SOCIAL BEHAVIOR INVESTIGATION OF AN INTELLIGENT

VIRTUAL AGENT WITH THE HELP OF TYPICAL WORKING

STUDENT’S LIFE SCENARIO MODELING

Dilyana Budakova

and Lyudmil Dakovski

Technical University of Sofia, Branch Plovdiv, 61, Sankt Petersburg Blvd., Plovdiv, Bulgaria

European Polytechnical University of Pernik, Pernik, Bulgaria

Keywords: Intelligent Virtual Agents (IVA), Architecture, Social behavior, Social scenario, Social emotion, Primary

and secondary emotions, Needs, Rationalities, Knowledge, Ambivalence, Choice, Deception.

Abstract: This paper investigates the social behaviour of an intelligent virtual agent (IVA) with PRE-ThINK

architecture with the help of typical working student’s life scenario modeling. Тhe program system and the

PRE-ThINK architecture, adapted for this scenario, are proposed, and their components are considered. The

dynamics of the decision making process in problem situations caused by the implementation of this

architecture is shown, when mixed emotions arise and the realization of what happened reflects on the

agent’s temper. IVA’s social behavior is shown, during which in the process of communicating with the

user the agent expresses learned from experience secondary emotions, which can be either in harmony or in

conflict with the realized secondary emotions, resulting both from the agent’s generalized condition and the

events. The investigated secondary emotions are: relief, confidence, prestige, uncertainty, confirmed fear,

disappointment; and also the socially expressed secondary emotions such as refrained sadness, refrained

anger, businesslike manners, politeness and authoritativeness.

1 INTRODUCTION

In general, social norms are commonly believed

rules in social interaction. These rules serve as a

guide for human behavior, and as the basis for their

beliefs and expectations about others. Without them,

communication can break down easily (Si Mei,

Stacy C. Marsella, and David V. Pynadath 2006).

Social interaction is accompanied by social

emotions. Social emotions can be defined as one’s

emotions projecting on or affected by others. They

are “valenced reactions”, which can occur as a result

of how people understand events. The first

modification to the OCC model, (that define a

cognitive approach for looking at emotions) is to

allow for the definition of social emotions (Ortonty,

A., Clore, G.L., Collins 1988).

(Damasio 1994) distinguishes at least two

classes of emotions, namely, primary and secondary

emotions, on the basis of his neurobiological

findings. According to them secondary emotions are

more dependent on the agent’s cognitive reasoning

abilities. The appraisal of secondary emotions

depends much more on the situational context and

an agent’s memory than that of primary emotions.

The releasers of secondary emotions might be

learned based on the history of primary emotions in

connection with memories of events, agents, and

objects. Secondary emotions also modulate the

agent’s simulated embodiment, such as its general

level of arousal. The agent expresses its awareness

of secondary emotions verbally. Secondary emotions

are based on more complex data structures than

primary ones.

Recently a number of investigators distinguish

between primary and secondary emotions,

underlining the importance of their modeling and

studying (André, E., Klesen, M., Gebhard, P., Allen,

S.,&Rist, T. (1999); Pelachaud, C., & Bilvi, M.

(2003).; Becker-Asano, C., & Wachsmuth (2008)).

(Rehm Matthias and Elisabeth Andre 2005)

focus on synthetic agents that may express emotions

that are in conflict with their appraisal processes and

they put a question: how to handle situations in

which the agent decides to display an emotion which

is in conflict with its internal appraisal processes in

social settings.

317

Budakova D. and Dakovski L..

SOCIAL BEHAVIOR INVESTIGATION OF AN INTELLIGENT VIRTUAL AGENT WITH THE HELP OF TYPICAL WORKING STUDENT’S LIFE

SCENARIO MODELING.

DOI: 10.5220/0003732103170324

In Proceedings of the 4th International Conference on Agents and Artiﬁcial Intelligence (ICAART-2012), pages 317-324

ISBN: 978-989-8425-96-6

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

Acording to (Benny Ping-Han Lee et.al. 2006)

the emotion models and architectures for virtual

agents are not yet advanced enough to be imbued

with coexisting emotions but this is very important

because mixed emotions, especially those in

conflict, sway agent decisions and result in dramatic

changes in social scenarios.

In order to contribute to further development of

the described problems, this paper investigates the

social behavior of an intelligent virtual agent (IVA)

with PRE-ThINK architecture by means of typical

working student’s life scenario modeling. Тhe

program system and the PRE-ThINK architecture,

adapted for this scenario, are proposed. The basic

modules of the programming system and the

implementation of this architecture are considered.

The mechanism of arising of the agent’s thoughts, as

well as their assessment, are cleared out. The

dynamics of the decision making process in problem

situations is shown, when mixed emotions arise and

the IVA’s behavior changes in a way, allowing

achieving all of the goals of the scenario,.

The social behavior, shown by the agent, as well

as his/her generalized condition of awareness, which

are not always in harmony, are explained and

presented. The proposed scenario allows for

modeling and expressing of both primary and

secondary emotions. The social emotions, shown by

the IVA, when giving answers to user’s questions in

the experimental on-line information site, are a

mixture of primary (spontaneously caused by the

current context emotions) and secondary emotions

(demonstrated - learned from IVA’s experience in a

similar situation; and realized – caused by the IVA’s

self-esteem related to his/her generalized condition

and by all the simultaneous events, hidden for the

users).

The rest of this paper is organized as follows:

Section 2 considers similar developments; the third

section presents the scenario for the goals of the

experiment; the fourth section comprises the

description of the programming system and the

model of IVA with PRE-ThINK architecture,

adapted for typical working student’s life scenario

modeling. Section 5 gives description of the

experimental results. The sixth section includes

discussion and directions for further development of

the programming system; a generalization of the

present work is proposed at the end.

2 BACKGROUND

There is a considerable body of work on social

norms and norms in conversations in particular,

including formalization of norms and obligations

(Boella, G., Torre, L.v.d. 2003), how norms emerge,

spread and get enforced in a society (Castelfranchi,

C. 1995), levels of cooperation in social

communications (Airenti, G., Bara, B. G.,

Colombetti, M. 1993), discourse obligations in

dialogues (Traum, D. R., Allen, J. F. 1994) etc.

In Thespian's decision-theoretic framework (Si

Mei, Marsella, Pynadath 2006), conversational

norms enable the characters to behave human-like

by making relevant responses, following natural

turn-taking patterns, and having appropriate

conversational flow. (deRosis Fiorella, C. Pelachaud

et. al. 2003) as well as Prendinger and colleagues

(Prendinger Helmut and Mitsuru Ishizuka 2001)

developed agents that are able to control their

emotional displays if the social situation requires it.

For instance, if the social distance between an agent

and its conversational partner is high, Prendinger’s

agent would not show anger to the full extent.

The virtual tutor COSMO (Lester J. C. et. al.

2000) intentionally portrays emotions with the goal

of motivating students and thus increasing the

learning effect.

In the WASABI architecture (Becker-Asano, C.,

Wachsmuth, I.) use their agent’s cognitive reasoning

abilities to model the mood-congruent elicitation of

secondary emotions as well. They explain how nine

primary emotions together with three secondary

emotions—namely the prospect-based emotions

hope, fears-confirmed, and relief —were integrated

in such a way that their mood congruent elicitation

can be guaranteed. They discussing results of a study

on the effect of secondary emotion simulation in a

card game scenario.

Many researchers model IVA behaviour aiming

at establishing a trust-based relationship between the

user and the IVA (Celso M. de Melo et. al. 2009,

Gratch et. al 2007, Bickmore et. al.2007,

Niewiadomski et. al. 2008).

In the present paper the agent with PRE-ThINK

architecture proposed in (Budakova D. 2011)

(abbreviation from the initial letters of the basic

components making it up - Principles, Rationalities,

Emotions, Thoughts, Investigations, Needs, and

Knowledge) is used to investigate social behavior in

typical working student’s life scenario.

3 THE SCENARIO

For the purposes of the experiment the behavior of

the typical working student is modeled, that have to

ICAART 2012 - International Conference on Agents and Artificial Intelligence

318

follows the polite form of social communication at

work; fulfills diligently its duties at work; manages

to pass all its exams; and manages to attend all

meetings and parties, organized by its friends.

According to this particular scenario the IVA is

modeled to be an attractive 20-year-old blond

student, giving information to the users of an

experimental, especially developed for the

experiment site, in the role of an on-line receptionist.

The users can only observe the IVA’s behavior

in its role of an on-line receptionist. The other

events, related to its modeled personal identity in

accordance with the modeled scenario (student’s

life, sleeping time, party time) as well as the course

of time, remain hidden for the users of the site and

are only simulated in a programming way.

The new aspect in the proposed paper is that a

situation is modeled, in which the hidden/refrained

emotions are caused by events that are outside the

current context and not related to the current events,

while the shown emotions are related to the current

context but they are well considered and result from

personal experience as well as from studying of

social standards.

The real secondary emotions of the agent, on the

one hand, could be: confidence, satisfaction with the

achievements and self-belief. They result from well

fulfilled tasks, passed exams, well spent time at

parties with friends, high ranking by the users of the

site, and they evoke at first happiness, relief and

pleasure and after their rationalization they turn in

the course of time into self-confidence, certainty and

authoritativeness.

This can be expressed by politeness, energy,

friendliness, readiness to help and understand the

others, willingness to communicate and join both the

others and various good causes.

On the other hand, these emotions could also be:

inferiority complex, disappointment and depression,

lack of self-belief, what is caused by failures at the

exams, unfulfilled tasks at work, lowered ranking by

the users, and they evoke at first fear and

anticipation of failures, as well as confirmed fear

and after their rationalization in the course of time

they turn into disappointment and inferiority

complex. They could be expressed by irritability,

aggression, dictatorial behavior or shyness, sadness,

depression and self-isolation from the others.

In order to reach all her goals, the IVA plans her

time for studying, working, sleeping and meetings

with friends, so that their optimal apportionment is

achieved. The scenario allows observing how in the

course of time, day by day, each deadline is met and

the IVA achieves this perfect time apportionment.

4 THE PROGRAM SYSTEM

DESCRIPTION

The basic modules of the program system are:

1) interface module; 2) module for simulating the

course of time – days and hours; 3) module for

decision making and defining the agent’s actions by

means of the PRE-ThINK architecture according to

her goals, principles, priorities and the assessment of

the generated thoughts; 4) module for initialization

of the scenario, comprising the relation need –

action, deadlines for achieving goals, related to the

different needs and emotions, necessary number of

working hours for successful fulfilling of a task,

hours for passing an exam and for normal attendance

of parties, as well as the agent’s luck and abilities

(fig.1).

Figure 1: Structural scheme of the program system.

Database

Time simulation.

Module for initialization of the

scenario. Schedule of the events,

deadlines and requirements for

achieving the assigned goals.

Module for realization of the scenario and

for modeling the IVA with the PRE-

ThINK architecture. IVA’s behavior,

decisions making for achieving the goals

according to the agent’s principles, priorities,

appraisals and experience.

Interface with the user

Experimental web site, where the IVA is a

receptionist.

Demonstrated social behavior.

SOCIAL BEHAVIOR INVESTIGATION OF AN INTELLIGENT VIRTUAL AGENT WITH THE HELP OF TYPICAL

WORKING STUDENT'S LIFE SCENARIO MODELING

319

4.1 The Interface Module

The experimental site is divided into three frames.

The first one is for choosing the topic in which the

user is interested; the second – for modeling the

IVA’s behavior, for initialization of the scenario, as

well as for simulating the course of time; and the

third is for visualization of the agent’s head and

realization of the agent’s multimodal social

behavior.

The following relation is realized: a chosen

question and a corresponding answer, generalized

condition of the agent, temper, temperament, self-

confidence, emotional but socially appropriate

pronunciation of the answer by the IVA.

The user can see and ask the IVA questions by

means of the site only. The rest of events and

experiences from the predefined scenario remain

hidden from him/her.

4.2 The Module for Initialization of the

Scenario

The relation need – action is preset as it follows:

physiological need – need of sleeping; need of safety

– working position as an on-line receptionist; need

of love and belonging – attendance at parties with

friends; need of self-actualization – university

course (fig.2).

Figure 2: Arrangement of the fulfilled and unfulfilled

needs in a crisis situation and their weight together with

the correspondence of the relation need-action according

to the considered scenario.

According to Maslow's theory the needs in

normal human development are arranged as follows:

physiological (ph), safety (s), love and belonging

(lb), esteem and self-assessment (es), self-

actualization (sa), aesthetics (a). In the architecture

considered here the needs are associated with

weights Wneed corresponding to their priority: Wph,

Ws, Wlb, Wes, Wsa, Wa.

When,

because of the occurrence of an event,

one or more needs prove to be unfulfilled, i.e., there

is a crisis situation, then the needs rearrange so that

the unfulfilled ones receive first priority. The

unfulfilled needs arrange in an order, opposite to the

order of needs weights in a normal state of the agent.

The unfulfilled needs are denoted by W

phu

, W

lbu

, W

esu

, W

. This concept is illustrated in

fig.2 where the predetermined (according to the

scenario) simplified relations of the type need –

action are denominated both for a normal and for a

crisis situation.

The dates of the exams, the dates of fulfilling the

working tasks and the dates of the parties are

predetermined. The hours necessary for the agent to

get ready for the exams, the hours for a task to be

fulfilled successfully, the sleeping hours and the

hours to enjoy a party are also predefined.

The IVA’s luck and abilities are predetermined

as well and they can facilitate or slow down the

fulfillment of a task, they can lead to a failure at an

exam, they can ruin one’s sleeping or his/her time at

a party. The thresholds of dissatisfaction for each

task, related to a particular agent’s activity are also

predefined.

An initial schedule for realization of the IVA’s

goals is predetermined. It defines the initial

apportionment of the hours in the agent’s day and

night. This is the exact initial apportionment of the

time which changes in the course of the events in the

agent’s life and their optimum arrangement is a goal

of the algorithm for decisions making by the IVA.

The IVA’s aspiration/goal is to succeed in each of

the directions, to be in good spirits and have self-

confidence.

The course of time is simulated by means of the

component Timer. Every hour is counted out and the

sequence of the morning, the noon, the evening and

the night is controlled; it is also controlled what is

the IVA supposed to do according to the initial

schedule in each particular moment; the real agent’s

behavior and actions in a moment are also controlled

and the activities from the schedule and the real

behavior according to the decision taken are

compared. All the agent’s activities are saved in the

database. The IVA knows how many tasks remain to

be fulfilled, how many his successes and failures

are, to which activities they are related.

4.3 Modelling of the IVA with the

PRE-ThINK Architecture and the

Process of Decision Making for

Defining the IVA’s Behavior

The IVA takes its decisions based on its principles.

esu

lbu

phu

10 20 30 40 50 60 70 80 90 100 110 120

ICAART 2012 - International Conference on Agents and Artificial Intelligence

320

Required hours for studying in order to pass the

exams and sleeping hours needed from date to date.

Required hours for studying for successful fulfilling

the working tasks and sleeping hours needed from

date to date.

Fig. 3a) Fig. 4a)

Really used hours according to the IVA’s schedule for

every exam preparation as well for sleeping from date

to date. The sleeping hours are less when the IVA has

been to parties during the considered time interval.

Really used hours according to the IVA’s schedule

for working tasks fulfillment as well as for sleeping

from date to date. The sleeping hours are less when

the IVA has been to parties during the considered

time interval. interval.

Fig. 3b) Fig. 4b)

Hours used for studying and sleeping with the use of

the PRE- ThINK architecture mechanisms.

Hours used for working and sleeping with the use of

the PRE- ThINK architecture mechanisms.

Fig. 3c) Fig. 4c)

Figure 3 and Figure 4: Experimental results.

The following IVA principles have been modelled:

“Choose the better possible action"; "Neglect the

basic needs until reaching a definite threshold of

dissatisfaction, giving priority to the highest-order

needs”,

“Follow a polite form of social behavior

regardless of your personal successes or problems”.

The IVA starts following her initial schedule and

realizes whether she manages to cope with the

SOCIAL BEHAVIOR INVESTIGATION OF AN INTELLIGENT VIRTUAL AGENT WITH THE HELP OF TYPICAL

WORKING STUDENT'S LIFE SCENARIO MODELING

321

assigned tasks or not. She also realizes how many

tasks from each activity remain to be fulfilled, e.g.,

how many exams to pass, how many working tasks

etc. According to the principles if the IVA does not

pass her exams, for example, she will try to

reallocate more hours for studying at the expense of

the sleeping hours at first; if this does not help, then

the new arrangement of hours is done at the expense

of the working hours; and, last, at the expense of the

time spent with friends. If all of these measures do

not help and her physiological need – her needs of

safety and friendship are seriously threatened, then,

according to her principles, realizing Maslow’s

theory and according to the predetermined at the

initialization very simple relation need-action, the

IVA will stop her studies in order to redress the

balance of the more basic needs, sacrificed in the

name of her assigned highest need/goal in this

scenario.

Each thought of the agent is related to an

emotional (positive (+) or negative (-) sign), rational

and need value and receive the weight of this need.

Each thought is addressed to the situation s be

denoted by Th_s. If the importance of the thought

Th_s is denoted by I

Th_s

, the weight of the need,

related to this thought is expressed by W

needTh_s

, , the

emotion implied by this thought is marked by

emot.Th_s

, then, following the formulae for

calculating the assessment value of the thought O

Th_s

corresponding to the situation s will be (Budakova

D., 2011):

sThsneedThsThemotsTh

IWEO

___._

**=

If a thought is partially related to more than one

need then the sum of the weight percentages of the

needs to which it is related is taken into account in

the formulae.

Each thought is related to an action. The

assessment values of the thoughts related to one and

the same action in one and the same situation are put

on the one basin of the “thoughts balance”. The

assessment values of the thoughts for the same

situation, but related to another action, are put on

another basin etc. Our “thoughts balance” will have

as many basins as the alternative actions considered

by the agent in the particular situation are.

(Budakova D., 2011).

The described mechanism for decision making is

adapted for the purposes of the considered scenario.

Instead of choosing the action from the basin having

the highest assessment value, the action with the

lowest ranking is sought for here, i.e., the less

important for the moment. The hours used for it will

be reduced in order to spare time for other more

important activities and, respectively, for achieving

other dissatisfied goals.

4.4 Mechanism of Origination and

Assessment of the IVA’s Thoughts

The IVA’s thoughts are generated in 24 hour each

clock round and they are related to the events from

the agent’s schedule, to her generalized condition, to

the successful or unsuccessful fulfillment of the

goals, as well as to the possibilities for action. Each

thought, related to achieving a goal/action, is

assessed from emotional, rational and need-related

point of view.

I can increase my hours for studying at the

expense of reducing the sleeping time. Emotion (+);

weight of the need for sleeping 10; rational

assessment – easy and safe 1; general assessment of

the thought related to this action (+10).

I can increase my hours for studying at the

expense of reducing the working hours. Emotion

(+); weight of the need for work 20; rational

assessment – easy and safe 2; general assessment of

the thought related to this action (+40).

I can increase my hours for studying at the

expense of reducing the hours for meeting friends.

Emotion (+); weight of the need for meeting friends

30; rational assessment – easy and safe 2; general

assessment of the thought related to this action

(+60).

In addition to choosing an action for improving

her generalized condition, the agent also generates

thoughts about her social behavior at her working

place.

5 THE EXPERIMENT

The first experiment illustrates the consequences for

the IVA if she follows the schedule without changes.

They are given in figure 3b for the exams and figure

4b for the working tasks. Having the graphs 3a and

3b compared, it can be seen that the IVA studies less

than needed and correspondingly she fails at the

exams. When comparing the graphs 4a and 4b it can

be seen that the working time turns out to be

sufficient for fulfilling the tasks and duties perfectly.

The sleeping hours are also sufficient and from the

graphs 3b and 4b it can be seen that the IVA does

not miss a party. The second experience shows how

the agent tries to cope with her problems at the

university by increasing at first the hours for

studying at the expense of the sleeping hours and

ICAART 2012 - International Conference on Agents and Artificial Intelligence

322

then, at the expense of the working hours (fig. 3c и

fig. 4c).

The results from the graphs 3a and 3c should be

compared here to see whether the actually used

hours are sufficient. The results from the graphs 4a

and 4c should also be compared to see if the actual

hours for coping with the working tasks are

sufficient to lead to success.

It can be seen from the graphs that, after she

fails at her first exam, the IVA increases the time for

studying at the expense of the sleeping time. After

the second failure she reduces the sleeping time even

more in order to study harder. After the third failure

she reduces the working time, too, and this is how

she manages to pass the fourth and the fifth exams.

The success turns out to be sufficient for bringing

back her happiness and self-confidence.

6 CONCLUSIONS

The paper investigates the social behavior of an IVA

with PRE-ThINK architecture. A scenario is

proposed, allowing for clearing out and revealing the

following: the agent’s decision making mechanism;

the mechanism of origination and assessing

thoughts, related to the alternative actions for each

situation; the mechanism for achieving optimal time

apportionment in order to achieve the best possible

coping with the assigned tasks. The origination and

the influence of the primary emotions such as

happiness and sadness is also cleared out for

building and investigating the secondary emotions

such as relief, confidence, prestige, uncertainty,

confirmed fear, disappointment. The peculiarities of

the program system, as well as of the IVA’s PRE-

ThINK architecture together with the conducted

experiments are described and explained.

New principles and behavioral rules can be

formed based on the accumulation of observations

and their generalization. This is realized by means

of the subprograms of the architecture component

named Investigations.

Creation of a more complex situation is

envisaged, in which the IVA’s goals will be related

to different needs to a different extent. The decision

making will be more complicated and more realistic

then.

Tracking the way the IVA gains experience in

communication with the users of the site will be a

subject of another investigation together with the

way she learns to be polite and businesslike

regardless of her problems, successes and failures.

It is assumed that, together with the further

development of the system, it could be possible to

realize a backward investigation, i.e., to

predetermine the decisions, taken in a problematic

situation in order to build a model of the investigated

person.

The investigation of a collaborative or

competitive behavior in a multi-agent system for

similar scenarios would be also of great interest.

Another direction for further development is to

investigate how the various ways of communication

with users would influence the ranking of the IVA.

Is it better for her to be more businesslike if the user

is interested in what is going on when he has

understood from the IVA’s multimodal behavior that

something is going wrong or that she is extremely

happy?

REFERENCES

Airenti, G., Bara, B.G., Colombetti, M.: Conversation and

bahavior games in the pragmatics of dialogue. In:

Cognitive Science. Volume 17(2). (1993) 197-256-48

André, E.,Klesen,M., Gebhard, P., Allen, S.,&Rist, T.

(1999). Integratingmodels of personality and emotions

into lifelike characters. In international workshop on

affect in interactions—Towards a new generation of

interfaces (pp. 136–149). Springer.

Becker-Asano, C., & Wachsmuth, I. Affect simulation

with primary and secondary emotions. Intelligent

virtual agents (IVA 08) (pp. 15–28). Springer, LNAI

5208. (2008).

Boella, G., Torre, L.v.d.: Obligations as social constructs.

In: Proc. of the Italian Conf. on Arti_cial Intelligence

(AI*IA'03). (2003) 27-38

Budakova D., Behavior of Home Care Intelligent Virtual

Agent with PRE-ThINK Architecture, pp.157-167,

ICAART’2011, Rome, Italy.

Castelfranchi, C.: Commitments: From individual

intentions to groups and organizations. In: ICMAS.

(1995) 41-48

Celso M. de Melo, Liang Zheng and Jonathan Gratch1,

Expression of Moral Emotions in Cooperating

Agents*, in IVA 2009, Sep 14-16, Amsterdam

Damasio, A. (1994). Descartes’ error, emotion reason and

the human brain. Grosset/Putnam.

deRosis F., C. Pelachaud, Isabella Poggi,Valeria

Carofiglio, and Berardina De Carolis. From Greta’s

mind to her gace: modelling the dynamics of affective

states in a conversational embodied agent.

International Journal of Human-Computer Studies,

59:81–118, 2003.

Gebhard P., ALMA – A Layered Model of Affect,

AAMAS'05, July 25-29, 2005, Utrecht, Netherlands.

SOCIAL BEHAVIOR INVESTIGATION OF AN INTELLIGENT VIRTUAL AGENT WITH THE HELP OF TYPICAL

WORKING STUDENT'S LIFE SCENARIO MODELING

323

Goleman D. Emotional Intelligence. Bantam Books, New

York, 1995.

Gratch J. G., N. Wang, J. Gerten, E. Fast, and Robin

Duffy, Creating Rapport with Virtual Agents,

Intelligent Virtual Agents, Paris, France 2007

Lee Benny Ping-Han, Edward Chao-Chun Kao, Von-Wun

Soo, Feeling Ambivalent: A Model of Mixed

Emotions for Virtual Agents, IVA 2006, LNAI 4133,

pp. 329-342, Springer., 2006

Lester J. C., S. G. Towns, C. B. Callaway, J. L. Voerman,

and P. J. FitzGerald. Deictic and emotive

communication in animated pedagogical agents. In

Embodied Conversational Agents. MIT Press, 2000.

Maslow A. H., Motivation and personality, Published by

arrangement with Addison Wesley Longman, Inc.

USA. 1970.

Morris W. N. Mood: The frame of mind. New York:

Springer-Verlag, 1989.

McCrae R.R. and John O.P.. An introduction to the

fivefactor model and its implications. Journal of

Personality, vol. 60, 1992, 171–215.

Niewiadomski R., Magalie Ochs, and Catherine

Pelachaud, Expressions of empathy in ECAs,

IVA’2008, Tokyo, pp. 37-44, LNAI, Vol. 5208, 2008

Ortonty, A., Clore, G.L., Collins, A.: The Cognitive

Structure of Emotions. Cambridge University Press

(1988).

Pelachaud, C., & Bilvi, M. (2003). Computational model

of believable conversational agents. In M.-P. Huget

(Ed.), Communications in multiagent systems.

Springer-Verlag.

Prendinger H. and M. Ishizuka. Social Role Awareness in

Animated Agents. In Proceedings of Agents ’01,

Montreal, Canada, pages 270–277, 2001.

Rehm M. E. Andre, Catch Me If You Can — Exploring

Lying Agents in Social Settings AAMAS’05, July

2529, 2005, Utrecht, Netherlands.

Si Mei, Stacy C. Marsella, and David V. Pynadath,

Thespian: Modeling Socially Normative Behavior in a

Decision-Theoretic Framework, IVA’06, 2006.

Traum, D.R., Allen, J.F.: Discourse obligations in

dialogue processing. In: ACL. (1994) 1-8

ICAART 2012 - International Conference on Agents and Artificial Intelligence

324