AMERICAN

INTELLIGENCE
JOURNAL THE MAGAZINE FOR INTELLIGENCE PROFESSIONALS

New Paradigms in Intelligence Analysis

__NMIA__________________________
Vol. 32, No. 1, 2015
 Propositional Diagrams for Intelligence
Sensemaking:
Examples and Case Studies
by Dr. Robert R. Hoffman, Tom Eskridge, Simon Henderson, Jonathan Jenkins, and Brian Moon

INTRODUCTION memory, present information at a glance, and shift some

of the burden of text processing over to the visual

N
ewspaper reports of intelligence-related activities perception system. In a team context, diagrams can support
conducted by the U.S. Intelligence Community have dialogue, help uncover hidden assumptions, facilitate the
occasionally referred to large Pens and Post-Its development of shared understanding, and act as a tool for
wall charts that were created, for example, to understand supporting the communication of meaning and intent.
and represent adversarial networks or the structures of
NGOs. Similar practices have been reported in other Concept Maps are meaningful diagrams composed of
nations intelligence services, such as in the UK. Another labeled nodes (concepts) and relational links. The original
rationale for such diagramming is that collaboration form of meaningful diagram called Concept Mapping was
requires the externalization of understanding and supporting invented in the 1970s by Joseph Novak of Cornell
conversations, to achieve a shared understanding in which University, who was interested in capturing the knowledge
concepts and their meanings are made precisely clear. Team of school children (Novak and Gowin, 1984). Since then, an
creation and analysis of meaningful diagrams encourage, extensive background and substantive research foundation
even force, people to achieve consensus and clarity. When has validated Concept Mapping for a variety of applications
the information that is being shared is critical understanding spanning primary education to professional brainstorming
or intent, upon which lives may depend, there is a clear (Moon, Hoffman, Caas, & Novak, 2011). Concept
imperative that both sender and receiver do everything in Mapping encourages critical thinking (Mintzes, Wandersee
their power to ensure that shared information leads to & Novak, 2000), and results in measurable gains in
shared, and accurate, understanding. knowledge. Building good Concept Maps leads to longer
retention of knowledge and greater ability to apply
Recent guidance on analytical methodology has included knowledge in novel settings (Caas et al., 2003; Mintzes et
recommendations regarding the applications of such al., 2000; Novak, 1991, 1998). Of particular interest in this
meaningful diagrams (Heuer and Pherson, 2011), but article is the use of Concept Maps to express expert
recommendations have not been accompanied by realistic, knowledge and to capture the complex concepts and
detailed examples showing how to make good diagrams and relations involved in analytical problems (Crandall, Klein
how they can be used to best results. This article presents and Hoffman, 2006).1
guidelines on what makes for a good meaningful diagram,
and expresses how meaningful diagrams can support the An example Concept Map is presented in Figure 1. This
process of intelligence sensemaking. diagram represents an attempt to explain a contradiction
about refugee statuswhy the Bhiari refugees do and do
MEANINGFUL DIAGRAMS not qualify as refugees under international law (Faranza,
2008). The icons beneath some of the nodes hyperlink to

R
esearch on diagrammatic reasoning, from fields text pieces and URLs that present supporting evidence. The
spanning geography, statistics, and instructional full set of diagrams that captured the analysis of the Bihari
design, has investigated the value of maps, schematic situation consisted of 15 Concept Maps, including this one.
diagrams, and many other forms of diagram. It has
converged on a set of conclusions concerning the value of
diagramming, and offers an explanation for why
diagramming has value (see Mandl & Levin, 1989; Vekirl,
2002). Diagrams can externalize cognition, guide
reasoning, reduce cognitive demands, support working

Vol 32, No 1 Page 122 American Intelligence Journal

 Concept Mapping supports a number of cognitive processes
that are crucial to critical thinking and fluid intelligence
(Hoffman et al., 2011): Assimilation (changing current
knowledge as a result of the discovery of new knowledge),
differentiation (distinguishing sub-concepts and their
relations), superordination (seeing how previously unrelated
concepts are in fact related), subsumption (seeing how
previously unrelated concepts actually fall under a higher-
order concept), and reconciliation (achieving coherence and
consistency). Concept Maps made by domain experts tend
to show high levels of agreement (see Gordon et al., 1993;
Hoffman, Coffey & Ford, 2000). Reviews of the literature
and detailed discussion of methods for making good
Concept Maps can be found in Caas et al. (2004),
Figure 1. A Concept Map used in the process of sensemaking
Crandall, Klein & Hoffman (2006), and Moon et al. (2011).
about the Bihari refugee situation.
It is important to think of Concept Mapping as a process,
Concept Maps include concepts and relationships among
versus the qualities of finished Concept Maps. Technically
concepts indicated by lines linking two concepts. Linking
stated, when creating a Concept Map, the Mapper uses
phrases specify the relationships. The most important
spatiality (i.e., different areas of the diagram space) as a tool
concepts and the context are generally toward the top and
to de-convolute meanings. As nodes and partially-linked
the more detailed concepts and relations generally toward
sets of nodes are grabbed and moved around in the diagram
the bottom. The diagram morphology is generally
space, the Mapper considers various relationships and ideas
hierarchical (although the cross-links mean that the diagram
to be expressed. The Mapper struggles to add in cross-links
is not a hierarchy in a graph-theoretic sense). In a well-
while avoiding the creation of a spaghetti graph having
formed Concept Map, each node-link-node triple can be
too many overlapping cross-links. Clusters of nodes will be
read as a stand-alone proposition, a feature that renders
parked somewhere, and that region of the Concept Map
Concepts Maps searchable on the basis of more than just a
space becomes, in effect, a memory aid.
bag of words.
In a study conducted with the support of DARPAs Rapid
It should be noted that Concept Maps are not argument
Knowledge Formation project, Concept Maps were made
maps. Concept Maps do not restrict the propositions in
by domain experts but were subsequently tidied up
terms of semantics of argument structure. Therefore, for
overnight by computer scientists. Upon next seeing their
instance, Concept Maps can represent causal relations or
Concept Maps so tidied up, the experts were upset because
temporal structures as well as arguments.
things werent where they were supposed to be (Hayes,
personal communication, 2003). The Mappers had been
With the support of the CmapTools freeware, Concept
using diagram spatiality as a tool.
Mapping is being used around the world, at all levels of
education (Caas, 1999; Caas et al., 2003) and in many
Node-link-node triples essentially make Concept Maps a
locations as part of the core infrastructure in schools and
surface notation for propositional logic. The expression of
entire school systems (Ford, Coffey, Caas, Andrews &
meanings in terms of propositions is central to the
Turner, 1996). Concept Mapping has come to be used in
construction of effective Concept Maps and the meaningful
numerous government and business applications as well
capture of knowledge. CmapTools includes capabilities for
(Moon et al., 2011). Concept Mapping is being used to
extracting and sorting propositions, automated suggestions
create and edit ontologies for intelligent decision aids and
of related concepts through searches on the Web, automatic
use on the semantic Web (Eskridge and Hoffman, 2013).
layout tools, recording and playback of the stages involved
in diagram construction, validation of map coherence, and
ADVANTAGES FOR ANALYTICAL WORK automatic fixing of broken links.

C
onsider the practical advantages of such diagrams Additional advantages of Concept Mapping for analytical
over typical analytical worksheets or data matrices. work stem from the capabilities of the CmapTools freeware.
While spreadsheets or synchronization matrices For example, the ability to hyperlink digital resources
make an analyst record and analyze certain kinds of such as text documents, images, video clips, and website
information in certain ways, they are usually not a useful addresses is a significant capability. Hyperlinks to resources
tool for conveying meaning to othersi.e., what the big are indicated by the small icons underneath concept nodes
picture is, or the so what?

American Intelligence Journal Page 123 Vol 32, No 1

(see Figure 1). Indeed, a Concept Map explaining the
situation under analysis and integrating the resources can be
the analysts report. This represents a potential major gain
in efficiency, perhaps eliminating the need to turn an
analysis into a slide presentation.

CmapTools has a number of capabilities that aid

networking. This includes the ability to search on the
occurrence of a concept to see if the same concept also
appears in other peoples Concept Maps. If it does, then
there exists the ability to pull in another Concept Map to
incorporate within ones own, or to point to the other
Concept Map with a link attached to the concept of interest.
In this manner, knowledge literally becomes networked and
the capability for knowledge reuse is created. CmapTools
contains a range of functions for data import and export that
can assist with better understanding of the data (for
example, concept propositions and hierarchies can be
exported as text).
Figure 2. Concepts in the Thai Protest article.

Example #1: Thinking in Terms of Propositions

The next step involves grouping related concepts spatially.
This helps direct thinking about the relationships that exist
We use this case study to illustrate the process of creating a
among concepts that are related to each other.
Concept Map-based analysis.

Thai protesters build barricades and toy with

talks. BBC News Website, April 21, 2010

Thai anti-government protesters have built

formidable barricades of tyres and sharpened
bamboo canes in Bangkok as tensions build in the
capital. But tentative hints of possible new talks
between protesters and the government have
emerged, as parliament met for the first time in two
weeks. Troops remain behind lines nearby in an
increasingly militarised standoff. The red-shirts are
demanding that Prime Minister Abhisit Vejjajiva
step down and parliament is dissolved. However,
analysts say both sides might feel the need for talks
as the prospect of another bloody crackdown looms.
A failed attempt to clear protesters on 10 April left 25
people dead. [[Link]
pacific/[Link]]
Figure 3. Concepts are grouped.
The first step is to extract the concepts of interest (Figure 2).
(In CmapTools, clicking on the diagram space creates a
The next step is to link the concepts and label the
concept node, ready for typing the concept label.)
relationships (links are easily created in CmapTools by
point and drag, and then the typing of the linking phrase).
The Concept Map should seek to exhibit propositional
coherence. This means that every node-link triple should
make sense when read alone, for example,

[Anti-Government Protesters] <are known as> [Red Shirts]

Vol 32, No 1 Page 124 American Intelligence Journal

Thinking in terms of propositions is a skill that takes some might be improved, our point being that there is not
practice. Learners often fall back on the inferences and necessarily one single best or right way to decompose
concepts that are tacit in ordinary syntax. For example, the and represent open text.
phrase parliament met for the first time in two weeks has the
concept of meeting expressed as a verb, and when going Example #2: Hypothesis Exploration
from text to propositions people sometimes mistakenly
make all the verbs into linking relations. Those who are new As CMappers review and reorganize their thinking, the
to Concept Mapping might craft the phrase protesters have Concept Map undergoes various transformations, revisions,
built formidable barricades of tires as additions, and deletions. The Concept Map can also support
the representation of causality, temporality, uncertainty, and
[protesters] <have built> [formidable barricades] <of> inferenceall features that are critical in analysis. An
[tires] assertion (in contrast with a fact) is a statement or
declaration, often expressed without supporting evidence or
when propositionally the source text reads as: accompanying reasoning. A suggested graphical method for
capturing assertions is to code assertions in the linking
[protesters] <have built> [barricades] phrase using both color and symbology since redundant
[barricades] <are> [formidable barricades] encoding is easier to process. This is illustrated in Figure 5.
[formidable barricades] <are made of> [tires] In this case, we know that it has been asserted that Tribe X
has been accused of political corruption, but we do not
Note also that the expression: know where this assertion has come from, or whether it is
true.
[formidable barricades] <of> [tires]

does not read as a stand-alone proposition.

Another crucial activity is the creation of cross-links;

Concept Maps are not pure hierarchies, but instead
accommodate the complexity and interconnectedness of
ideas and events. A Concept Map representing all of the
propositions in the Thai protest text is presented in Figure 4.
The reader is invited to look for things in Figure 4 that

Figure 4. A completed Concept

Map based on the Thai
protests article.

American Intelligence Journal Page 125 Vol 32, No 1

 BBC News Website, July 13, 2010

An Iranian nuclear scientist at the centre

of an abduction row between the United
States and Iran is free to leave, the US
State Department says.
[[Link]
10617656?utm_source=twitterfeed&utm_m
edium=twitter]

In Figure 7, a number of relationships are inferred and

diagrammed explicitly. For example, while the nuclear
Figure 5. Assertions can be represented using color coding and scientist is said to be at the center of the abduction dispute,
together with front and back question marks. it is not explicitly stated that it was he who was abducted.
However reasonable it might be to infer this, the
In Figure 6, an inference is made that, if there is political relationship is expressed and color-coded as an inference.
corruption present in Tribe X, then there are officials within Notice also the creation of (and notation for) an inferred
the tribe who are themselves corrupt. Note that Figure 6 concept (? Released ?), and also the use of intersecting
also indicates that two linked concepts are inferred from the nested nodes to capture relationships among sets of
premise of political corruption. The inference is not just that concepts. This analysis shows clearly how even simple
Tribe X has officials but that those officials are corrupt. assertions can contain many implications and entailments,
Thus, the proposition is nested, indicating that it is a which get hidden by the syntactic conventions of ordinary
proposition that is inferred. As with the representation of language. It also shows how one proposition can refer to, or
assertions, the use of the double question mark to label the comment on, another proposition.
linking phrase in an inference enables it to be output
collectively as ordered triples for intelligent search.

Figure 6. Inferences can be represented using color coding

along with front- and back-end question marks

Example #3: Inferencing

These techniques will now be exemplified by analysis of the

following text. Although this is comprised of a simple Figure 7. A Propositional-Inferential representation.
sentence, it contains a great deal of information and can be
used to speculate about a wide range of issues. The analysis Taking the analysis a step further can involve adding
involves not just extracting the key information contained in annotations, comments, and questions to make the analysts
the text, but converting the information into a form that is thinking process an explicit part of the representation. In
propositionally coherent and inferring additional concepts Figure 8 a range of questions has emerged, which are
and relationships. recorded as their own supplementary Concept Map.

Vol 32, No 1 Page 126 American Intelligence Journal

 organizational expertise (Ford et al., 1996). In capturing the
expert knowledge within an organization, practitioners can
always add to and modify the Concept Maps in the existing
pool.

One such Knowledge Model is called STORM (System To

Organize Representations in Meteorology) (Hoffman et al.,
2001, 2006). It consists of two dozen Concepts Maps
created by forecasters at the Naval Training Oceanographic
and Meteorology Facility at Pensacola Naval Air Station,
FL. Although the project involved creating many dozens of
Concept Maps about all aspects of weather and weather
forecasting, the knowledge model focuses on weather of
particular concern to naval aviation in the Gulf Coast region
(e.g., turbulence, fog, thunderstorms, and hurricanes). It
covers forecasting processes, such as the use of the radar.
STORM Cmaps can be viewed at
[[Link]
rid=1147120059423_996189320_18181/ROCK-
Figure 8. Questions that emerge from the inferences. TA%[Link]]. Another, and larger, Knowledge
Model called ROCK (Representation of Conceptual
COMMUNICATING THE BIG PICTURE Knowledge) was created for the U.S. Army and focuses on
intelligence preparation of the battlefield (Eccles et al.,

W
hen mapping a large domain or more extensive 2003). It consists of about 200 Concept Maps and has
set of material, a single Concept Map can become hyperlinked topographic maps, aerial photos,
unmanageable for the user to comprehend, photointerpretation keys, and information about
display, and manipulate. To facilitate the construction of trafficability of such terrain features as types of dunes.
large representations, the CmapTools allows the user to Cmaps can be viewed at [[Link]
create and hyperlink collections of Concept Maps, enabling rid=1103739939432_102411597_6499/STORM-LK].
the navigation from one Concept Map to another. Another
capability is that different authors (for example, from An example STORM Concept Map is shown in Figure 9
different functional desks or technical disciplines) can and an example ROCK Cmap (with a screen shot of some
support the generation of integrated collections of Cmaps, open resources) is shown in Figure 10. Note that Cmaps in
each from its own perspective, and show explicitly how Knowledge Models can have many hyperlinks. The
their map relates to those produced by others. Cmap Tools STORM hyperlinks include digital videos of expert
provides the ability collaboratively and synchronously or discussion of weather forecasting procedures and links to
asynchronously to construct a joint Concept Map (see real-time weather and radar data. The ROCK hyperlinks
Caas, Suri, Snchez, Gallo, & Brenes, 2003). stitch the many Cmaps together and link to text pieces
including aerial photos and text about trafficability. The
Hyperlinks can connect Concept Maps to other Concept ROCK Cmap also has, at its left side, a Cmap piece that
Maps; a set of Concept Maps hyperlinked together is shows the place of the particular Cmap within the larger
regarded as a Knowledge Model. The hyperlinking Knowledge Model. Using this piece, and the hyperlinks
permits navigation among the Concept Maps and serves as a within it, the user can always tell where he/she is in the
navigational tool that prevents getting lost in hyperspace. Knowledge Model, and how he/she can get from anywhere
Within the context of a Knowledge Model, an overarching to anywhere in the model, in only one or two mouse clicks.
Concept Map can be created to communicate the big More information about the construction of knowledge
picture. Individual concepts within this Cmap then link to models can be found in Caas, Hill, & Lott (2003);
detailed Cmaps that expand the concept into a series of Crandall, Klein, & Hoffman (2006); and Moon et al. (2011).
lower-level components. Also part of the big picture, any
digital resource can be hyperlinked into a Cmap, bringing in CAUSAL REASONING
the supporting evidence: imagery, reference documents,
video, and websites. Knowledge Models can serve as living

C
repositories of expert knowledge to support knowledge oncept Maps can express causality and temporality.
sharing as well as knowledge preservation. This too Of particular concern in analysis is indeterminate
represents a significant capability for preserving and sharing causation where the goal is to anticipate individual

American Intelligence Journal Page 127 Vol 32, No 1

 Figure 9. A Concept Map from the STORM Knowledge Model about Navy weather forecasting.

Figure 10. A Concept Map from the ROCK Knowledge Model for intelligence preparation of the battlefield.

Vol 32, No 1 Page 128 American Intelligence Journal

or aggregate human activity (Moore and Hoffman, 2011). would be: Low interest rates caused people to purchase
Causal analysis helps explain how the current situation homes they could not really afford, which caused the
came to be, and on this basis supports analysis of where housing bubble, which in turn caused the economic
the situation is likely to go (and the impact that different decline.
actions might have on shaping that path). Research on the
causal reasoning strategies used by practitioners across a We present three somewhat richer examples in the
range of domains, including economic, political, and following figures. Figure 11 presents both a generalized
military, revealed a dozen patterns (Hoffman and Klein, form for the abstractiona template if you willand a
2009). A few of these are: specific example. Figures 12, 13, and 14 illustrate the
onion, spiral, and snark hunt, respectively.
The Abstraction. This is a generalization over
evidence (events or conditions). This causal
attribution takes several causes, sometimes including
counterfactuals, and synthesizes these into a single-
cause explanation.

The Domino. This is a chain or sequence of causes

and effects culminating in the primary effect or
phenomenon that is to be explained.

The Swarm. This is when a number of independent

causes converge to bring about some effect.

The Spiral. Events X and Y were both causal of

Event Z, but Event X increased (or decreased) the
power of Event X.

The Clockwork. This is when one or more causes

have effects that influence other causes, culminating
in the primary effect or phenomenon that is to be
explained.

The Onion. This is when the analyst wonders about

what caused the effect that is used to explain the
primary effect or phenomenon that is to be explained.

The Snark Hunt. The Snark is a mythical animal for

which one can search, but which can never be found Figure 11. A generic form for the Abstraction causal
because it does not really exist. The Snark Hunt is structure, along with an example.
when the explainer is seeking some particular kind
of cause when in fact the to-be-explained effect has
some cause that is hidden or might be unknown. The
Snark Hunt can be considered a form of counterfactual
or disconfirmational reasoning.

All of these themes, and more, can be expressed as a

Concept Map, or as a simple Concept Map Piece that can
be embedded in a larger Concept Map. Awareness of the
different causal structures can support critical thinking, that
is, the search for alternative causes or causal structures. An
example clockwork is from economics: bank deregulation Figure 12. An example of an Onion causal structure, focused
permitted mortgaging that entailed relaxed lending criteria; on the primary phenomenon of the surrender of Poland to
these resulted in risky loans that were used to leverage Germany.
mortgaging. In other words, the key causal factors
interacted. An example chain, also from economics,

American Intelligence Journal Page 129 Vol 32, No 1

 the reader to apply such methods as Analysis of Competing
Hypotheses. The challenge is to decide what is most likely
to happen in the regional conflict, annotated by conflicting
evidence. The realism of the scenario is highlighted by the
fact that there is no single, clear best answer, and analyses
of the scenario by experienced analysts not only do not
always agree, but result in hypotheses and findings that
were not anticipated by the scenarios creator (Moore,
personal communication).

An attempt was made to represent the scenario text

Figure 13. An example Spiral causal structure involving the exhaustively in propositionally-coherent Concept Maps.
Israeli-Palestine relation. The first result was 14 Concept Maps, each representing one
of the major paragraphs of the scenario (four Cmaps for
each of the four main paragraphs in the Background,
three Cmaps for scenario sections that each focused on one
of the hypothetical nations, a Cmap for a paragraph about
mysterious events occurring on an island in the region, a
Cmap covering the discussion of the regional religions and
historical religious conflicts, a Cmap about recent news
reports, a Cmap about events at a particular shrine, and a
Cmap about regional wars). As these were created, separate
diagrams were made expressing the hypotheses and
speculations that occurred during the analysis (in the
manner of Figure 8, above, but not unlike the process
involved in the Analysis of Competing Hypotheses). Next,
versions of all of the Cmaps were created that expressed
assertions and hypotheses (in the manner of Figures 5-7,
above). All this took timethe better part of three days of
full-time effort.

The Concept Map about one of the nations is presented in

Figure 15 (next page). This diagram is representative of the
degree of detail one would expect in a useful Concept Map
in analytical work. Our heuristic is that for the clearest
presentation of meaning a good Cmap should have no more
than about 40 concepts (or about 45 propositions). As a
Figure 14. A template plus an example for the Snark Hunt Cmap gets larger than this, it is appropriate to break it up
causal structure. into meaningfully appropriate smaller Cmaps that are then
hyperlinked together.
Our final case study brings many of these ideas together,
and also hints at the richness and complexity that can be One of the Concept Maps about the analysts hypotheses is
involved in conducting analytical work using meaningful presented in Figure 16. A majority of other analysts who
diagrams. had attempted the scenario (up to the time that the Cmap-
based analysis was conducted) had attempted conclusions
Case Study: The Klathu Scenario about whether or when a war would break out (Moore,
personal communication). The conclusion for this Cmap-
We illustrate the Cmap-based analysis method using the based analysis was that a war had already broken out.3
Klathu Scenario developed by David Moore (Moore,
2010). While hypothetical, it is a detailed, rich, and realistic
scenario about a regional conflict.2 The text is 13 pages,
covering history and background, current situation, assets,
recent events, and culminating events as reported by open
sources and intelligence sources. The scenario includes
maps and listings of evidence, followed by an invitation to

Vol 32, No 1 Page 130 American Intelligence Journal

 Figure 15. A Concept Map expressing the evidence, assertions, and inferences about one of the nations in the Klathu Scenario.

focused primarily on Concept Mapping we note also that

meaningful diagrams of other kinds can be useful in
analytical work.

REPRESENTING ANALYTICAL INTENT

T
he Analysts Intent diagram is designed to help
practitioners set out, at the beginning of problem
analysis, their intended strategy and line of attack.
This can serve a number of purposes. It can enable analysts
to return to their foundation during time of information
overload or distraction, in order that they can help keep
themselves on track. It can help other analysts understand
how the problem is being tackled (for example, if the
problem was handed over to somebody else midway
through). It can also serve as a training device to help
communicate tough cases and exemplars to less experienced
analysts. By filling in the blanks in the template for
Analysts Intent, the analysts address a range of
Figure 16. A Concept Map used to express some of the considerations and captures their resultant thought process.
questions and hypotheses formulated by the analyst. When drafted at the beginning of an analytical activity, the
Intent Diagram helps frame and contextualize the work
process, but it can be iteratively refined as the analytical
It is not our contention that Cmap-based analysis will work progresses. It can be used to frame reports on, and
uniquely support the creation of high-value analytical other products associated with, the activity.
products or results. We only suggest, as do Heuer and
Pherson (2010), that analysis using meaningful diagrams
has its appropriate and valuable uses. While we have

American Intelligence Journal Page 131 Vol 32, No 1

 Figure 17. A template for a meaningful diagram to express Analysts Intent.

Figure 18. An instance of a completed Analysts Intent diagram.

A template Analysts Intent diagram is presented in Figure software-assisted knowledge, brainstorming, knowledge
17 (note that this is not a Concept Map though it is easily sharing (getting data points and information for others who
created using Cmap Tools). In using this template, the nodes view Cmaps), contrasting alternative perspectives, training,
are filled by phrases or short sentences and then meaningful identifying knowledge gaps, creating new knowledge (for
relationships among nodes are added, as deemed useful, example, turning tacit knowledge into an organizational
using linking phrases to express interrelationships, resource), representing team knowledge, constructing and
contingencies, or dependencies. An example of a completed representing shared understanding, structuring conceptual
Analysts Intent diagram is presented in Figure 18 queries, expressing and comparing methodologies,
(descriptions, templates, and examples of other causal structuring linguistic definitions, designing competency
explanation structures can be provided upon request). questions, representing networks and organizations, and
decomposing analytical problems.
CONCLUSION
Thus, we have seen a great diversity of diagrams, posted on

T
he methodology of analysis includes techniques of walls and workboards/workplaces such as command posts
diagramming. Uses of meaningful diagrams within and analyst cubicles. Diagrams range from the well-
the Intelligence Community include: eliciting composed and formatted, to nearly useless spaghetti
knowledge from experts analysis, designing new graphs. There has been little discussion of what makes for a
technology by domain experts (bridging the gap between good diagram, and why. Thus, in this article we have
analysis and systems engineering requirements and needs presented some principles or heuristics for the creation of
statements), revealing expert-novice differences, acquiring meaningful diagrams used in problem decomposition, based

Vol 32, No 1 Page 132 American Intelligence Journal

on findings from research on diagrammatic reasoning in Caas, A.J., Hill, G., Carff, R., Suri, N., Lott, J., Eskridge, T.,
psychology and other disciplines. We have illustrated a Gomez, G., Arroyo, M., & Carvajal, R. (2004).
variety of forms of meaningful diagrams that have been and CmapTools: A knowledge modeling and sharing
are being used in analytical work, accompanied by environment. In A.J. Caas, J.D. Novak & F.M.
Gonzalez (eds.). Concept Maps: Theory, Methodology,
templates and examples (for a discussion of the use of
Technology, Proceedings of the 1st International
meaningful diagrams in structuring conceptual queries and Conference on Concept Mapping, Pamplona, Spain:
forming ontologies, see Eskridge and Hoffman, 2012). Universidad Publica de Navarra, pp. 125133.
Caas, A.J., Leake, D., & Wilson, D.C. (1999). Managing,
One of the most potentially valuable and effective uses of Mapping and Manipulating Conceptual Knowledge:
meaningful diagrams may stem from the fact that the Exploring the Synergies of Knowledge Management and
diagram supports the analytical activity and at the same time Case-Based Reasoning. AAAI Workshop Technical
can serve as a key part of the analysts report. It is widely Report WS-99-10. Menlo Park, CA: AAAI Press.
known that report preparation can be a huge drain on an Caas, A., Hill, G., & Lott, J. (2003). Support for Constructing
Knowledge Models in CmapTools. Technical Report
analysts time, and that reporting is a major bottleneck. For
IHMC CmapTools 93-02. Available at:
an example of using Concept Maps to summarize material, [Link]
see the synoptic diagrams for the chapters in Hoffman and for Constructing Knowledge Models in [Link]
Militello (2006). Last accessed 20/12/2010.
Caas, A., Novak, J.D., & Gonzlez, F.M. (eds.) (2004).
[Authors Note: Portions of this material were created with Proceedings of the First International Conference on
the support of an SBIR to Perigean Technologies, Inc. Concept Mapping: Concept Maps: Theory,
(Contract No. W31P4Q-08-C-0229) in the Rapid and Methodology, Technology. Navarra, Spain: Direccin de
Accurate Idea Transfer SBIR Program sponsored by the Publicaciones de la Universidad Pblica de Navarra.
Caas, A., Suri, N., Snchez, C., Gallo, J., & Brenes, S. (2003).
Defense Advanced Research Projects Agency, Washington,
Synchronous Collaboration in CmapTools. Technical
DC. Additional material in this article has been adapted Report IHMC CmapTools. Available at:
from Henderson, S., Hoffman, R.R., and Jenkins, J. (2011). [Link]
A Guide to Influence Mapping: Analysing and Synchronous Collaboration in Cmap [Link]. Last
Representing Influence Activity over Time. Report accessed 20/12/2010.
11.00106/1.0, QinetiQ, Farnborough, UK.] Crandall, B., Klein, G., & Hoffman R.R. (2006). Working Minds:
A Practitioners Guide to Cognitive Task Analysis.
Notes Cambridge, MA: MIT Press.
1
Other diagramming schemes have been created. Some of these Eccles, D.W., Feltovich, P.J., and Hoffman, R.R. (2003, April).
borrow Novaks ideas. Most are limited, for example, by not using Modeling Terrain Analysis Expertise to Support
labelled links or not having a principled layout or morphology. Command and Control in the Field. In Proceedings of
2
Readers may benefit by attempting to work the Klathu Scenario the 2003 Collaborative Technology Alliances
on their own, using their preferred method, before reading about Symposium. Sponsored by the Army Research
the Cmap-based analysis. Laboratory and held at the University of Maryland,
3
While our figures offer some hints as to how this conclusion was College Park, MD.
reached, we do not present the full details, by way of inviting Edwards, J. & Fraser, K. (1983). Concept Maps as reflectors of
conceptual understanding. Research in Science
readers to conduct their own analysis.
Education 13, 19-26.
Farzana, K.F. (2008). The neglected stateless Bihari community in
References Bangladesh: Victims of political and diplomatic
Ausubel, D.P. (1960). The use of advance organizers in the onslaught. Journal of Humanities & Social Sciences
learning and retention of meaningful verbal material. 2(1), 1-19.
Journal of Educational Psychology 51, 267-272. Ford, K.M. & Bradshaw, J.M. (1993) (eds.). Knowledge
Ausubel, D.P. (1968). Educational Psychology: A Cognitive View. Acquisition as Modeling. New York: John Wiley &
New York: Holt, Rinehart and Winston. Sons.
Ausubel, D.P., Novak, J.D., & Hanesian, H. (1978). Educational Ford, K.M., Coffey, J.W., Caas, A.J., Turner, C.W., & Andrews,
Psychology: A Cognitive View (2nd ed.). New York: E.J. (1996). Diagnosis and explanation by a Nuclear
Holt, Rinehart and Winston. Cardiology Expert System. International Journal of
Bauer, M.I. & Johnson-Laird, P. N. (1993). How diagrams can Expert Systems 9(4), 499-506.
improve reasoning. Psychological Science 4, 372-378. Ford, K.M., Coffey, J.W., Caas, A., Andrews, E.J., & Turner,
Caas, A.J., Coffey, J.W., Carnot, M.J., Feltovich, P., Hoffman, C.W. (1996). Diagnosis and explanation by a nuclear
R., Feltovich, J., & Novak, J.D. (2003). A Summary of cardiology expert system. International Journal of
Literature Pertaining to the Use of Concept Mapping Expert Systems 9, 499-506.
Techniques and Technologies for Education and Glasgow, J., Narayanan, N.H., & Chandrasekaran, B. (eds.)
Performance Support. Report to the Chief of Naval (1995). Diagrammatic reasoning: Cognitive and
Education and Training, prepared by the Institute for
Human and Machine Cognition, Pensacola, FL.

American Intelligence Journal Page 133 Vol 32, No 1

 computational perspectives. Cambridge, MA: MIT Moon, B., Hoffman, R., Novak, J., & Caas, A. (2011). Applied
Press. concept mapping: Capturing, analyzing, and organizing
Gordon, S.E. & Gill, R.T. (1997). Cognitive task analysis. In C. knowledge. New York: CRC Press.
Zsambok & G. Klein (eds.), Naturalistic Decision Moore, D.T. (2010). What is going on in Klathu? NSA
Making, Mahwah, NJ: Erlbaum, pp. 131-140. APPROVAL NUMBER: PP-08-135.
Gordon, S.E., Schmierer, K.A., & Gill, R.T. (1993). Conceptual Moore, D.T. and Hoffman, R.R. (2011). A practice of
graph analysis: Knowledge acquisition for instructional understanding. Chapter 5 In D.T. Moore, Sensemaking:
system design. Human Factors 35, 459-481. A structure for an intelligence revolution. Washington,
Hackett-Fisher, D. (1970). Historians Fallacies: Towards a Logic DC: National Defence Intelligence College Press.
of Historical Thought. New York: Harper Torchbooks, Novak, J.D. (1991). Clarify with concept maps: A tool for
pp. 164-186. students and teachers alike. The Science Teacher 58, 45-
Henderson, S., Hoffman, R., & Haines, P. (2010). Influence 49.
Support Concepts: Simple Tools for Complex Problems. Novak, J.D. (1998). Learning, Creating, and Using Knowledge:
QinetiQ Report 10/00788/1.0, QinetiQ, Farnborough, Concept maps as Facilitative Tools in Schools and
UK. Corporations. Mahwah, NJ: Lawrence Erlbaum
Henderson, S., Hoffman, R.R., and Jenkins, J. (2011). A Guide to Associates.
Influence Mapping: Analysing and Representing Novak, J.D. & Caas, A. (2008). The Theory Underlying Concept
Influence Activity Over Time. QinetiQ Report Maps and How to Construct and Use Them. Technical
11.00106/1.0, QinetiQ, Farnborough, UK. Report IHMC CmapTools 2006-01 Rev 2008-01.
Heuer, R.J. & Pherson, R.H. (2010). Structured analytic Available at [Link]
techniques for intelligence analysis. Washington, DC: ResearchPapers/TheoryCmaps/
CQ Press. [Link] Last accessed 3/14/
Hoffman, R.R. (ed.) (2007). Expertise out of context: Proceedings 2010.
of the Sixth International Conference on Naturalistic Novak, J.D. & Gowin, D.B. (1984). Learning How to Learn. New
Decision Making. Boca Raton, FL: Taylor and Francis. York: Cambridge University Press.
Hoffman, R.R., Coffey, J.W., & Ford, K.M. (2000). A Case Study Vekirl, I. (2002). What is the value of graphical displays in
in the Research Paradigm of Human-Centered learning? Educational Psychology Review 14, 261-298.
Computing: Local Expertise in Weather Forecasting.
Report to the National Technology Alliance on the
Contract, Human-Centered System Prototype. Institute Dr. Robert R. Hoffman is a recognized world leader in
for Human and Machine Cognition, Pensacola, FL. cognitive systems engineering and human-centered
Hoffman, R.R., Coffey, J.W., Ford, K.M., and Carnot, M.J. (2001, computing. He is a Fellow of the Association for
October). STORM-LK: A human-centered knowledge
Psychological Science, a Fellow of the Human Factors and
model for weather forecasting. In Proceedings of the
45th Annual Meeting of the Human Factors and Ergonomics Society, and a Fulbright Scholar. Hoffman has
Ergonomics Society (p. 752). Santa Monica, CA: HFES. been recognized in disciplines including psychology, remote
Hoffman, R.R., Coffey, J.W., Ford, K.M., and Novak, J.D. sensing, weather forecasting, and artificial intelligence for
(2006). A method for eliciting, preserving, and sharing his research on human factors in remote sensing, the
the knowledge of forecasters. Weather and Forecasting psychology of expertise, the methodology of cognitive task
21, 416-428. analysis, and the design of complex cognitive work systems.
Hoffman, R.R., Henderson, S., Moon, B., Moore, D.T., and His current research focuses on methodological issues in
Litman, J.A. (2011). Reasoning difficulty in analytical the analysis of complex systems and performance
activity. Theoretical Issues in Ergonomic Science 12,
measurement for complex cognitive work systems.
225-240.
Hoffman, R.R. and Markman, A.B. (eds.) (2001). The
interpretation of remote sensing imagery: The human Simon M. Henderson is a Lecturer in Information Design at
factor. Boca Raton, FL: Lewis Publishers. Cranfield University, based at the Defence Academy of the
Klein, G. & Hoffman, R. (2009, June). Causal Reasoning: Initial United Kingdom, where he has been since 2011. Earlier he
report of a naturalistic study of causal inferences. was a researcher at QinetiQ (and its forerunner
Proceedings of NDM9, the 9th International Conference organizations) focusing on the psychological basis of
on Naturalistic Decision Making, London, UK. sensemaking, decision-making, and teamwork. He began
Mandl, H. & Levin, J.R. (eds.) (1989). Knowledge acquisition his career in the areas of knowledge elicitation and system
from text and pictures. New York: Elsevier.
dynamics modeling, developing exploratory models of
Markham, K., Mintzes, J., & Jones, G. (1994). The concept map
as a research and evaluation tool: Further evidence of future command information systems. He then joined a
validity. Journal of Research in Science Teaching 31(1), group studying the impact of technology on command and
91-101. intelligence teamwork and decision- making. His
Mintzes, J.J., Wandersee, J.H., & Novak, J.D. (2000). Assessing contributions to methodology include the development of a
Science Understanding: A Human Constructivist View. range of data collection and knowledge elicitation tools and
San Diego, CA: Academic Press. techniques; methods for the experimental and field-based
assessment of teamwork, sensemaking, and naturalistic

Vol 32, No 1 Page 134 American Intelligence Journal

decision-making; methods for the enhancement of team is a visiting speaker at the NATO Information Operations
debriefing and self-learning approaches; and methods for Senior Officers Course.
the design and evaluation of team support technologies.
Work over the past five years has focused on studying the Brian Moon is Chief Technology Officer for Perigean
psychological basis of collective problem solving in complex Technologies LLC. A leading practitioner and researcher in
settings, with particular emphasis on the development of the field of expertise management, he applies diagrammatic
knowledge mapping and representational systems. This and naturalistic interviewing techniques across a spectrum
work has been widely published. Simon has been the UK of knowledge management problems, from the elicitation
representative on a number of NATO and other and capture of expert knowledge to the development of
international panels on decision making and teamwork. decision support and training systems that enable its
effective and efficient transfer. He has been widely
Jonathan A. Jenkins is a Director at i to i research. He leads recognized for his innovative applications of Concept
the organization in its application of cutting-edge human Mapping, which are highlighted in the edited volume,
science to address theoretical and practical challenges Applied Concept Mapping: Capturing, Analyzing, and
across the domains of influence, intelligence, and Organizing Knowledge. His clients have included several
counterterrorism. His passion and expertise lie in the private corporations, the Department of Defense, the New
practical science of behavior change within groups, York Power Authority, the Federal Bureau of Investigation,
organizations, and populations, with his work focusing on and Sandia National Laboratories. Brian is also an adjunct
practical methods, techniques, and approaches. Having faculty member for Kent State Universitys Information
originally studied psychology, Jon has spent the last ten Architecture and Knowledge Management, an online MS
years focused on the areas of behavioral research, behavior program.
change tools and techniques, and measurement for influence
campaigns. He is the author of guidance for change
practitioners and speaker on behavior change and influence
to practitioners across the defense and security domain, and

American Intelligence Journal Page 135 Vol 32, No 1