Machine Understanding and Avoidance of Misunderstanding
in Agent-directed Simulation and in Emotional Intelligence
Tuncer Ören
1
, Mohammad Kazemifard
2
and Levent Yilmaz
3
1
School of Electrical Eng. & Computer Science,University of Ottawa, 800 King Edward Ave.,Ottawa,ON,Canada
2
Department of Computer Engineering, Razi University, Kermanshah, Iran
3
Department of Computer Science and Software Engineering, Auburn University, Auburn, AL, U.S.A.
Keywords: Agents with Emotional Intelligence, Emotional Intelligence Simulation, Machine Understanding,
Avoidance of Misunderstanding.
Abstract: Simulation is being applied in many very important projects and often it is a vitally important infrastructure
for them. Several types of computational intelligence techniques have been part of the abilities of simula-
tion. An important aspect of intelligence is the ability to understand. Agent-directed simulation (ADS) is a
comprehensive paradigm to cover all aspects of synergy of software agents and simulation and our approach
is to develop agents with understanding abilities. After a brief review of ADS, our paradigms of machine
understanding is presented. The article clearly indicates types of misunderstandings that might occur. Our
research plans are to avoid some of the misunderstandings which could occur and especially to have self-
attesting abilities in our applications to document which types of misunderstandings are avoided.
1 INTRODUCTION
Simulation is being applied in many very important
projects and often it is a vitally important infrastruc-
ture for them. Several types of computational intelli-
gence techniques have been part of the abilities of
simulation (Ören, 1995); (Yilmaz and Ören, 2009).
An important aspect of intelligence is the ability
to understand. Our research on machine understand-
ing started with understanding of simulation pro-
grams (Ören et al., 1990) and evolved to understand-
ing software in general (Ören, 1992), then to
understanding systems (Ören, 2000), to agents with
ability to understand emotions (Kazemifard et al.,
2009), and finally to machine understanding in emo-
tional intelligence simulation (Kazemifard, et al.,
2013).
Failure avoidance has been recently intro-duced
to advanced simulation studies as a paradigm in
addition to validation and veri-fication studies (Ören
and Yilmaz 2009).
This article is built on our previous work
especially on machine understanding as applied to
agents used in simulation and to misunderstanding.
However, in this article, we provide three additional
machine understanding paradigms in addition to our
basic machine understanding paradigm.
After this introduction, we start with a concise
review of our view of agent directed simulation
which provides a comprehensive framework to
consider all aspects of the synergy of software
agents and simulation. In section 3, we discuss our
four paradigms for machine understanding. Section
4 covers a systemati-zation of most types of
misunderstanding appli-cable to machine
understanding. Sections 5 covers the conclusions
and some of our plans for future research.
2 AGENT-DIRECTED
SIMULATION
Agent-Directed Simulation (ADS) is a unifying and
comprehensive framework that allows integration of
agent and simulation technologies (Ören, 2000).
Agents are often considered as model design
metaphors in the development of simulations. Yet,
this narrow view limits the potential of agents in
improving various other dimensions of simulation
(Yilmaz and Ören, 2009). To this end, ADS is
comprised of three distinct, yet related areas that can
be grouped under two categories as follows:
Simulation for Agents (i.e., agent simulation)
318
Ören T., Kazemifard M. and Yilmaz L..
Machine Understanding and Avoidance of Misunderstanding in Agent-directed Simulation and in Emotional Intelligence.
DOI: 10.5220/0004635003180327
In Proceedings of the 3rd International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH-2013),
pages 318-327
ISBN: 978-989-8565-69-3
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
involves the use of simulation modeling method-
methodology and technologies to analyze, design,
model, simulate, and test agent systems. This
includes, but is not limited to using agents as
model design elements (i.e., agent-based
modeling).
Agents for Simulation: (1) Agent-supported
simulation involves the use of agents as support
facilities to enable computer assistance in
simulation-based problem solving (e.g., simulation
experiment management); (2) Agent-based
simulation, on the other hand, focuses on the use
of agents for the generation of model behavior
(e.g., simulator coordination, run-time models) in a
simulation study as well as agent-initiated
simulation.
In agent simulation, agents possess high-level
mechanisms that include communication protocols
for interaction, task allocation, coordination of
actions, and conflict resolution at varying levels of
sophistication. Agent-based simulation focuses on
the use of agent technology to monitor and generate
model behavior. This is similar to the use of
Artificial Intelligence techniques for the generation
of model behavior (e.g., qualitative simulation and
knowledge-based simulation). Agents can provide
cognitive architectures that allow reasoning and
planning and serve as run-time models of simulation
model behavior management such as dynamic model
updating and symbiotic simulation. That is, context-
awareness of intelligent agents can facilitate
simulator coordination, where runtime decisions for
model staging and updating takes place to facilitate
dynamic composability. On the other hand, agent-
supported simulation enables the use of agents to
support simulations as well as simulation studies by
enhancing cognitive capabilities in problem
specification, simulation experiment management,
and behavior analysis.
Often, agent-supported simulation is used for the
following purposes (Yilmaz and Ören, 2009):
To provide computer assistance for frontend and/or
backend interface functions;
to process elements of a simulation study
symbolically (for example, for consistency checks
and built-in reliability); and
to provide cognitive abilities to the elements of a
simulation study, such as learning or
understanding abilities.
3 MACHINE
UNDERSTANDING
In the study of natural phenomena, the role of simu-
lation is often cited as “to gain insight” which is
another way of expressing “to under-stand.” Under-
standing is one of the important philosophical topics.
From a pragmatic point of view, it has a broad appli-
cation potential in many computerized studies in-
cluding program understanding, machine vision,
fault detection based on machine vision as well as
situation awareness and assessment. Therefore, sys-
tematic studies of the elements, structures, archi-
tectures, and scope of applications of com-puterized
understanding systems as well as the characteristics
of the results (or products) of understanding pro-
cesses are warranted.
Dictionary definitions of “to understand” include
the following: to seize the meaning of, to accept as a
fact, to believe, to be thoroughly acquainted with, to
form a reasoned judgment concerning something, to
have the power of seizing meanings, forming rea-
soned judgments, to appreciate and sympathize with,
to tolerate, and to possess a passive knowledge of a
language.
For machine understanding, or computerized un-
derstanding, we aim a limited scope as was ex-
pressed in a previous publication: “We say that a
system ‘knows about’ a class of objects, or relations,
if it has an internal relation for the class which ena-
bles it to operate on objects in this class and to
communicate with others about such operations.
Thus, if a system knows about X, a class of objects
or relations on objects, it is able to use an (internal)
representation of the class in at least the following
ways: receive information about the class, generate
elements in the class, recognize members of the
class and discriminate them from other class mem-
bers, answer questions about the class, and take into
account information about changes in the class
members” (Zeigler 1986)." (Ören, 2000). For addi-
tional clarification of understanding and its philo-
sophical roots see Ören (2000).
3.1 Machine Understanding:
Basic Paradigm
As seen in Figure 1, an understanding system re-
quires the provision of a meta-model, the perception
of the source in terms of the elements and constrains
depicted in the meta-model, and an analyser that
allows mapping of the perceived elements to con-
structs of the meta-model. In this context, a model is
an abstraction of phenomena or system, whereas a
MachineUnderstandingandAvoidanceofMisunderstandinginAgent-directedSimulationandinEmotionalIntelligence
319
meta-model provides an abstraction of the properties
of the model itself. A model conforms to its meta-
model in the way that a computer program conforms
to the grammar of the programming language in
which it is developed.
Figure 1: Machine understanding – Basic paradigm.
A system A can understand an entity B (Entity,
Relation, Attributes) if and only if three conditions
are satisfied (Ören et al., 2007):
A can access C, a meta-model of Bs. (C is the
meta-level knowledge of A about Bs.) The meta-
model can be unique or multiple, fixed, evolvable,
replaceable, or functionally equivalent (similar but
not identical) to another one. In the basic para-
digm, we assume that the meta-model is unique
and fixed, i.e., is non-evolvable and non replacea-
ble.
A can analyze and perceive B to generate D. (D is
a perception of B by A with respect to C.)
A can map relationships between C and D for
existing and non-existing features in C and/or D to
generate result (or product) of understanding pro-
cess.
As shown in Figure 2, a functional decomposi-
tion reveals that an understanding system has a me-
ta-model, an analyzer, and an evaluator. The meta-
model stores knowledge about Bs. The analyzer
analyzes inputs with respect to C to produce a per-
ception of B. The evaluator can compare the percep-
tion of B with the meta-model to provide additional
information about B, such as its non-observable
characteristics and how this instance of B relates to
other Bs. The product of the understanding process
has the following characteristics:
It depends on the understanding system; that is,
another understanding system may have a different
understanding of the same entity.
For a system A, understanding depends on: (1) its
meta-model, (2) its analyzer, and (3) its evaluator;
that is, with a different meta-model, analyzer, or
evaluator, the understanding may differ.
Figure 2: Functional decomposition of an under-standing
system (Arrows indicate information flow).
3.2 Machine Understanding:
Extended Paradigms
The basic paradigm of understanding can be extend-
ed
three ways to (1) rich understanding, (2) explor-
tory understanding, and (3) theory-based under-
standing. As will be clarified in the sequel, the four
metaphors for machine under-standing have the
following characteristics: Basic paradigm of under-
standing: system has background knowledge (i.e.,
meta-knowledge) to understand. Rich understand-
ing: All or some of the understanding elements may
have more than one version. Exploratory under-
standing: Background knowledge (meta-model) has
to be found or developed to process the perception.
Theory-based understanding: A theory (or theoreti-
cal model) if formulated without any observation;
then technology has to be developed for observation
of phenomena. Once the phenomena are observed
(perceived) they can confirm the theory which in
turn is used to explain the phenomena.
3.2.1 Rich Understanding Paradigm
A model of rich understanding is represented in
Figure 3. The difference of basic model of under-
standing and rich understanding stems from the
following:
There can be more than on meta-model in rich
understanding – some may focus on different as-
pects or may have different resolutions.
There can be more than on perception of the entity
to be understood.
There can be different interpretations of the per-
ception(s) with respect to meta-model(s).
B: entity to be understoo
d
Understandin
g
s
y
stem A
Analyze
r
D: perception of B (by A)
with respect to C
C: meta-model of Bs
(meta-knowledge
about Bs
)
Evaluator (interprets
the relationship
between C & D)
U: an understanding of B
b
y A
with respect to C
Entity B
System A
Cs:
meta-models
Ds: percep-
tions
Relationships: C(s)-D(s)
can understand
1
2
3
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
320
The total number of understandings may be the
Cartesian product of the meta-model(s), percep-
tion(s) and interpretations. Rich understanding can
allow multi-understanding and switchable under-
standing.
Figure 3: Rich understanding.
3.2.2 Exploratory Understanding
Paradigm
Exploratory understanding process (see Figure 4) starts
with a perception D. Formulation of basic knowledge to
interpret perception D requires
Figure 4: Exploratory understanding.
meta-knowledge to be formulated and/or to be found. This
would require formulation and testing of hypotheses. In
exploratory understanding, changing the point of view
may be very useful to understand the phenomenon or the
entity.
3.2.3 Theory-based Understanding
Paradigm
Theory-based understanding starts with a hypothesis
(or theory); then necessary technology would be
developed to perceive (detect) relevant phenomena
that would be tested later. A well known example is
the gravitational waves (ripples of spacetime caused
by events such as colliding neutron stars and merg-
ing black holes) which were predicted in 1916 by
Einstein based on his theory of general relativity.
Still technology to detect gravitational waves is not
available
.
As another example, in nuclear physics, several
models to explain elementary particles have been
developed over the years; this exemplifies the exist-
ence of several meta-models. Pictorial representation
of theory-based understanding would be similar to
Figure 3 representing rich understanding.
3.3 Machine Understanding
of Emotions:
Emotional Intelligence Simulation
According to the theory of emotional intelligence
(Mayer and Salovey, 1997), four psychological
abilities that enable humans to relate emotionally to
one another are: (1) emotion perception, (2) thought
facilitation using emotions, (3) emotion understand-
ing, and (4) emotion management. The ability to
understand emotions is desirable in intelligent agents
(Dias and Paiva, 2009), (Kazemifard et al., 2009),
(Kazemifard et al., 2012).
"Emotion understanding is a cognitive activity of
making inferences using knowledge about emotions
about why an agent is in an emotional state (e.g.,
unfair treatment makes an individual angry) and
which actions are associated with the emotional state
(e.g., an angry individual attacks others)" (Kazemi-
fard et al., 2013).
A functional decomposition of our emotion un-
derstanding framework –which is an extension of
our basic machine understanding paradigm– is de-
picted in Figure 5.
Our emotion understanding framework consists
of four elements (Kazemifard et al., 2013):
a meta-model or knowledge about agents and emo-
tions. It consists of an episodic memory to store
observed details of experienced events and a se-
mantic memory to store general knowledge about
emotions, such as their similarities and relation-
ships among emotions and experiences in episodic
memory. The semantic memory includes semantic
graphs to represent knowledge about past emotion-
al experience(s).
a perceptor (or analyzer) to perceive agents and
emotions. It assigns similar agents to types and
perceives the emotional states of agents.
an evaluator of the perceived agent and the emo-
tion(s) with respect to the meta-knowledge, that is
the states of the episodic and semantic memories.
a memory modulator to update meta-model based
on observed emotional reactions of agents to act
ions.
Entity B
System A
Cs: Meta-
models
D: percep-
tion
Relationshi
p
s: C
(
s
)
-D
(
s
)
can understan
d
2
1
3
Entity B
System A
Cs:
meta-models
Ds: percep-
tions
Relationshi
p
s: C
(
s
)
-D
(
s
)
can understan
d
1
2
3
MachineUnderstandingandAvoidanceofMisunderstandinginAgent-directedSimulationandinEmotionalIntelligence
321
Figure 5: Functional decomposition of the framework of
emotion understanding (Kazemifard et al., 2013).
4 MISUNDERSTANDING
There are three possibilities for the outcome of an
understanding system,: (1) the system can provide
an understanding of the entity or phenomenon, (2)
the system cannot understanding it and can or cannot
declare its inability to understand, and (3) the system
provides a flawed understanding, i.e., it misunder-
stands the entity or phenomenon and does not warn
the (human or another agent) user about its short-
coming(s). Failures in understanding have been first
elaborated by Ören and Yilmaz (2011).
Before starting to develop and implement misun-
derstanding avoidance algorithms, a systematic
approach to study causes of misunderstanding would
be very useful. This is the aim in this article.
As depicted in Figure 6, there are two main
groups of sources for an understanding system not to
function properly. They are inability to understand-
ing and filters causing misunderstanding.
Failures in understanding have been first elabo-
rated by Ören and Yilmaz (2011). In this article, we
further discriminate two sources of filters, namely
internal or self-imposed filters and externally im-
posed filters for context, biases, and fallacies. How-
ever, externally imposed filters are not elaborated
extensively. Furthermore, in this article, our basic
machine understanding paradigm has also been
extended with three other machine understanding
paradigms.
Inability to
understand
Filters causing
misunderstanding
due to:
external
Context
Meta-model
internal
Perception
Biases
Interpretation
Fallacies
Figure 6. Inabilities and filters that can induce misunder-
standing.
4.1 Inabilities to Understand Properly
Inabilities to understanding properly may depend on
the meta-model, perception, and interpretation of the
perception with respect to meta-model.
4.1.1 Misunderstanding Due to Meta-model
In the sequel, misunderstandings due to meta-model
are elaborated on for the basic understanding para-
digm, rich understanding, exploratory understand-
ing; and theory-based understanding as well as with
respect to the memories used in emotional intelli-
gence. Misunderstanding based on meta-model is
knowledge-deficient misunderstanding.
4.1.1.1 In Basic Understanding Paradigm
Misunderstanding due to meta-model may be one of
the four types:
not having necessary knowledge (uninformed
system),
not having necessary knowledge of proper resolu-
tion (superficially informed system) [superficial
understanding],
use of erroneous, incomplete, inconsistent, irrele-
vant, or corrupt meta-model (ill-
B: agent and its emotion
to be understood
Emotion understanding system A
Anal
y
zers
for emotions
for agents
D: perception of
B (by A)
with respect to
status of C
Memory mod-
ulator
Evaluator
(interprets the
relationship
between C &
D
)
Episodic memory
Semantic memo
r
y
S. graph
C:
meta-model of Bs
(meta-knowledge
about Bs)
Target action
U: an understanding of agent B and
its emotional state by A
with respect to C
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
322
informed/misinformed system) [ill-informed un-
derstanding] [misinformed understanding];
deliberately applying wrong meta-model (dogmat-
ic point of view). [This type of dogmatic under-
stand can be called meta-model induced dogmatic
understanding.]
4.1.1.2 In Rich Understanding Paradigm
Misunderstanding may be due to:
limited knowledge base (lack of additional meta
model(s) [knowledge-deficient misunderstanding],
inability to switch to other meta-model(s).
4.1.1.3 In Exploratory Understanding
Paradigm
Misunderstanding may be due to:
non existence of pertinent meta-model,
inability to find a pertinent meta-model,
inability to formulate needed hypotheses about
meta-models and to test them,
inability to adopt a different perspective [misun-
derstanding due to rigid perspective].
4.1.1.4 In Theory-based Understanding
Paradigm
Lack of understanding may be due to:
lack of appropriate theory or paradigm,
non acceptance of appropriate theory or paradigm
[theory-induced lack of understanding].
4.1.2 Misunderstanding Due to Perception
What cannot be perceived may normally not be
understood. Exception is the case of theory-based
understanding where theoretical knowledge precedes
experimental validation. Misunderstanding based on
perception is perception-deficient misunderstanding.
4.1.2.1 In Basic Understanding Paradigm
Some sources of problems for misunderstanding due
to perception (lack of perception, misperception)
are:
lack of appropriate ability to perceive [inability to
perceive],
inability to discriminate [perceptual confusion],
focus on an irrelevant aspect (domain, nature,
scope, granularity, modality) [irrelevant percep-
tion],
inability to discern goal(s) behind action(s) [super-
ficial perception],
hallucination in the absence of stimulus.
"The halo effect is a type of cognitive bias in
which our overall impression of a person influ-
ences how we feel and think about his or her char-
acter" (Cherry). Hence, halo effect may cause in-
appropriate and false perception; therefore may
cause misunderstand.
Perception component of an understanding system
should be able to discriminate deception [deception-
induced misunderstanding].
4.1.2.2 In Rich Understanding Paradigm
Misunderstanding may be due to:
inability to perceive reality from different perspec-
tives. [Tunnel vision understanding is only one
way to perceive and interpret; which is not the ap-
propriate way].
Hence, the following types of misunderstandings can
be distinguished: meta-model focused dogmatic
understanding, perception focused dogmatic under-
standing, and interpretation focused dogmatic un-
derstanding.
4.1.2.3 In Exploratory Understanding
Paradigm
Misunderstanding may be due to misperception
[misperception-induced misunderstanding].
4.1.2.4 In Theory-based Understanding
Paradigm
Misunderstanding may be due to instrumentation
error. An example is the claim made in early 2012
that "particles can travel faster than the speed of
light" as physicists operating the Large Hadron Col-
lider at the CERN laboratory claimed before detect-
ing a bad connection which invalidated the claim
[instrumentation-induced misunderstanding].
4.1.3 Misunderstanding
Due to Misinterpretation
Inappropriate pairing of meta-model(s) and percep-
tion(s) may lead to misunderstanding. Misinterpre-
tations may be done unwillingly or willingly. Mis-
understanding based on interpretation is
interpretation-deficient misunderstanding.
4.1.3.1 In Basic Understanding Paradigm
Misinterpretation is a source of misunderstanding
and may be due to:
lack of pertinent knowledge processing ability in
interpretation,
MachineUnderstandingandAvoidanceofMisunderstandinginAgent-directedSimulationandinEmotionalIntelligence
323
misinterpretation of motivation [misunderstanding
due to misinterpretation of motivation],
illusion which is a misinterpretation of a true sen-
sation, and
schizophrenic understanding which –as an aberra-
tion– leads to misinterpretations of reality.
4.1.3.2 In Rich Understanding Paradigm
Misunderstanding may be due to:
lack of appropriate meta-model(s) [knowledge-
deficient misunderstanding],
inability to access appropriate meta-model,
inability to use and to pair relevant perception and
relevant meta-model (misinterpretation).
4.1.3.3 In Exploratory Understanding
Paradigm
Misunderstanding may stem from the following
facts:
there is not yet an appropriate meta-model as a
basis for evaluation of the perception,
the granularities of the perception and the meta-
model may not match.
4.1.3.4 In Theory-based Understanding
Paradigm
Lack of interpretation or misinterpretation may be
due to the following facts:
theory was wrong and
technology is not yet ripe to observe with needed
precision.
4.1.4 Misunderstanding in Emotional
Intelligence
Emotions may have contradictory manifestations.
For example, the behaviour of an athlete crying after
winning a match, may be due to his emotional status
and distress while he is extremely joyful.
The contents and/or misinterpretations of the two
types of memories involved in emotional intelli-
gence can also be source of misunderstandings. For
example, strong past psychological experiences as
coded in the episodic memory may cause unbal-
anced behaviour.
The other causes of misunderstanding in emo-
tional intelligence are, as seen in section 4.1 (inabili-
ties to understand properly) and as discussed in
section 4.2 (filters affecting misunderstanding).
4.2 Filters Affecting Misunderstanding
Three types of filters such as context, biases, and
fallacies may affect understanding and cause misun-
derstanding. Filters can be internal or imposed ex-
ternally.
4.2.1 Context in Misunderstanding
Perception and/or interpretation in an improper con-
text can be source of misunderstanding [context-
induced misunderstanding]. Hence, one can identify:
context-sensitive understanding, context-insensitive
understanding, and double standards in understand-
ing. Context-dependent understanding would require
specification of the context. It would be desirable to
have context-aware understanding. The types of
misunderstandings are context unaware misunder-
standing and context-dependent misunderstanding.
4.2.2 Biases
Several types of biases such as group biases, cultural
biases, cognitive biases, emotive biases, personality
biases as well as effects of dysrationalia and irra-
tionality affect quality of understanding. Biases may
lead to biased understanding which may be errone-
ous understanding, incomplete understanding, in-
consistent understanding and irrelevant understand-
ing.
4.2.2.1 Group Bias in Misunderstanding
The group can be limited by a family, company,
institution, region, nation, interest, affinity, and/or
religion. A group member may have tunnel vision
which might affect understanding process [tunnel-
vision dogmatic understanding]. Sometimes mem-
bers may be instructed and even be indoctrinated
about a certain way of understanding. At extreme
cases, understanding can be blocked to lead to
blocked understanding.
4.2.2.2 Cultural Bias in Misunderstanding
Values and symbols differ for various cultures;
hence a same entity may be interpreted differently
based on the cultural background to lead culture-
induced misunderstanding.
4.2.2.3 Cognitive Bias in Misunderstanding
Cognitive bias is a "common tendency to acquire
and process information by filtering it through one's
own likes, dislikes, and experiences. [cognitive bias-
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
324
induced misunderstanding]
"Dunning-Kruger effect "those with limited
knowledge in a domain: (1) they reach mistaken
conclusions and make errors, but (2) their incompe-
tence robs them of the ability to realize it." "High
cognitive complexity individuals differ from low
cognitive complexity individuals not only in
knowledge processing abilities in general but in
understanding, in particular.
4.2.2.4 Emotive Bias in Misunderstanding
Certain types of emotions affect reasoning abilities
to cause misunderstanding [emotive bias-induced
misunderstanding, emotion-induced misunderstand-
ing]. For example, anger negatively affects reason-
ing, hence understanding ability. Effect of anger in
misunderstanding leads to anger-induced misunder-
standing. Joy may lead to euphoria which in turn
may affect understanding [joy-induced misunder-
standing].
4.2.2.5 Personality Bias in Misunderstanding
Some personality types are prone to anger; hence
their understanding ability can easily be affected to
lead misunderstanding [personality-induced misun-
derstanding].
4.2.2.6 Effects of Dysrationalia
in Misunderstanding
Dysrationalia is the inability to think and behave
rationally despite adequate intelligence (Stanovitch,
1993). It affects ability to understand properly [dys-
rationalia-induced misunder-standing].
4.2.2.7 Effects of Irrationality
in Misunderstanding
Irrationality may have two types of effects in mis-
understanding (Ariely, 2008) [irrationality-induced
misunderstanding]:
lack of ability to understand properly and
ability to distort understanding of others to cause
distorted understanding.
4.2.3 Fallacies in Misunderstanding
Fallacy is misconception resulting from incorrect
reasoning. A logical fallacy is an element of an ar-
gument that is flawed, essentially rendering invalid
the line of reasoning, if not the entire argument.
Fallacies in information distortion as well as delib-
erate misperception and misinterpretation are
sources of misunderstanding [fallacy-based misun-
derstanding]. They may exist as deliberate use of
unfit metamodel in understanding. Two categories of
fallacies are paralogism and sophism.
4.2.3.1 Paralogism in Misunderstanding
Paralogism is unintentional use of invalid argument
in reasoning. It causes misunderstanding due to
misperception, mis-interpretation, and/or mis-
justification of background knowledge (meta-model)
[paralogism-based misunderstanding].
4.2.3.2 Sophism in Misunderstanding
Sophism is deliberately using invalid argument
displaying ingenuity in reasoning in the hope of
deceiving someone. Some recent techniques in lie
detection in text analysis can also be used to detect
sources of attempt to misguide in understanding.
Misunderstandings due to fallacies can be delib-
erate misunderstanding (giving the illusion of not
understanding) and induced misunderstanding. Data
and evidences may be tempered or doctored by the
entity which attempts to understand and/or by an
outside agent. The individuals (or their representa-
tives, such as software agents) need to notice that
their understanding is being tempered [doctored or
tempered-evidence-based misunderstanding]. Hen-
ce, recognizing why a reality is presented in a cer-
tain way is helpful not to be trapped in misunder-
standing.
A type of misunderstanding is mutual misunder-
standing. Avoiding mutual misunderstanding is very
important to find reconciliatory solutions at different
levels of relationships.
4.3 Documentation of Understanding
It would be very desirable for an understanding
system to be able to document its abilities and limi-
tations. In this way, a user (human or another agent)
can have an informed trust to the results of an under-
standing system. Based on the systematization used
in this article, this type of documentation may in-
clude the following:
Meta-model(s) available and used
Perception(s)
Interpretation(s)
Contents of episodic and semantic memories
Types and contents of filters used.
A challenging situation in understanding is the case
when an understanding system does not have any
knowledge (or meta-model) about the entities it is
MachineUnderstandingandAvoidanceofMisunderstandinginAgent-directedSimulationandinEmotionalIntelligence
325
asked or required to understand. In this case, the
system would need to search and get appropriate
background knowledge and meta-model(s) and/or be
able to formulate and test hypotheses to formulate a
meta-model.
5 CONCLUSIONS
This article is a sequel to our joint work on multi-
understanding especially applied to understand hu-
man behavior and failure avoidance in simulation
studies. On understanding, we expanded our basic
multi-understanding paradigm and continue to sys-
tematize our exploration of sources of misunder-
standing.
We plan to implement some cases of misunder-
standing to avoid misunderstanding in agent simula-
tion of human behavior and especially in emotional
intelligence simulation.
Another line of research we plan to continue is to
realize context-aware agents for advanced simula-
tion studies. Context aware agents may also be use-
ful in other applications.
In both cases, we will attempt to develop soft-
ware agents capable to attest their limits of under-
standing by generating proper detailed documenta-
tion of their limits of understanding.
For human misunderstanding, the books by
Heyman (2012) and Young (1999) may be useful. In
addition to them, the book by Herman and Chomsky
(1988) would be useful for external distortions of
understandings [distortion-induced misunderstand-
ing].
REFERENCES
Ariely, D., 2008. Predictably irrational: The hidden forces
that shape our decisions. Harper Collins, 2
nd
edition in
2012.
Cherry, K. What is the halo effect? About.com.
http://psychology.about.com/od/socialpsychology/f/ha
lo-effect.htm
Dias, J., Paiva, A., 2009. Agents with emotional
intelligence for storytelling. 8th International
Conference on Autonomous Agents and Multi-agent
Systems, Doctoral Mentoring Program Budapest,
Hungary.
Herman, E., Chomsky, N., 1988. Manufacturing consent –
A propaganda model, Pantheon Books, New York.
Heyman, D., 2012. The Little Book of Big Misunderstand-
ings, AllPolitics Communications Inc. at Smashwords.
Kazemifard, M., Ghasem-Aghaee, N., Ören, T. I., 2009.
Agents with ability to understand emotions. In:
Summer Computer Simulation Conference, 2009,
Istanbul, Turkey. SCS, San Diego, CA, 254-260.
Kazemifard, M., Ghasem-Aghaee, N., Ören, T. I., 2012.
Emotive and cognitive simulations by agents: Roles of
three levels of information processing. Cognitive
Systems Research, 13, 24-38.
Kazemifard, M., Ghasem-Aghaee, N., Koenig, B. L.,
Ören, T. I., 2013. An emotion understanding
framework for intelligent agents based on episodic and
semantic memories. Journal of Autonomous Agents
and Multi-Agent Systems.
(DOI) 10.1007/s10458-012-9214-9.
Mayer, J. D., Salovey, P., 1997. What is emotional
intelligence. In: Salovey, P., Sluyter, D. (eds.)
Emotional Development and Emotional Intelligence:
Educational Implications. New York: Basic Books.
Ören, T. I., 1992. Towards program understanding
systems. (Invited Plenary Paper). In: Proceedings of
the First Turkish Symposium on Artificial Intelligence
and Neural Networks, Oflazer, K., Akman, V.,
Guvenir, H. A., Halici, U., (eds.), 25-26 June, 1992,
Bilkent University, Ankara, Turkey, pp. 3-12.
Ören, T. I., 1995. (Invited contribution). Artificial intelli-
gence and simulation: A typology. In: Proceedings of
the 3rd Conference on Computer Simulation, Raczyn-
ski, S., (ed.), Mexico City, Mexico, Nov. 15-17, pp. 1-
5.
Ören, T. I., 2000. (Invited Opening Paper). Understanding:
A taxonomy and performance factors. In: Thiel, D.,
(ed.) Proc. of FOODSIM’2000, June 26-27, 2000,
Nantes, France. SCS, San Diego, CA, pp. 3-10.
Ören, T. I., 2012. Evolution of the discontinuity concept in
modeling and simulation: From original idea to model
switching, switchable understanding, and beyond.
Special Issue of Simulation –The Transactions of SCS.
Vol. 88, issue 9, Sept. 2012, pp. 1072-1079. DOI:
10.1177/0037549712458013.
Ören, T. I., Abou-Rabia, O., King, D. G., Birta, L. G.,
Wendt, R. N., 1990. (Plenary Paper). Reverse
engineering in simulation program understanding. In:
Problem Solving by Simulation, Jávor, A., (ed.)
Proceedings of IMACS European Simulation Meeting,
Esztergom, Hungary, August 28-30, 1990, pp. 35-41.
Ören, T. I., Ghasem-Aghaee, N., Yilmaz, L., 2007. An
ontology-based dictionary of understanding as a basis
for software agents with understanding abilities. In:
Proceedings of the Agent-Directed Simulation Sympo-
sium of the Spring Simulation Multiconference
(SMC'07), pp. 19-27. Norfolk, VA, March 2007.
Ören, T. I.,.Yilmaz, L., 2009. Failure avoidance in Agent-
Directed Simulation: Beyond conventional V&V and
QA. In L. Yilmaz and T.I. Ören (eds.). Agent-Directed
Simulation and Systems Engineering. Systems
Engineering Series, Wiley-Berlin, Germany, pp. 189-217.
Ören, T. I., Yilmaz, L., 2011. Semantic agents with
understanding abilities and factors affecting
misunderstanding. In: Elci, A., Traore, M. T., Orgun,
M. A. (eds.)
Semantic Agent Systems: Foundations
and Applications, Springer-Verlag. pp. 295-313.
SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand
Applications
326
Stanovich, K. E., 1993. Dysrationalia: A new specific
learning disability. Journal of Learning Disabilities,
26(8), 501–515.
Yilmaz L., Ören, T. I., 2009. Agent-directed simulation,
Chapter 4 in Yilmaz, L., Ören, T. I. (eds.) Agent-
directed Simulation and Systems Engineering, pp.
111-144. Wiley Series in Systems Engineering and
Management. Wiley.
Young, R. L., 1999. Understanding Misunderstand-ings:
A Guide to More Successful Human Interac-tion. Uni-
versity of Texas Press, Austin, TX.
Zeigler, B. P., 1986. Systems knowledge: A definition and
its implications. In: Modelling and Simula-tion Meth-
odology in the Artificial Intelligence Era, Elzas, M.S.,
Ören, T. I., Zeigler, B. P. (eds.), North-Holland, Am-
sterdam, pp. 15-17.
MachineUnderstandingandAvoidanceofMisunderstandinginAgent-directedSimulationandinEmotionalIntelligence
327
