless the idea
that system behavior can be depicted and analyzed can be traced to industrial dynamics.
Although the influence of Toulmin analysis and industrial dynamics were, at best,
indirect, we can identify several immediate precursors to causal maps.
Immediate Precursors
Axelrod identifies five fields from which he has drawn inspiration to develop his cognitive
mapping approach: (a) psycho-logic, (b) causal inference, (c) graph theory,(d) evaluative
assertion analysis, and (e) decision theory.
Psycho-logic. Abelson and Rosenberg (1958) developed a mathematical system to deal
with a person’s cognitive processes called psycho-logic. Their system uses points and
arrows, with points referring to “thing-like” concepts, and arrows expressing associations
between concepts. As Axelrod notes, there are two major differences between
cognitive maps and psycho-logic. First, nodes are variables that can have different
values in a cognitive map, and not “things” as in psycho-logic. This makes cognitive
mapping an algebraic system, not a logic system. Second, arrows in cognitive maps are
representations of causal assertions, not attitudinal associations. Axelrod goes on to
suggest, “Although the interpretations of the two systems are different, from a strictly
mathematical point of view, a cognitive map can be regarded as a generalization of
psycho-logic.”
Causal inference. The statistical literature of causal inference was developed by Simon
(1957) and Blalock (1964) to estimate the parameters appropriate to describe a given body
of data. This literature is credited by Axelrod with the idea that points can be regarded
as variables, and arrows can be regarded as causal connections between the points.
However, cognitive mapping does not incorporate the complex calculations typically
involved in the causal inference literature.
Graph theory. Graph theory and its mathematical ideas have been employed in both
psycho-logic and cognitive mapping. It includes concepts such as paths, cycles, and
components that are useful in the analysis of complex interconnections. Cognitive
mapping uses graph theory, “but generalizes it by allowing the points as well as the
arrows to take on different values.”
Evaluative assertion analysis. Osgood, Saporta and Nunnally (1956) developed this
analysis, which provides a method for systematically and reliably coding the structural
relationships between pairs of concepts from a document. Axelrod’s cognitive mapping
method owes the coding process to evaluative assertion analysis.
Decision theory. This field, which has close affinity with the Operations Research
discipline, was well developed by the time Axelrod formulated the cognitive mapping
approach. The ideas of choice and utility from decision theory were transported by
Axelrod to cognitive mapping, since one of the intended contributions of the cognitive
mapping approach was to shed light on decision-making processes.
These five immediate pre-cursors, acknowledged as such by Axelrod, found their way
to the original formulation of the cognitive mapping approach.
Axelrod’s Seminal Work
In 1976, Axelrod published and edited his book, Structure of Decision: The Cognitive
Maps of Political Elites, which heralded the advent of cognitive mapping in the literature.
In early 1970, while at the University of Berkeley, Axelrod along with his colleagues
Matthew Bonham and Michael Shapiro turned their attention to the study of the beliefs
of elite policy makers. Axelrod’s initial work culminated in the development of a new
approach to decision making based on the idea of a cognitive map of a person’s stated
values and causal beliefs. This approach was presented as a paper at the Conference on
Mathematical Theories of Collective Decisions at the University of Pennsylvania, and
published as a monograph (Axelrod, 1972a). Later, Axelrod used the verbatim transcripts
of the British Eastern Committee to derive the cognitive maps of the committee members
according to the coding rules he had developed, and the resulting analysis was presented
to the Peace Research Society at their London Conference in 1971 (Axelrod, 1972b).
Meanwhile Bonham and Shapiro collaborated to produce a preliminary report on their
work (Shapiro & Bonham, 1973). To quote Axelrod:
“By this time, the project seemed to have a life of its own, as different people
found different uses for cognitive maps.”
Axelrod pulled together the works of several of these people working on cognitive
mapping to produce his classic, Structure of Decision.
Axelrod’s work consisted of five major sections. The first section dealt with an introduction
to cognitive mapping. The second section provided five empirical studies including
Axelrod’s study of the British Eastern Committee and Bonham and Shapiro’s work. The
remaining studies focused on Governor Morris in the Constitutional Convention, the
Energy Crisis, and the politics of the international control of the oceans. The third
section, which consisted of only one chapter, summarizes the conclusions of the
empirical works, with particular emphasis on cognitive maps. The fourth section dealt
with the limitations of the approach, and enumerated several projects for future work. The
final section, the Appendix, contained the coding rules, and approaches to cognitive
mapping including the questionnaire method, mathematics, simulation techniques, and
a guide to source materials.
Axelrod’s work thus provided several methodological ideas that are still with us today.
Key among them are:
1) Definition. “A cognitive map is a specific way of representing a person’s assertions
about some limited domain such as a policy problem. It is designed to capture
the structure of the person’s causal assertions and to generate the consequences
that follow from this structure.”
2) Method of coding. Axelrod provided a detailed system by which a document may
be coded. These rules have served the two following generations of researchers
and will be covered in Chapter II.
3) Sources of data. Various sources of data from documents to interviews to
questionnaires were illustrated by Axelrod.
4) Analysis. Although qualitative analysis is the most commonly used form of
analysis in cognitive mapping, Axelrod presented several—even now infrequently
used—analysis approaches, ranging from statistical analysis to simulations.
In short, the Structure of Decision was vast in its scope and profound in terms of the ideas
it set forth. From the vantage point of this book, Axelrod’s influence on the writings in
organization sciences was immense. It is in this latter regard that I view this work as
seminal. Indeed, almost all the contributors to the evolution of causal mapping owe a
considerable debt to his work.