Early Precursors

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
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


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.