Table of Contents
Go Back
Perception and handling of complex problems in dynamic settings
— three cases of relevance to military command and crisis
management
Bjorn T Bakken
Norwegian Defence Leadership Institute (FIL)
Norwegian Defence Academy (FSS)
Oslo mil/Akershus
NO-0015 OSLO, Norway
Tel +47 23 09 73 95
Fax +47 23 09 37 71
btbakken@fil.mil.no
Martin Gilljam
Division for Analysis
Norwegian Defence Research Establishment
PO Box 25
NO-2027 KJELLER, Norway
Tel +47 63 80 77 56
jmg@ffi.no
Thorvald Haerem
Department of Leadership and Organisational management
Norwegian School of Management BI
PO Box 580
NO-1302 SANDVIKA, Norway
Tel +47 67 55 73 17
thorvald.haerem@bi.no
Abstract
Command of military operations requires leaders and teams who are able to make decisions
and respond in an appropriate, timely manner even in highly uncertain situations. The degree
of situational uncertainty has continued to increase as military requirements have evolved in
response to changing conditions around the world and advances in information systems that
have made more information available. To provide a basis for better understanding of the
dynamics of operations, we suggest an approach linking Effects Based Operations (EBO)
modelling and analysis with System Dynamics (SD) principles, yielding valuable insights in
the complexity of modern warfare and conflict resolution. To illustrate our approach, we
present the following dynamic cases: 1) operational logistics, 2) strategic investments, and 3)
crisis management.
Introduction
Command of military operations requires leaders and teams who are able to make decisions
and respond in an appropriate, timely manner even in highly uncertain situations. The degree
of situational uncertainty has continued to increase as military requirements have evolved in
response to changing conditions around the world and advances in information systems that
have made more information available, faster than ever before (NATO, 2003).
To respond to the terrorist threat, for example, military forces must “anticipate the unexpected
and be prepared for the unimaginable” (NATO, 2003). The people and organizations of the
military must be adaptable and technology must enable their adaptability. Adaptability is both
a proactive and reactive process and can be seen in what military forces do and how they
operate.
Information systems are a critical component of adaptable performance — especially of
distributed, decentralized, yet highly interdependent military forces required to deftly
transition between modes of operation and levels of intensity (e.g., from high-intensity war-
fighting to low intensity peacekeeping, or vice versa) in collaboration with joint and coalition
forces.
As Horn (2003) makes clear: “Commanders will be required to operate in, and be comfortable
with, ambiguous and uncertain surroundings. Their options for using force will often be
restricted. In addition, of necessity, they will require the capability of adapting physically and
theoretically to changes in the immediate operational area as well as in the larger international
security environment. These sorts of uncertain situations will also demand that individuals,
units and formations be agile, flexible and capable of responding to the unforeseen and
unexpected.” The future conflict will first and foremost be of an asymmetric! nature.
Network Centric Warfare (NCW) and Effects Based Operations (EBO)
The development of concepts such as Network Centric Warfare (NCW) and Effects Based
Operations (EBO) are attempts to structure, understand and take advantage of the
opportunities that are now exposed in the transformation processes involving technology,
organisation, doctrine, training, leadership and culture. NCW essentially encompasses the
idea of interconnecting a heterogeneous range of actors and objects (typically called effectors,
sensors, decision makers etc) in the battle-space through telecommunication and computer
networks. The advantages of NCW are (Diesen, 2004):
e¢ The most ideally located decision maker may engage any target within range of the
networked sensors, and may employ the best-suited effector for the engagement mission
(self-synchronisation).
e¢ Multi-role objects allow interchange of information and flexible usage of all resources
(multi-functionality).
e All levels of the command organisation may receive and utilise the available information,
at no incremental cost (information “overlaying”).
The above benefits not only provide solutions and opportunities, but also pose central
dilemmas. As an example, consider the self-synchronisation and overlaying aspects of NCW.
In effect, these will lead to a “collapse” of the traditional command hierarchy (tactical,
operational and strategic levels). When all actors receive all information, there is a high risk
of information overload, let alone potential confusion when it comes to who is going to
' An asymmetric threat is a concept used to describe attempts to circumvent or undermine an opponent’s
strengths while exploiting his weaknesses, and using methods that differ significantly from the opponent’s usual
mode of operation.
implement what decision. Will it be the strategic command or rather the individual solider in
the battlefield (the paradox of the “strategic corporal”)?
EBO may in many ways be regarded as the defence analyst’s attempt to turn the multi-faceted
implications of NCW into analysable and well-structured problems, in that it provides a
modified and fruitful perspective on operational dynamics. That is, it offers insight into the
“driving forces” governing modern warfare and conflict resolution. Davis (2001) defines EBO
as “operations conceived and planned in a systems framework [emphasis made by this author]
that considers the full range of direct, indirect, and cascading effects—effects that may, with
different degrees of probability, be achieved by the application of military, diplomatic,
psychological, and economic instruments.”
Instead of the established force structure and procedures providing a more or less given effect
in relatively well-known settings (platform centric warfare), the problem is reformulated to
being one of determining what combination of assets may yield a desired effect (network
centric warfare). The advantage to the latter approach is that it is more appropriate for
handling a wider range (even a priori unknown) circumstances.
As pointed out by Davis (2001), the main challenge is that current methods of analysis and
modelling appear to be inadequate for representing EBO. Davis proposes the following
principles as a guide for defining and conducting analyses, in order to take the “broad view”
on defence (force and operations) planning:
e Defence planning should focus on mission-system capability, which refers to the ability to
accomplish missions under a wide range of operational circumstances.
e Analysis dealing with EBO should fully confront the scope and magnitude of uncertainty
and should deal explicitly with probability and randomness. E.g., assessments of
capability should refer specifically to most-likely, best-case, and worst-case outcomes.
e Dealing with uncertainty will require low-resolution exploratory analysis for breadth, as
well as more-detailed modelling and gaming for both depth and insight into underlying
phenomena. This suggests a “family-of-models-and-games” approach in which
information obtained from different members of the “family” is used to inform and cross-
calibrate the whole body of knowledge.
e A key element of analytical work should be qualitative modelling, including cognitive
modelling of the decision making and behaviour of commanders, political leaders, and
even societies. Such modelling should be undertaken in an “uncertainty-sensitive”
framework.
¢ Much of the related modelling should be organized around adaptive systems for command
and control and other matters, rather than around the mass and physical characteristics of
forces (focus on decisions and behaviours of people and organizations. This implies
emphasis on the concepts and technology of agent-based modelling, as well as on system
engineering.
e Analysts should vigorously pursue a new base of empirical information—including
information obtainable from history and from a combination of gaming, man-in-the-loop
simulation, and experiments in battle laboratories and the field. This information should
be collected and framed in ways that illuminate complex and subtle relationships and that
support uncertainty analysis.
Cognitive models of operations = Command concepts
Essential to the successful conduct of a military operation is that the commander, his/her
supporting staff, and subordinate units together have a “shared” mental conception of the
tasks to be accomplished, the goal(s) to be reached, and the methods to employ. Builder et al
(1999) have developed a concept called “Command Concepts”, which in essence constitute
the commander’s mental visualisation of the course of the operation, including desired end
state(s). At the most general (cognitive) level, it is required that the commander has the ability
to create, disseminate and update/maintain a “command concept” for the planned operation.
Closely linked to command concepts is the Situational Awareness (SA). Endsley (1988)
defines SA like this: “Situational awareness is the perception of the elements in the
environment within a volume of time and space, the comprehension of their meaning, and the
projection of their status in the near future”. The SA concept is therefore essential to
accomplish rapid and “correct” decision making in operations. According to Brewster (2002),
many studies within military and scientific communities conclude that (military) commanders
actually rely more heavily on an intuitive versus an analytic approach when in a field
environment. The intuitive approach to decision making appears to be chosen when facing:
ill-structured problems; uncertain or dynamic environments; time stress; and/or high stakes.
The intuitive approach is based on pattern recognition and experience, and goes within the
military profession under terms such as “fingerspitzengefuhl” and “coup d’oeil”.
The contribution from System Dynamics (SD)
A military operation or campaign presents the commander with a dynamic decision problem.
During the campaign, the commander continuously receives status information, and on the
basis of this information he will manage his resources (by issuing directives to his staff and
subordinate commanders). As the directives are acted upon, subordinates will report back on
outcome and new status (updated situation assessment). This cyclic procedure will repeat for
the duration of the campaign.
As a practical example to illustrate, consider this general description of crisis management,
taken from the NATO/PfP “Generic Crisis Management Handbook” (1997 interim version):
“Procedures and activities in crisis management range across; information acquisition and
assessment; the analysis of the situation; the establishment of goals to be achieved; the
development of options for actions and their comparison; the implementation of chosen
options, to (finally, as feedback to close the loop) the analysis of the reaction of the parties
involved.”
According to Brehmer and Allard (1991), a dynamic decision problem has the following
characteristics:
e It requires a series of decisions
e The decisions are interdependent.
e The state of the problem changes, both autonomously, and as a consequence of the
decision maker’s actions
¢ Decisions must be made in real time
Following Brehmer (2000, 2002) this kind of task can be seen as a problem of finding a way
to use one process to control another process, and it is the relation between the processes that
determines what strategies are possible. It also means that the time scales of the task are
important. The decision maker thus has two essential tasks:
e To control the system
e To control his workload and avoid being overwhelmed by the task
That a series of interdependent decisions is required means that the decision maker must learn
to allocate (in time and space) his resources in a way that provides for the best possible
overall outcome. That the task changes both autonomously and as a consequence of the
decision maker’s action means that he must be able to distinguish between the effects of his
own actions from autonomous effects.
That the decisions must be made in real time means that the decision maker must be able to
find a strategy that protects him from being overwhelmed (according to Brehmer (2002)
expert decision makers are recognized by their capability to adapt their work strategies to
control their work load); he must learn not only about the task, but also about his own
abilities. He must understand how to use a process for controlling another process, find ways
to handle the delays, and possess an understanding of the time scales.
Intuition vs. analysis?
Intuition and analysis represent the end points of the “cognitive continuum” (Dunwoody et al
2000), but the cognitive mode is rarely purely intuitive or analytical. More often, it is a
mixture of both. This is referred to as “quasi-rationality” (Brunswik 1956, Hammond 1996).
This “middle course” is characterized by a robust and adaptive decision making/problem
solving process, and is closely associated with “common sense”. Tasks that contain
uncertainty, dynamics and many (redundant) perceptual cues (thereby making the task hard to
analyse) will benefit from a largely intuitive approach (Dunwoody et al 2000). Intuitive
decision making is quick (almost instantaneous), and happens with low cognitive control and
low conscious awareness. Perhaps paradoxically, good intuition is not commonplace, and in
constant need of improvement!
That people have problems when applying common sense (or intuition) to static situations
involving simple probability judgment is well known (see Kagel and Roth (1995), for a
comprehensive review). Several authors now point to decision makers’ failure to consider
feedback in complex, dynamic systems. Let two recent studies illustrate the magnitude of this
problem: In his studies of management of renewable resources, Moxnes (1998) observes that
experienced decision makers over-invest and over-utilize their resources. He attributes this
behavior to systematic misperceptions of stocks and flows, and of non-linearities. Sweeny and
Sterman (2000) took a different approach when they gave system dynamics case problems to
students at an elite business school. The students, who were highly educated in mathematics
and science (but had received no prior schooling in system dynamics concepts), were found to
have a poor level of understanding of the basic system dynamics concepts: stock and flow
relationships, and time delays.
Inspired by the Sweeny and Sterman (2000) study, Ossimitz (2002) has conducted an
investigation where 154 subjects were given different tasks in dynamic thinking in general,
and in the interpretation of stock-flow related graphs in particular. The results were alarming,
in that the mean performance of the subjects was approximately at the level of tossing a coin
for each answer. He suspects that the lacking ability is to grasp that a positive net-flow results
in an increase in the corresponding stock. In a related study, Kainz and Ossimitz (2002) finds
that a 90-minute crash course introducing basic stock-flow concepts between pre- and posttest
was suitable to bring about an improvement in performance.
Jensen and Brehmer (2003) let a collection of laypeople attempt to establish equilibrium in a
simple predator-and-prey simulation. They find that even though the task was structurally
simple, it was still perceived as difficult. They conclude that subjects may have a low ability
to apply indirect reasoning, at the same time as they resort to thinking in terms of discrete
time steps rather than in terms of continuous time.
It is thus a general observation that people perform quite poorly in systems with even modest
levels of complexity. Sterman (2000) labels this kind of cognitive dysfunction
“misperceptions of feedback”. The solution would be to develop “systems thinking” abilities.
Tversky & Kahneman (1987) recognized early the shortcomings of a static, one-shot approach
to learning. They describe the prospects for learning in dynamic environments like this:
“Effective learning takes place only under certain conditions: it requires accurate and
immediate feedback about the relation between the situational conditions and the appropriate
response. The necessary feedback is often lacking for the decisions faced by managers,
entrepreneurs, and politicians because:
Outcomes are commonly delayed and not attributable to a particular action
Variability in the environment degrades the reliability of the feedback, especially where
outcomes of low probability are involved
e There is often no information about what the outcome would have been if another
decision had been taken
¢ Most important decisions are unique and therefore provide little opportunity for learning
(see Einhorn and Hogarth 1978)”
Though decision-making errors may disappear with guided experience and reflection, little
research has focused on the improvement in decision strategies — in particular, changes in
decision making with experience have not been revealed to any extent. Notable exceptions
exist, however. Bakken (1989) finds that in a simulated economy, subject’s performance
improves over trials. Paich & Sterman (1993), in their investigation of human performance in
a product lifecycle task, finds that performance suffers as dynamic complexity increases.
Brehmer (1988) finds that action lags decrease performance in a simulated forest fire, and at
the same time conditions for learning get worse.
Three dynamic cases
The aim of presenting the following cases is to demonstrate how structurally simple,
deterministic tasks may induce complex dynamic behaviour, and that this behaviour is not
fully perceived even by fairly competent decision makers. The importance and generality of
the tasks (within a defence organization) makes strong cases for improvement, for example
through simulation based training.
Case 1 — Operational logistics
The following case employs a model simulating a humanitarian task, where decision makers
were either in command of an operational supply chain; a supporting supply chain; or had to
control both in combination. Practice on the simulation model was organized at two levels of
complexity: Practice on each of the single supply lines in turn, before embarking on the
combined task (simplified approach); and as practice on the combined task from the start
(full-scale approach). Similarly the cover stories existed in two versions — a very brief one —
and a more verbose one (more than twice the number of words compared to the former).
CONTROL PANEL || | A
Order number of vaccines
Zz —-
Call up number of personnel
conn to te sec a 20
Next day |
——— 4 a
— RNB
At
RESULT: [_ #3 1 ‘. eee f \
3ae oph 20h ope 480M wr 0404
Diagram 1: User interface of the simulation model
When practicing, the decision makers were instructed to complete as many trials as they could
manage within a limited time — 100 minutes. The time limit was equal across all treatments.
The effect of practicing was measured as performance on a subsequent evaluation task —
similar to the full-scale task with verbose cover story, but with slightly adjusted contextual
information. The best participants would get monetary rewards.
In addition to the fundamental problem of matching supply with capacity, good decisions in
the simulation model are characterized by three capabilities:
e “Ordering ahead”, taking into account transportation lags in the supply chains
e “Cut-off” in ordering — when capacity utilization approaches its maximum
e Near perfect balance between the two supply chains
The problem of matching supply with capacity is present in sub-models (a sub-model
represents either one of the standalone supply chains) as well as in the full model, except on a
smaller scale. Moving from sub-models to the full model requires merely recognition of the
fact that the two supply chains that previously were separated now have to move in
synchronization, and that different time lags between the chains make this a non-trivial effort.
The experimental results indicate that learning does occur from training, and that the number
of practice trials explains performance, not the treatments as such (Bakken and Gilljam 2003).
A key training issue then becomes how such increased training frequency can be obtained in
practice. The experiment indicates that a simpler cover story, and a simpler simulation model
will not in itself be sufficient. Our results suggest that participants may have employed a
certain “style” or “rhythm” when making decisions in the experimental environment. If the
cognitive complexity is decreased through a simpler cover story or a decomposed model
structure, participants probably to some extent compensate by creating and reflecting on
additional hypotheses about cause and effect relationships. The continued research attention
within training improvement should therefore be a divided one: on how to facilitate and guide
participants’ mental processes, as well as on techniques and technology for environment
compression.
Case 2 — Strategic investments
The following case is an extremely simplified model of the logistics chain in a “Peace Support
Operation”, and focuses on the dynamic interrelations between budget allocations and cost of
operations (initial purchases and running sustenance), and the short- and long-term
consequences of a limited time budget expansion. The paradox encountered is that a one-time
budget increase may lead not only to a performance boost in the short term (primary effect),
but also to a significant penalty in the longer term (unintended side-effect). The primary cause
of this behaviour is the “chaining” and thus preservation of discrete age classes (cohorts) in
the model, which through time lag and negative feedback conserves an imbalance of structure
from the one-time increased purchases and throughout the lifetime of the resources. In fact,
the one-time perturbation sets off a pattern of oscillations in performance, which, though
dampened, could continue for the whole duration of the operation. Note that the task is fully
deterministic, and complete information is contained on an A4 page. Based on the task
description alone, subjects will be asked to depict the performance curve as a function of the
budget input.
The task was first tested on N=7 defence analysts (Bakken et al 2004). When prompted, the
subjects did not manage to reproduce the cyclic performance resulting from the one-time
budget increase. Instead, graphs sketched (by hand) resembled more the “plain” budget
profile, but stretched out in time. Only one subject correctly perceived and rendered the
“unintended” performance penalty, but he had also a background in system dynamics (SD).
Detailed inspections of the graphs reveal that subjects stipulated an accumulated performance
boost of more than three times the actual increase.
150
1) secre reser ae eae ean teeta ate eaten aon ana
Subject 1
|—Subject 2
|—Subject 3
I—Subject 4
|—Ssubject 5
I—Ssubject 6
|—Subject 7
70
60
50
012.3 4 5 6 7 8B 9 10141213 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
Diagram 2: Estimated Total Effects sketched by subjects #1-7. The sketch by subject 4
is the only one closely resembling the analytic solution.
The second study involved N=10 military college cadets. The task was repeated as in study 1,
but this time with a debrief session immediately following the task itself. The subjects were
also presented with four questions aimed at assessing subjects’ confidence in own judgment
as well as their understanding of the problem at hand. To simplify, subjects were this time
presented with six template graphs (diagram 3), and instructed to choose individually the one
that in their opinion best resembled the actual performance profile. None of the subjects
managed to pick the best one (graph E).
la le le Ie Jel
i
Fe le le le |
C; 0 aE: 3
Diagram 3: Selection of template graphs to choose from (number of subjects choosing
the actual graph is indicated)
Despite the low number of experimental subjects, the obvious conclusion is that intuition is a
poor judge of dynamic (feedback) processes. The subjects appear to use the first order
intuitive judgment as an “anchor” to which subsequent adjustments may be made (processing
in spreadsheet or SD modelling software might reveal the analytic solution, but this approach
usually takes valuable time which otherwise might be spent on environment monitoring and
other management/executive activities). When the necessary SD insight is lacking, in this case
the ability to recognise that a significant time lag coupled with negative feedback tends to
produce oscillations around a baseline, such adjustments rarely happen. This kind of
misperception results in over-optimism when it comes to estimate the performance of the
system, because over-shoots are perceived as extended for a longer time than appropriate, and
“under-shoots” are not detected at all.
Case 3 — Crisis management
Rudolph and Repenning (2002) use a fairly simple system dynamics model to study how
organizations react to an ongoing stream of interruptions to normal activities. The key
assumption in the model is that stress is linked to the resolution rate through a relationship
known as the Yerkes—Dodson curve (Yerkes and Dodson 1908). The Yerkes—Dodson Curve
depicts a curvilinear relationship between stress and performance. Initially, stress improves
performance, but, as it continues to increase, performance eventually declines. When people
are under stress, autonomic arousal, which manifests as faster breathing and a higher heart
rate, heightens awareness and focuses attention on the sources of stress. For modest levels of
stress, this reaction improves task performance in most contexts. However, as stress grows,
autonomic arousal also rises and consumes scarce cognitive resources. That is, people become
distracted by the physical sensations and, consequently, performance declines (Repenning,
2003).
C—O
Interruption Interruption
Arval Rate Resolution Rate
DELAY Effect of Straas on Performance
Desired 2
Resolution Time a]
zx
a_
Desired 5
Rissolution Fate 3 E 1
i z
3
% 1 2
Normal __ 4”
Resolution Rate
Diagram 4: Structure of disaster model. Facsimile from Repenning (2003).
The system has two equlilibria, one stable and one unstable. If the level of interruptions
pending, and therefore the amount of stress, stays below the unstable equilibrium, the system
will tend to return to the stable equilibrium. If, however, the level of stress pushes the system
beyond the unstable equilibrium, then performance quickly collapses in a vicious cycle of
accumulating unresolved interruptions, mounting stress, and declining performance.
Rudolph and Repenning (2002) effectively used system dynamics to outline a new crisis
archetype, the quantity-induced crisis. While the existing literature focused on how novel
events could precipitate a crisis, their simple model suggested that disaster could also result
from a stream of more routine interruptions. Distinguishing between two types of
organizational crises, novelty-induced and quantity-induced, and their associated dynamics
has several implications for both practice and research. Most notably, their analysis suggests
that the strategies often proposed for mitigating novelty-induced crises — stepping back and
reframing the situation — can be counterproductive when confronting a potential quantity-
induced crisis. Any action that temporarily slows the interruption resolution rate can push the
system closer to the unstable equilibrium, making collapse more likely.
In his research framework from 1967, Perrow reasons along the same lines when he classifies
tasks to be solved in organisations.
Low
(Information search by|
chance & guesswork) . .
Craft industries
Analyzabili 7
Engineering
High
(Logical & analyticall
information search) Low Task variability High
Few exceptions Many exceptions
Diagram 5: Perrow’s complexity framework
The framework consists of two dimensions: The task’s variability, defined as the number of
exceptions encountered in the task resolution, ranging from few to many. The other dimension
is the analysability of the exception encountered. The exceptions may be analysable, which
means that the search process triggered by the exception is systematic and predictable. The
exceptions may also be un-analysable, which means that one does not know how the
exception should be handled. The search process is characterized by guesswork and a more
elaborate problem solving is required. This framework has been the source of much of the
research on the relationship between technology and organizational structure (see Haerem
2002).
Contrary to how the framework was originally used, it is not so that a task always belongs in
one of the quadrants. Organizations are more or less good at learning. Learning in
organizations is often defined as changes in routines (Argyris og Schon 1978) and learning in
this framework can be analysed as changes in the analysability and variability dimensions.
From the framework it is also clear that it is the exceptions in the tasks and the analysability
of these that make the task resolution processes challenging to handle. Good models for
exception handling provides those managing work processes necessary tools to handle the
processes more efficient.
Integrating CC, SA and SD
Brehmer (2000) points to two main principles when it comes to interpret people’s problems in
handling dynamic settings:
e Overemphasis on the present: Decision makers tend to attend to only the information
currently at hand, and as a result experience difficulties in accommodating feedback
delays. The world is perceived “here and now”, and the information is not processed any
further.
e Lack of systems thinking: the tendency to think linearly, that is, to believe that actions and
results are directly related and ignore the side effects of actions. This tendency can also be
viewed as an over-reliance on information that is readily available, along with a tendency
to ignore what must be inferred — such as side effects.
When discussing “command concepts”, Brehmer (2002) issues a warning against relying
solely on (direct) feedback control of a command process. Without taking into account
inherent delays, the command system would inevitably become reactive and lagging, resulting
in loss of initiative and subsequently loss of personnel and equipment. A (mental) model
needs to be able to produce reliable predictions of future events if it is to provide the basis for
a functional command concept.
Therefore, systems thinking imply first and foremost striking a balance between the feed-
forward and feedback modes of information processing. The feedback mode involves
perceiving the outcome of an (single or series of) action or event, in order to interpret its
causes (“meaning”). Feed-forward, on the other hand, is about being proactive — using
interpretations to generate possible future causes of events (mental simulation). The quality of
mental simulation will generally be dependent on the experience (expertise”) of the decision
maker, as well as the complexity of the situation.
The table below is an attempt to integrate command concepts (CC), situational awareness
(SA), and system dynamics/thinking (SD/ST) abilities. The leftmost column summarises the
requirements posed by CC and SA. The remaining columns exemplify challenges found in the
categories of time lags, feedback and non-linearities, respectively.
SD/ST challenges
CC/SA requirements
Time lags
Feedback
(positive/negative)
Non-linearities
1: Perception
Detect and classify
relevant objects (static
and dynamic) and
“features” in text and on
map.
The “Common
Operational Picture”
(COP) may be outdated,
ie., reflects a previous
state rather than the
current.
Misperception of the
“age” of a given COP
may inadvertently
accelerate feedback
processes.
A broad range of
heterogeneous objects
requires that mental
models be more
elaborate and complex.
2: Understanding
Mental model (picture)
of the current situation,
including events and
causes leading up to it.
Assessment of forces “at
play”, their traits,
capabilities and
command (personalities).
Different types of
objects may have
different times to target
and effect (the
difference being several
orders of magnitude at
the extreme).
Consider feedback
processes when
analysing possible
causes of events
leading up the current
situation.
(e.g. questioning the
“true” motivation of
the enemy).
“Force multipliers”:
is the idea that two or
more effectors
engaging a target
simultaneously yield a
higher total effect than
if the same effectors
was employed
sequentially.
3: Projection
Mental visualisation of
probable future
developments of the
situation. Assess
congruence between plan
and resources. Structure
adapted to desired effect.
CCIR including
confirming and refuting
signs.
Enemy development.
Evaluate C2 criticalities.
Contingency plans.
Mental models need to
incorporate the effect of
delays resulting from
e.g. congestion along
lines of transportation;
the time it takes to
disseminate orders and
intentions through the
chain of command; the
time from engagement
and to results.
Positive
(accelerating)
feedback may
dominate when the
operation is on a
“winning course”,
ie., gaining a
foothold on enemy
land may accelerate
the progress.
Negative (stabilising)
feedback is dominant
in the traditional
attrition situation;
when left
undisturbed, forces
will wear each other
down to extinction.
Plans should
incorporate the
possibility that while
forces engage and are
interconnected in
complex ways, a
“chaotic” state may
occur. While appearing
to be uncontrollable, it
is still important to
understand what
determines the
seemingly “chaotic”
condition.
Table 1: Examples of dynamic challenges in military operations and crisis management
Further work - A training solution for commanders and staff
Experimentation is required to define the training requirements, organizational design, and
information system requirements for adaptable performance of military coalitions (NATO,
2003). In connection with the development of NCW concept(s), it is necessary to create a
range of arenas for exercising and testing, where one can get experiences with both technical
and human aspects of the NCW concept. We suggest the development of a Minimalist
Decision Trainer’, or “learning lab”, for NCW with focus on the human aspects.
The training solution will have as its main ambition to expose and point to challenges,
dilemmas and opportunities inherent in the NCW concept. By providing personnel with a
foundation for reflection on these aspects, it is possible to contribute to both individual
understanding and organisational development. As a prerequisite for NCW lies the
development of a technology infrastructure, facilitating information gathering, sharing and
presentation. In a “learning lab” setting for NCW, we assume the required technical
infrastructure to be in place and fully operative, and will emulate’ the most essential features
of this infrastructure. The solution will on the basis of this assumption provide a “practice
arena” for organising, information gathering, communicating and decision making (within
certain boundaries).
Scenario
Interface
Decisions
Information
Diagram 6: Organisation layout of Learning Lab for NCW
? A Minimalist Decision Trainer (MDT) provides a solution for simplifying and compressing the training
environment. MDT is designed to be a very simple and pedagogically focused simulation-supported system (i.¢.,
microworld) for use in the training of General-level commanders (both existing and to-be). MDT is aimed at
putting a commander or the command group in charge of own logistics and operations resources in a scenario.
The scenario may contain any implied or explicit mission. The resources reflect a combined joint operation;
typically there will be less than a hundred units representing land, sea, air and other resources (Bakken and
Gilljam, 2003).
> Thus, in gaming mode we operate some time into the future.
The “learning lab” provides the arena where two (or more) teams, consisting of different units
(a unit may in turn be a single person or a team), can meet to solve problems subject to a
scenario. Every unit will have access to a networked personal computer. A simulation game
will be run over the network, where participating units receive information and submit
decisions. Together, the units’ decisions will have consequences for the progress of the game.
By developing a broad spectrum of scenarios, and enforcing restrictions on information access
before and during the game (while allowing communication within team and between
commander and team), it is possible to create valid experiences for participating players.
Conclusions
In this paper we have suggested a new approach to learning and understanding within the
concepts of Network Centric Operations and Effects Based Operations. To arrive at a solution
rooted in theories of both modern military leadership as well as dynamic decision making, we
have linked the principles of “best practice’ EBO modelling and analysis with System
Dynamic insights.
Experimental results from case #1, Operational logistics, indicate that learning does occur
from training, and that the number of practice trials explains performance, not the treatments
as such. A key training issue then becomes how such increased training frequency can be
obtained in practice. Training improvement research should have a divided focus: on how to
facilitate and guide participants’ mental processes, as well as on techniques and technology
for environment compression.
Case #2, Strategic investments, shows us (as expected) that intuition may be a poor judge of
dynamic feedback processes. When the necessary insight is lacking, decision makers rarely
arrive at correct conclusions. The consequences may be (as in this case) overly optimistic
estimates of cost and effect.
Case #3, Crisis management, draws on Repenning’s (2003) work in formulating a dynamic
model of a crisis archetype. Most notably, he shows that the strategies often proposed for
mitigating novelty-induced crises — stepping back and reframing the situation — can be
counterproductive when confronting a potential quantity-induced crisis. Any action that
temporarily slows the interruption resolution rate can push the system closer to the unstable
equilibrium, making collapse more likely. Interestingly, this approach is also consistent with
Perrow’s (1967) framework for classifying task complexity in organisations.
The cases have (among others) exposed a need for a “learning lab” for NCW. The purpose of
the learning lab would be, through man-machine and inter-team interaction, to improve
mental models of commanders and teams, and thus improved outcomes of future operations.
References
Bakken B et al. 2004. The intuitive vs. Analytic approach to Real World Problem Solving:
Misperception of Dynamics in Military Operations. Forthcoming in Proceedings from
the Human Factors Conference. Dunblane, Scotland. September 2003.
Bakken B, Gilljam M. 2003. Dynamic intuition in military command and control: why it is
important, and how it should be developed. Cognition, Technology and Work 5: 197-
205.
Bakken B E. 1989 Learning in dynamic simulation games: Using performance as a measure.
In Millin P M and Zahn E O K (eds). Computer-based management of complex
systems: Proceedings of the 1989 International Conference of the System Dynamics
Society. Springer, Stuttgart.
Brehmer B. 1988 Strategies in real time dynamic decision making. University of Chicago.
Chicago. (Einhorn Memorial Conference).
Brehmer B and Allard R. 1991 Dynamic decision making: Effects of complexity and
feedback delays. In J Rasmussen et al (eds). Distributed decision making: Cognitive
models of cooperative work. Wiley, Chichester.
Brehmer B. 2000 Dynamic Decision Making in Command and Control. In McCann & Pigeau.
The Human in Command: Exploring the Modern Military Experience. Kleuwer, New
York.
Brehmer B. 2002 Learning to control a dynamic system. Unpublished manuscript, Swedish
National Defence College, Stockholm.
Brewster FW. 2002. Using Tactical Decision Exercises to Study Tactics. Military Review,
Nov-Dec2002: 3-9.
Brunswik E. 1956 Perception and the representative design of psychological experiments.
Univ of California Press, Berkeley.
Builder CH et al. 1999. Command concepts. A theory derived from the practice of command
and control. RAND.
Davis PK. 2001. Effects-Based Operations (EBO): A Grand Challenge for the Analytical
Community. RAND.
Diesen. 2004. Nettverksbaser forsvar — hva er egentlig det? LUFTLED 1/2004: 4-6.
Dunwoody P T et al. 2000 Cognitive adaptation and its consequences: A test of cognitive
continuum theory. Journal of Behavioral Decision Making 13: 35-54.
Einhorn H J and Hogarth R M. 1978 Confidence in judgement: Persistence of the illusion of
validity. Psychological Review 85: 395-416.
Endsely M. 1988. Situation Awareness Global Assessment Technique (SAGAT). Proceedings
of the National Aerospace and Electronics Conference: 789-95. IEEE, New York.
Haerem T. 2002. Task Complexity and Expertise as Determinants of Task Perceptions and
Performance. Unpublished PhD dissertation. Norwegian School of Management, Oslo.
Hammond KR. 1996. Human judgement and social policy: Irreducible uncertainty, inevitable
error, unavoidable injustice. Oxford University Press, New York.
Jensen E. and Brehmer B. 2003. Understanding and control of a simple dynamic system.
System Dynamics Review, 19(2), 119-137.
Kagel JH and Roth AE. 1995. Handbook of Experimental Economics. New Jersey: Princeton
University Press.
Kainz D and Ossimitz G. 2002 Can students learn stock-flow thinking? An empirical
investigation. Proceedings of the 2002 International Conference of the System
Dynamics Society. Palermo, Italy.
Moxnes E. 1998 Not only the tragedy of the commons: Misperceptions of bioeconomics.
Management Science 44: 1234-1248.
NATO. 2003. Project description (unpublished): Leader and Team Adaptability in
Multinational Coalitions: Cultural Diversity in Cognition and Teamwork.
Ossimitz G. 2002 Stock-flow thinking and reading stock-flow related graphs: An empirical
investigation in dynamic thinking abilities. Proceedings of the 2002 International
Conference of the System Dynamics Society. Palermo, Italy.
Paich M and Sterman J D. 1993 Boom, bust, and failures to learn in experimental markets.
Management Science 39, 1439-1458.
Perrow C. 1967. A Framework for the Comparative Analysis of Organizations. American
Sociological Review 32 : 194-208.
Repenning J. 2003. Selling system dynamics to other (social) scientists. System Dynamics
Review 19(4): 303-328.
Rudolph J, Repenning N. 2002. Disaster dynamics: understanding the role of stress and
interruptions in organizational collapse. Administrative Science Quarterly 47: 1-30.
Sterman J D. 1989 Modeling managerial behavior: Misperceptions of feedback in a dynamic
decision making experiment. Management Science 35: 321-339.
Sterman J D. 2000 Business dynamics: Systems thinking and modelling for a complex world.
McGraw-Hill, Boston.
Sweeny L B and Sterman J D. 2000 Bathtub dynamics: Preliminary results of a systems
thinking inventory. Presented at the 2000 International System Dynamics Conference,
Bergen, Norway.
Tversky D and Kahneman D. 1987 Rational choice and the framing of decisions. In Hogarth
RM and Reder M W (eds). Rational choice: The contrast between economics and
psychology. University of Chicago Press, Chicago.
Yerkes RM. and Dodson JD. 1908. The relation of strength of stimulus to rapidity of habit
formation. Journal of Comparative Neurological Psychology 18: 459-482.
Back to the Top
