11-1

Click here to view PowerPoint presentation; Press Esc to exit

Human Behaviour Representation -

Definition

Dr. Uwe Dompke

NC3A, ORFS Division
Oude Waalsdorper Weg 61
2501 CD The Hague
The Netherlands
+31(70) 374 3636
Uwe.Dompke@nc3a.nato.int

Abstract

This paper is taken from the Final report of LTSS SAS-017 on Human Behaviour Representation in which A.
v. Baeyer and J. Mylle gave this definition of HBR.

1 The importance of “Human Behaviour Representation”

The end of the Cold War has brought new military tasks and types of operations to NATO. These include
regional contingency operations, Crisis Management and support of non-NATO missions (UN, PfP, WEU,
etc). All these new types of missions have one newly emerging characteristic that implies a better modelling
of human behaviour: they need methodologies to describe on various detail levels, how individuals and
social entities (e.g. teams, groups) influence the course and outcome of military conflicts. Those emerging
technologies will have a great impact on the implementation and on the military use of simulation systems in
the future. Therefore, building better simulation systems and decision support tools, which include HBR, is
of primary importance.

2 Definitions

2.1 Human behaviour

Human behaviour (B) is a purposive reaction of a human being (P) to an idiosyncratic meaningful
situation (S).

Formally expressed: B= f (P,S). In words: the observed variability in behaviour is attributable to differences
in the person’s characteristics, to differences in the situation and/or to the interplay of both.

• Mathematically spoken: the variation in the measured behaviour can be explained by the
variation in P, the variation in S and interaction between P and S (measurement error not taken
into account). This definition implies that human behaviour:

• is a change from one state into another state (bodily and/or mentally);

• is always goal-oriented (but not necessary in a one to one relation)

• is a reaction to an external observable stimulus or to an internal covert stimulus,

• has three interrelated components: a cognitive, a psycho-motor and a socio-affective component

Paper presented at the RTO SAS Lecture Series on “Simulation of and for Military Decision Making”, held in
Rome, Italy, 15-16 October 2001; Stockholm, Sweden, 18-19 October 2001; Virginia, USA, 23-25 October 2001;
The Hague, The Netherlands, 10-11 December 2002, and published in RTO-EN-017.
11-2

• is an integration of several physiological and mental processes

• is individualised because each individual interprets the objective characteristics of the situation;

• is neither necessary „rationale“ nor the most appropriate reaction under given circumstances.

2.2 Representation

Representation means mapping (f) characteristics of empirical phenomena (Pe) into values of parameters in
an artificial world (Pa). Thus a representation is determined by {Pe, Pa, f}.

This means that we need a (formal) system to describe Pe Pa and the mapping “function” f.

It must be stressed that mapping does not necessarily mean mimicking or portraying. For example, a plane
that flies does not mimic the behaviour of a flying bird.

2.3 Model

A model is a simplified representation at a conceptual level of (a part of) the real world and/or the way it
behaves, that suffices to make some deductions concerning the real world and/or it’s functioning. A model
consists of components and the relationships between those components, which are generally cause-effect
relationships. A model can be visualized in some graphical form.

For example, the three stage-model of memory (here rendered in its oversimplified form).

The graph below means that information that is kept by receptors is temporary stored in the sensory memory,
is then transferred to the working memory, and after elaboration in this part, is permanently stored in the
long term memory. Furthermore, information can be retrieved from the long term memory by the working
memory (and then used for some behavioural purposes).

Sensory memory Short term Memory Long term memory

(working memory) (permanent memory)

Figure 1: Example of a Model

Essential characteristics of a model are:

• Reduction of the complexity of the real world

• Highlighting what is considered as essential or important

• Transparency of the relationship(s) between the components

• Putting the representation in a certain perspective, based on the choice of the components and
the relationship between components

• Productivity: models allow for the discovering of (working) hypotheses, for new insights at a
certain level of quantification, for the verification of the impact of changes (by changing
 11-3

starting values, by adding or dropping components and/or relationships: cfr path analysis; e.g.
LISREL).

A model implies that:

• A choice must be made which characteristics will enter the model and which not

• Therefore, one or more selection criteria are needed (which define the perspective)

• Characteristics are to be reflected in variables that can be quantified

In the context of this LTSS the notion “model” does NOT refer to the physical device that shows how
something works in reality (e.g. a model of the wings of a plane and the air streams around it) nor to the
notion as defined in social learning, i.e. the person whose behaviour is mimicked.

Aspects of modelling human behaviour

• Given that human behaviour is purposive, not only the behaviour itself must be modelled
but also the goal, which has to be converted in a SMART objective (Specific, Measurable,
Acceptable, Realistic, Timed).

• This objective is the touchstone in determining what is the optimum behaviour under the given
circumstances.

• In its most simple expression modelling behaviour means: determining initial values for P and S
(input), run a process (throughput) which lead to an outcome (output); i.e. a change in P and/or
S.

• The model requires a “level of behavioural analysis”:

A single action an individual

A contingent series of actions by a team

A series of actions executed in parallel an organisation

HBR modelling relies, as to the content of the models, on:

• Military experience

• Findings of human science research

The latter defines and structures, using established terms and theories, all relevant types of human behaviour.
There are additional concepts that should be used as in for any further discussion and HBR science
development. These concepts are:

• Individual

• Team

• Group (small group and large group)

• Organisation

• Crowd

• Public opinion (shaped by the media)

11-4

The basic concept is that of the individual.

Team, group, organisation and crowd are specific collections of individuals, which are defined according to
different behavioural goals and interactions (both within the collective entity and as a collective entity
interacting with others).

As for public opinion see Chapter 4 Section 4.2.1.

The scientific task with respect to the individual, team etc is to find and define the specific goals and
interactions of the individual and of each collection of individuals, thus specifying what goals and
interactions:

• build a military team from few skilled individuals

• make a small or a big group under certain special conditions

• create or change an organisation

• constitute a crowd, when many people suddenly behave in a uniform manner.

The task of modelling individuals, teams etc is

• to take the typical behaviour expressed in goals and interactions

• to put quantitative measures on the internal and external behaviour of the respective entity.

So far the characteristic behaviours that differentiate a team from any smaller or larger group or organisation
or crowd have not been elaborated to the point where they can be used for modelling in a military context.
Therefore, the first scientific endeavour is to collect on a systematic and scientifically accepted manner this
body of knowledge, which can then be used by for modelling purposes.

• For each characteristic (= variable) a scale of measurement1 has to be determined and

“baseline”, i.e. an initial parameter value.

• A distinction has to be made between trait and state:

• a trait is a covert cross-situational and cross-temporal stable characteristic of a person but which
expression is modulated by the perceived situation.

• A state is a condition not manifesting any marked change with respect to some quality or
property.

For example, a person who shows often anxiety in his behaviour in different circumstance is an
anxious person. This person is characterised by anxiety as a trait.

A person who is rarely seen anxious but has been confronted with a traumatic event is temporary
anxious. This person is in a state of (intense) anxiety.

1A scale is the triple {U, N, f} in which U stand for the observed variations in the empirical phenomenon, N for the
numerical values that are assigned to the observed “states” of the empirical phenomena using a function f.
 11-5

2.4 Simulation

Simulation means:

• A method to implement a model in some environment and/or in a device, which may be totally
or partially artificial (instead of real).

• A technique for analysing, testing, evaluating the effect of some values of the parameters of the
model on other parameters. For example, a decision making process (e.g. choose a tactical plan
among three alternatives) can be simulated on a computer, using some algorithm (among others,
a decision tree).

• Using a tool (a simulator) for instruction or training purposes, e.g. flight simulator.

The following categories of simulation are used commonly in the NATO context/language of Modelling &
Simulation:

Live simulation: real humans operate real equipment in a real environment except for some parts or
aspects (e.g. OPFOR is not a real foe but a unit that acts as enemy; blank ammunition or hit/kill indicators
instead of wet ammo; terrain is a training camp or a civilian area but not the operation zone)

Virtual simulation: real humans operate simulated equipment in a simulated environment (e.g.
Computer Assisted (Command Post) Exercises; a pilot in a flight simulator).

Constructive simulation: simulated people (or units) operate simulated equipment and/or behave in a
simulated environment. The intervention of real humans is limited to “initialise” the simulation run.

2.5 Instruction/learning

During instruction, a person learns a new cognitive, psychomotor or socio-affective “item” relevant to the
organisation he is part of. Instruction takes place in a specific environment such as a school. Instruction is
given by qualified personnel who are required to show the optimal behaviour and is thus better skilled than
the trainee.

Learning in this sense, is defined as a long lasting desired behaviour modification under the influence of the
repeated exposure to adequate stimuli. The learning process is (constantly) monitored by the instructor. The
behaviour of the trainee is corrected on the spot. Feedback is procured during and after the training process.

E.g. an officer learns how to write an operations order for his subordinates, a soldier learns how to aim and
fire a weapon, a new formed squad/crew learns what cohesion means to their functioning as a group.

2.6 Training

Training means to repeat a learnt behaviour in order to enhance the performance, or to internalise a norm, or
to develop an attitude in a rather simplified environment. For example, a pilot is trained in applying flight
procedures (in a simulator); discipline is trained throughout different situations, which require obedience to
the established rules. Training does not differ from instruction with respect to the trainer/trainee interaction.

2.7 Exercise

An exercise aims at using what has been learned and trained but now in a less or more real like situation. In
an exercise real people use real means in a real environment. For example, command post exercises; or full
troop exercises live firing exercises in battle runs.
11-6

An exercise is controlled by an observer-controller (or a coach) who does not intervene in the course of
action, but who is responsible for detailed after-action review. Moreover, the observer is not necessary better
skilled than the people who are exercising.

2.8 Operation
Operation refers to real life (military) activities in order to realise a given mission. For example, the
deployment of a task force, a brigade attack, peace keeping as a particular case of peace support operations.

2.9 Performance
Performance refers to the behaviour itself or its result/outcome in one of the following forms:

• Executing and finishing a certain task. For example, fire until hit.

• A numerical expression related to the behaviour. For example, defend a position during 4 hours

• Expression of the competence, moderated by some personal and/or situational variables. For
example, physical performances under bad weather conditions (too cold, too warm) are lower
than under good weather conditions.

2.10 Competence
Competence means the best possible behaviour of a given person with respect to certain ability for a
particular job or vocation. Competence is a higher order notion for a series of performances that meet high
standards. For example, a competent leader is a leader who is often successful in conducting operations (of
different kind).

2.11 Optimal behaviour

Optimal behaviour is the best possible performance given under the circumstances; i.e. taking personal and
situational constraints in to account; thus: B=f (P,S). This means also that an optimal performance is lower
than or equal to the maximal performance (=competence).

2.12 Situational awareness

Situational awareness is the result of the perception of a number of elements in the environment within a
given timeframe and space, their meaning with respect to the mission at hand and their possible evolution in
the near future that must be taken into account in determining one’s own behaviour.

Situational awareness is multidimensional because it deals with spatial information (e.g. where is who/what),
with information contained in the mission (e.g. what to do within which time interval), own means (e.g.
“readiness” of crews) and available resources (e.g. which fire support).

2.13 Validation, Verification, Accreditation

• Validation means determining the degree to which a model is an adequate representation of the
real world (or a part of it) for the purposes it has been conceived for.

• Verification refers to the process of determining to what extend the implementation of a model
corresponds to the design specifications determined by the customer and of providing the proof
that the model runs (as its has to run).

• Accreditation is the official certification that a model or simulation is acceptable for use with
respect to its purpose(s).
 11-7

3 A Concise Structural Approach of HBR as a Framework

This section tries to give a global approach of HBR in military applications. However, a global theory
cannot be the goal of this LTSS and must be postponed, when more information about the practical use of
HBR for any field is available. Therefore, we have to limit an approach to a framework.

3.1 General Approach

The concise theory must avoid two fallacies:

• on the one hand to be too short and therefore trivial

• and on the other hand to be too scientific and lengthy.

This dilemma will be avoided by presenting a logical deduction and phenomenological description of those
human behaviours, which are essential in military operations (including CRO) and the way they should be
represented for analysis, training and simulation.

By logical deduction it is understood that the theory is coherent in itself, and not just an arbitrary collection
of abstract statements.

By phenomenological description a description is meant, which focuses on typical and relevant events and
describes them as close to (military) common sense as possible.

3.2 Objects of HBR-Theory

The ultimate goal of HBR is (as this is the case in every military simulation), to represent behaviour that is
typical and relevant in military operations. Relevant is every human behaviour, which is decisive for the
mission success. But in order to make it the object of a theory and therefore the object of well elaborated
models and simulations, the human behaviour must be “reduced” to certain “system simplifications”.

What is the “level of detail” that must be employed to describe the human behaviour?

The level of detail of a rough task analysis but reduced to general mental (i.e. cognitive, psycho-motor, social
and physiological) functions (examples see below).

What is relevant for the representation of behaviour?

Every behaviour that is needed in analysis, training and simulation is considered as relevant.

3.3 Elementary Behaviours (mental functions)

The following is a taxonomy, which lists mental functions in a systematic way appropriate for modelling.

¾ Actions – Observable behaviour in the outside world

♦ Interactions with real objects and real environment - physical skills (e. g. driving a car,
digging a hole)

♦ Symbolic interactions (e.g. communication by speech, in documents by gesture)

♦ Social interactions (e. g. speaking to another person)

ƒ Interpersonal relations (with superiors, subordinates, colleagues)

11-8

ƒ Group dynamic influences (small group effects; e.g. cohesion)

ƒ Organisational influences

¾ Decisions – not directly observable behaviour within the person – making up one’s mind.

♦ Situational awareness (e. g. observing an area)

♦ Information collection (e. g. map reading)

♦ Information processing – from sensory input to mental models (e. g. identifying an object as
foe)

♦ Schemata (e. g. knowledge, skills)

ƒ Declarative knowledge

ƒ Procedural knowledge and skills

ƒ Meta-cognitive skills (control of behaviour, evaluation)

♦ Deciding for action (e. g. shooting to a foe object, giving an order to shoot)

¾ Psychical and physiological traits / states

♦ Age

♦ Values (e. g. personal standards, cultural values, beliefs, attitudes)

♦ Moods / emotions (e. g. anxiety)

♦ Motives (e. g. high performance motive)

♦ Alertness / vigilance (e. g. high vigilance)

♦ Stress (e. g. loss of sleep, heavy workload)

¾ Dynamic changes of behaviour

♦ Learning and Instruction (slow)

♦ Traumatic experience (quick)

♦ Duration of stress

For the scope of this paper it is sufficient to use this simple taxonomy. It can be used for three purposes:

• to describe a global structural concept of the interrelationships between the behaviours (human
behaviour “model”)

• to establish a list of compound behaviours

• to evaluate the feasibility of special elementary human behaviour model

• to define the different kinds of representation for analysis, training and simulation.
 11-9

3.4 Compound Behaviours

No single behaviour type exists by itself. All types are always combined and have to be represented at least
together with some of the other types. The global behaviour model, therefore, serves the purpose of
identifying the interrelationships of the behaviour types. Internal behaviour (within the person) and external
behaviour (towards other persons and with objects in the environment) are combined into compound
behaviours. See the following picture.

Person
Other Persons
Other Persons
Decis. Other Persons
Actions
{☛}
{;}
internal
behaviour
External behaviour of the acting person

State
{Ψ+Φ}

Objects

Picture 2: Person and its behaviour in an environment

¾ Objects and persons provoke decisions and actions.

¾ Motives influence the choice of individually available schemata.
¾ Schemata influence decisions.
¾ Decisions make persons act.
¾ Age values, moods, alertness and stress moderate decisions.
¾ Age values, moods, alertness and stress moderate directly actions.
¾ Actions change values, moods, motives, alertness
¾ Actions change decisions.
¾ Decisions change values, moods, motives.
¾ Objects, persons and stress cut off the decision-making processes.

Table 1: Internal compound behaviour

11-10

¾ Actions change objects.

¾ Actions influence other persons.
¾ Other person’s actions and objects of the environment make persons change their psychical and
physiological states.

Table 2: External behaviour

3.5 Data Sources

Every behaviour produces different data. The data can be classified into two groups:

• Directly accessible data, which are subdivided into

- measurable and observable data {m+o}

- only observable data {o}

• Data that must be constructed indirectly by theories {c}.

Applying this simple data taxonomy to the behaviour types, it can be seen, whether it is (relatively) easy to
build models about the behaviours in question (on model building in HBR see below):

• Directly accessible data {m} and {o} make model building easier; many models do already
exist and are used

• Constructed data make model building difficult.

The reason lies in the validation and accreditation of the data. Directly accessible data can by verified
through objective methods and the corresponding theory. Constructed data are entirely based upon the
validity of the theory.

Directly accessible data can be sub-divided into:

• Observable data at a high level of measurement; i.e.

- on an interval scale. For example: body temperature, precision of aiming at a target.

- on a ratio scale. For example: speed of movement, estimation of the distance to a target.

• Observable data at a low-level of measurement, i.e.

- at a nominal level or classifying objects. For example, friend or foe, identifying a object
as a main battle tank

- at an ordinal level or ordering objects with respect to a certain criterion. For example,
“intensity” of threat of different enemy manoeuvres.

- at the hyper ordinal level or comparing differences between (two) sets of two objects.
For example, different balances of forces.

Unobservable data refer to psychological constraints. The way they are “operationalised” i.e. made
observable. The process of operationalisation determines also their level of measurement.
 11-11

For example, intelligence is made observable through the responses to an intelligence test and can be
expressed in an intelligence quotient IQ (which is at the interval level). For example, group cohesion can be
expressed through a ranking of the co-workers by each of the group members. Using multi-dimensional
scaling these rankings can be converted into a map with distances between the members. The greater the
distances the less cohesion is.

Complex constructs, for example combat readiness –involving several domains of human behaviour (i.e.
cognitive, motor and/or socio-affective) constitute a specific challenge for modelling because the whole is
more and something other than the sum of its components. Therefore, it is suggested to approach such
constructs in a holistic way rather than in an analytical way.

Behaviours Data
Actions {;}
Interactions with real objects and real m+o
environment (e. g. driving a car, digging a
hole)
Symbolic interactions (e. g. writing) m+o
Social interactions (e. g. speaking to another person) m+o
Decisions {☛}
Situational awareness (e. g. observing an area) o
Information collection (e. g. asking an expert) o
Information processing (e. g. identifying an object as foe) o
Deciding for action (e. g. shooting to a foe object, giving an order to shoot) m+o
Psychical and physiological states {Ψ+Φ}
Values (e. g. cultural values) c
Moods / emotions (e. g. anxiety) c
Motives (e. g. high performance motive) c
Alertness / vigilance (e. g. high vigilance) m+o
Stressors (e. g. loss of sleep, heavy workload) m+o
Schemata (e. g. knowledge, skills) c
Dynamic changes of psychical and physiological states {∆ψϕ}
Learning (slow) c
Traumatic experience (quick) o
Duration of stress m+o
Table 3: Matrix of elementary behaviours and the related data sources

4 Model Building in HBR

Models of human behaviour are always models of compound behaviours. It is still uncertain, whether an all-
embracing model of the Human Behaviour is feasible and necessary for military simulation. The following
matrix gives an overview on the feasibility models. Every compound behaviour is analysed as to whether
models are

• Easy to develop, even if they do not yet exist {e}

• Difficult to develop, even if they already exist {d}

• Impossible to develop {i}.

11-12

Feasibility of Model Building

Behaviours Easy Difficult
Objects and persons provoke decisions and actions. e
Motives influence the choice of individually available schemata. d
Schemata influence decisions. d
Decisions make persons act. e
Values, moods, alertness and stress moderate decisions. d
Values, moods, alertness and stress moderate directly actions. d
Actions change values, moods, motives, alertness d
Actions change decisions. d
Decisions change values, moods, motives. d
Objects, persons and stress cut off the decision making processes. e
Actions change objects. e
Actions influence other persons. d
Other person’s actions and objects of the environment make persons d
change their psychical and physiological states.
Table 4: Matrix of behaviours and the feasibility of model building

Conclusion: There is no human behaviour, which cannot be made the object of a model. However, the
majority is difficult to model and relies mainly on theories of the elementary behaviours (mental functions),
which themselves are difficult to validate.

5 References
Dompke, U., Scheckeler, K., Final Report on Long Term Scientific Study (LTSS/SAS-017) on Human
Behaviour Representation (HBR) Technology, RTO-TR-047, AC/323(SAS-017)TP/25, Brussels, 2001