Modeling Personality in the Affective Agent Architecture GenIA

Joaquin Taverner, Bexy Alfonso, Emilio Vivancos and Vicente Botti

Departamento de Sistemas Inform

aticos y Computaci

on, Universitat Polit

ecnica de Val

encia, Spain

Keywords:

Jason, Agent, Emotion, Personality.

Abstract:

In the last few years there has been a growing interest in affective computing. This type of computation tries

to include and use emotions in different software processes. One of the most relevant areas is the simulation

of human behavior where various affective models are used to represent different affective characteristics such

as emotions, mood, or personality. Personality is deﬁned as a set of individual characteristics that inﬂuence

motivations and behaviors when a human being faces a particular circumstance. Personality plays a very

important role in modeling affective processes. Through the simulation of emotions we can improve, among

others, the experience of users dealing with machines, and human simulations in decision-making processes

using multi-agent systems. In this work we propose a model for the use of personality in the general purpose

architecture for affective agents GenIA

, as well as the development of the model in the current GenIA

platform.

1 INTRODUCTION

When we analyze human behavior from a rational

point of view, we observe inconsistencies when hu-

man beings face a particular situations like a decision-

making problem. These inconsistencies are due to

the inﬂuence of emotions and affective characteris-

tics. These characteristics inﬂuence the decision mak-

ing and reasoning processes to a greater or lesser ex-

tent (Broome, 2002). Researchers in this area ana-

lyze the inﬂuence of emotions, mood and personal-

ity (Zelenski, 2007; Frijda, 1986). It is generally ac-

cepted that emotions are reactions to a certain stim-

uli (Ortony et al., 1990). Emotions are divided into

two types: primary emotions, that have a direct re-

lationship with expressive abilities such as facial ex-

pressions, body postures or voice inﬂections; and sec-

ondary emotions, that arise as the result of reasoning

about current events taking into account expectations

and past experiences. On the other hand mood is dif-

ferent from emotions or feelings: it is less speciﬁc,

less intense, and less likely to be triggered by a par-

ticular stimulus or event. In addition, mood is less

volatile than emotions (i.e. mood vary less over time

and have a longer duration) (Becker, 2001). Finally

personality is deﬁned as a set of individual character-

istics that inﬂuence motivations, behaviors, and emo-

tions when facing a particular circumstance (Damasio

and Sutherland, 1994).

Personality also refers to the characteristic way

in which a person thinks, feels, behaves, and relates

to others. Personality is the only affective trait that

is maintained in the long term and reﬂects individ-

ual differences in mental characteristics (Ortony et al.,

1990).

In recent years, the interest in analyzing and us-

ing these affective characteristics is increasing in the

computing area. Affective computing is the area that

deals with the treatment of emotions in computing

science (Picard and Picard, 1997). It draws on dif-

ferent theories of the psychological and cognitive sci-

ences. In the last years, different studies to incorpo-

rate emotions and affective characteristics in the soft-

ware processes have been conducted to improve the

simulation of human behavior (Becker-Asano, 2008;

Gebhard, 2005; Alfonso et al., 2014). These pro-

posals employ computational models to deﬁne emo-

tions, mood, and personality. Generally the emo-

tions are represented with the OCC (Ortony et al.,

1990) model which classiﬁes emotions into twenty-

two categories based on reactions to different situa-

tions. The mood is usually represented using the PAD

model. PAD is a three dimensional model proposed

in (Mehrabian, 1996) that deﬁnes the mood as an av-

erage of individual emotional states across a repre-

sentative variety of life situations. The three dimen-

sions are: Pleasure, Arousal, and Dominance

. Plea-

In some references this model is named VAD replacing

the Pleasure by the Valance.

236

Taverner, J., Alfonso, B., Vivancos, E. and Botti, V.

Modeling Personality in the Affective Agent Architecture GenIA

DOI: 10.5220/0006597202360243

In Proceedings of the 10th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2018) - Volume 1, pages 236-243

ISBN: 978-989-758-275-2

sure denotes how pleasant or unpleasant a stimulus

is, excitement is the activation that stimulus produces,

and domination is the level of dominance or submis-

sion to that situation. Finally, one of the most popular

models to represent the personality in affective com-

puting is the Five Factor Model (FFM) (McCrae and

John, 1992). The FFM divides personality into ﬁve

dimensions: Extraversion, Agreeableness, Conscien-

tiousness, Neuroticism, and Openness to Experience.

Each trait summarizes a large number of distinct and

more speciﬁc personality characteristics. The ﬁve fac-

tors together determine how a person will respond to

different stimuli during his or her life.

One of the technologies that is becoming more rel-

evant in simulation environments are the multi-agent

systems (Wooldridge and Jennings, 1994). These sys-

tems provide adaptability, scalability, versatility, au-

tonomy, and have high fault tolerance. Currently there

are representations and formalizations of agents that

take into account the affective processes to make de-

cisions simulating human behavior (Gebhard, 2005;

Becker-Asano, 2008). These formalizations model

the evaluation process, the dynamics of the emotions,

or the effect that emotions have on the cognitive and

behavioral processes of the agents. However, there

are not many architectures for affective agents that

make a general formalization of the interrelation be-

tween rational processes and affective processes. One

of the formalizations that takes into account this in-

terrelation is the GenIA

architecture (Alfonso et al.,

2017) that is the base of our a proposal to manage

the personality in a multi-agent system. Our proposal

allows the use of different personality theories and

groups agents according to their personality by incor-

porating personality proﬁles.

2 BACKGROUND AND

SUPPORTING THEORIES

Personality is a construct that is used in psychol-

ogy to explain the individual differences that con-

stitute a person and allows him/her to differentiate

from others. Personality inﬂuences the way in which

emotional responses to speciﬁc stimuli occur (Ortony

et al., 1990), but there is no absolute consensus on

what should be understood by personality. One of the

most accepted deﬁnitions deﬁnes personality as the

dynamic individual organization of the psychophysi-

cal systems responsible for their characteristic think-

ing and behavior (Allport, 1937). In a general way,

it can be said that personality is a set of distinctive

features of human beings that remain relatively stable

over time and are invariant to different situations. Be-

ing a stable trait, personality allows to predict certain

behaviors in people when they are in a concrete sit-

uation. In addition, personality theories also include

other elements such as cognition, affection, or moti-

vations that deﬁne the way people behave and allow to

explain the inconsistencies that sometimes destabilize

the personality.

There are two approaches to the classiﬁcation

of individuals based on their personality. The ﬁrst

one is the type approach, in which personality can

be deﬁned using a ﬁnite number of categories such

as: optimistic, depressed, irascible, or melancholic.

These types are used as categories of people with

similar characteristics and each individual may or

may not belong to a particular category (Clonninger,

1993). This model is criticized because it does not

provide any information on the degree to which an

individual belongs to a particular category. For this

reason, in psychology the use of the traits approach is

more popular. The traits approach allows to quantify

the degree that a person has of a certain trait. Traits

are characteristics that distinguish people from the

rest and that affect the way they behave (Matthews

et al., 2003). For example, a person can be very

active and somewhat depressed. The traits allow to

describe in a more precise way the personality and

the behavior that the types.

2.1 Personality Affects on Cognition,

Emotions, and Mood

Psychologists discuss of different cognitive processes

that inﬂuence on humans own being. These pro-

cesses include reasoning, memory, attention, decision

making, problem-solving, and perception. There is a

strong relationship between emotions and these cog-

nitive processes, so that emotions are able to deform

these processes and produce different results depend-

ing on the type of emotion. In fact, empirical evidence

suggests a critical impact of emotions on cognition

and a high variability of these effects among people

with different personality traits. For example, people

who have a high level of openness and a low level of

neuroticism have a greater ability to perform differ-

ent cognitive task. Extraverted people are better than

intraverted people in reaction-based task, while intro-

verted people are better than extraverted people in the

processing and reasoning task (Allen et al., 2017).

Personality plays a very important role in emo-

tions and mood. Personality can make the person

more or less susceptible to experience certain types of

emotions (Zelenski, 2007). For example, in the FFM,

the extraversion trait is related with positive emotions

Modeling Personality in the Affective Agent Architecture GenIA

237

and moods like to joy, enthusiasm, emotion, energy,

and also with daring and trust (Watson and Naragon-

Gainey, 2014). On the other hand, neuroticism trait

is related with negative emotions and moods like fear,

anxiety, sadness, guilt, depression, dissatisfaction, or

anger among others (John and Srivastava, 1999; Der-

ryberry and Reed, 1994). Mood and emotions also

inﬂuence cognitive processes. For example, negative

mood affect people’s judgments in a negative direc-

tion, increasing the perception of risk (Ditto et al.,

2006). Therefore, the personality will be related to the

appraisal process. The appraisal process comes from

the psychological theory of appraisal (Lazarus, 1991)

that argues that the emotions are the result of the

interpretation and explanation that each person per-

forms based on his/her circumstances and concerns

(Scherer, 2001). So, emotions are not mere responses

to stimuli but also the result of the individual’s as-

sessment of this stimuli (Roseman, 1996). Through

this appraisal process, a person interprets his/her re-

lationship with the environment (Smith et al., 1990;

Scherer, 2001).

3 AGENT PERSONALITY IN

GenIA

There are some previous works using the personal-

ity in multi-agent systems. (Santos et al., 2011) pro-

pose a multi-agent system model which employs per-

sonality to simulate a group of people in a negotia-

tion task. Each agent was programmed according to

the personality of the user, obtained by performing

a personality test. The personality was modeled us-

ing the FFM grouping the agents in different person-

ality proﬁles: negotiator, aggressive, submissive, and

avoidant. Each proﬁle was associated with a different

behavior.

Another example is the OA3 architecture (Alfonso

et al., 2015). They deﬁne different basic personality

proﬁles: sociable, mediating, negotiating, and realis-

tic, and compare several agents following these pro-

ﬁles through the use of classic games, such as the pris-

oner’s dilemma and the trust game. One of the conclu-

sions obtained is that personality is one of the factors

that best explains the differences between the agents

and the variability that occurs during the simulations.

That conclusion is coherent and in accordance with

the theories of personality previously explained.

3.1 GenIA

Architecture

GenIA

(Alfonso et al., 2017) is a general-purpose

architecture for intelligent agents based on the BDI

(Believe, Desire, Intention) architecture using Jason

(Bordini et al., 2007). GenIA

facilitates the design

of affective agents in a general way. Psychological

and neurological theories have traditionally focused

on the description of the characteristics and processes

related to emotion and personality. Emotion-related

processes are usually studied from a cognitive per-

spective and can be grouped into the generation of

emotion, the experience of emotion, and the effects of

emotion. The GenIA

architecture includes the cen-

tral processes of these three groups, as well as the

processes of a traditional BDI agent architecture (see

Figure 1). Currently GenIA

offers a default design

that includes an appraisal process based on (Marsella

and Gratch, 2009) and uses Jason as a base platform

for multi-agent systems, the FFM for representing the

personality, and the PAD model for the mood. How-

ever, the GenIA

architecture can be easily expanded

and adapted to other psychological theories. For ex-

ample, in (Taverner et al., 2016) the management of

expectations is incorporated into GenIA

The affective processes in GenIA

include (see

Figure 1): an appraisal process in which the ap-

praisal variables (desirability, expectedness, likeli-

hood, causal attribution, and controllability) are de-

rived from the current situation and the emotions are

generated; an affect generator process which deter-

mines the possible emotional behaviors and coping

responses for a given situation; an affective modula-

tor of beliefs which determines if and how the affec-

tive state biases the beliefs of the agent, contributing

to the beliefs maintenance according to the affective

state; and the affect’s temporal dynamic, which deter-

mines the duration of the affective state’s components

as well as how their intensities decay over time.

The design of the GenIA

architecture proposed in

(Alfonso et al., 2017), allows the introduction of per-

sonality as a set of traits followed by the agent’s ra-

tionality level and a list of coping strategies

. But ini-

tially, in the GenIA

default design, personality is not

used in the cognitive process of the agent. Our goal is

to provide GenIA

with different personality proﬁles

so that users can implement different agent behaviors.

In addition, we have identiﬁed and deﬁned the neces-

sary modiﬁcations for using the personality in both

the cognitive and affective processes of the agents in

the GenIA

default design.

Coping strategies are both physical and physiological

responses produced by the individual to face a situation

such as anxiety or stress. Coping strategies deﬁne how the

individual reacts to an event involving emotional changes,

and these reactions may be involuntary manifestations or

more planned actions.

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

238

Figure 1: The GenIA

architecture (Alfonso et al., 2017). The processes modiﬁed by our proposal are highlighted.

3.2 Deﬁning the Personality Model

In this section we present an extension of GenIA

facilitate the use of agents with personality proﬁles

and to incorporate any personality theory based on

types. Personality proﬁles are very useful in modeling

different behaviors since they allow to create groups

of agents with similar personality traits. We propose a

model to deﬁne these proﬁles following the structures

of Figure 2.

personality profiles→

“personality profiles:”

personality profile

(“,”personality profile)*“;”

personality profile→ <

<ATOM>

[“(”traits range“)”]

traits range→ (<

>|range)

(“,”<

>|range)*

range → “<” (<

> | “-”<

>) :

> | “-”<

>)“>”

personality types→ “personality types:”

<ATOM>

> (“,”<

<ATOM>

>)*“;”

Figure 2: Extension of GenIA

multi-agent system syntax

to deﬁne personality proﬁles. The complete EBNF of the

GenIA

architecture can be found in (Alfonso et al., 2017).

We also propose to modify the speciﬁcation of

the multi-agent system to incorporate the deﬁnition

of the necessary personality proﬁles. In this model

each personality proﬁle is deﬁned by the ranges

of the different traits (in case of employing a trait

based theory). For example, if using the FFM the

traits are openness, conscientiousness, extraversion,

agreeableness, and neuroticism. Two proﬁles using

this model: Proﬁle

A and Proﬁle B can be deﬁned

as follows:

personality profiles:

Profile A(< 0.6, 1.0 >, < 0.5, 0.7 >,

< 0.5, 0.7 >, < 0.7, 1.0 >, < 0.1, 0.6 >).

Profile B(< 0.3, 0.5 >, < 0.0, 0.6 >,

< 0.2, 0.6 >, < 0.5, 1.0 >, < 0.0, 0.2 >).

were each pair of numbers correspond to the interval

of values for the traits openness, conscientiousness,

extraversion, agreeableness, and neuroticism respec-

tively. An agent will be considered in Proﬁle A if it

has a value between 0.6 and 1.0 of openness, from

0.5 to 0.7 of conscientiousness, from 0.5 to 0.7 of

extraversion, from 0.7 to 1.0 of agreeableness, and

between 0.1 and 0.6 of neuroticism. For example, an

agent whose personality is deﬁned as [0.7, 0.5, 0.6,

0.8, 0.1] will be part of the Proﬁle A. But, it will

not be part of the Proﬁle B because it does not sat-

isfy the value of openness (< 0.3, 0.5 >) in Proﬁle B.

But in the personality model incorporated to

GenIA

is also possible to use a type-based model of

personality. For example, if we use the Myers-Briggs

Type Indicator (Myers, 1962) to deﬁne the person-

ality using the types ISTJ, ISFJ and INFJ we can

deﬁne each type as follows:

personality types:

ISTJ

ISFJ

INFJ

This is only used to deﬁne the personality types

allowed to deﬁne the agent’s personality. For each

type we can deﬁne different behaviors as we do with

proﬁles. Therefore, we have also extended the agent

personality model to allow the use of personality

theories based on types. For example, an agent can

have a personality deﬁned as:

personality : { types : [ISTJ] }

In this example the agent’s personality type is

ISTJ. To make this type of representation possible, the

Modeling Personality in the Affective Agent Architecture GenIA

239

agent speciﬁcation includes the personality deﬁned as

a list of types or traits (see Figure 3).

personality → “personality :”“{”

(traits|types) [“,” rat level]

[“,”coping strats]“}”

types → “types :”

“[” <

<ATOM>

> (“,” <

<ATOM>

>)* ) “]”

Figure 3: Extension of GenIA

agent syntax to allow the use

of type theories.

3.3 Improving the Personality

Management in the Default Design

of GenIA

As we said in section 2.1, personality has a rele-

vant effect on cognitive and affective processes. So

if we want to simulate human behavior using affec-

tive agents, it is necessary that personality inﬂuences

these processes. We have also modiﬁed the GenIA

architecture to make the personality affect the cogni-

tive and affective processes.

The modules that are affected by the incorporation

of our personality model are (see Figure 1): the Ap-

praisal process, where the personality affects the gen-

eration of emotions; and the Affect Generator process

where we propose a model where the personality af-

fects the displacement of the mood. In the proposed

model, intentions and affective events related to per-

sonality proﬁles are generated.

Finally, we have modiﬁed the personality on the

platform incorporating the use of theories based on

types and a model that allows the use of personality

proﬁles to deﬁne different behaviors. For this pur-

pose, we have modiﬁed the agent selection of plans

process generating new intentions related to person-

ality proﬁles. Therefore, users can deﬁne different in-

tentions for each proﬁle allowing to create different

behaviors according to agent’s personality in a simple

way.

In GenIA

, the selection of plans is done through

two processes: The Jason plan selection process

which returns the list of possible actions sorted

according to their priority, and the selecting affective

actions process which returns a list of possible

affective plans sorted by priority. A GenIA

affective

plan is any plan including the annotation affect ()

in the plan’s label

. This annotation is used to

determine the affective state that the agent must have

In Jason, plans have a label that allows, in addition to

naming the plan, to add annotations that can be later used in

the system.

to select that plan. For example, in the following plan:

@plan1[affect (sadness)]

!event : true <- .print("I’m sad")

the plan’s annotation is affect (sadness), so this

plan will be only chosen if the agent is in a sadness

state. The lists obtained by these two processes are

used to decide the plan to be executed based on the

level of rationality of the agent. A plan is selected

using the Formula 1 proposed in (Alfonso, 2017).

min

i∈I

Rl ∗ Rr

+ (1 − Rl) ∗ Ar

(1)

were I is the set of intentions, Rl is the rationality level

of the agent, Rr

is the priority of the intention i in

the rational list of plans, and Ar

is the priority of the

intention i in the affective list of plans.

Our proposal for the selection of intentions in-

cludes a new annotation for the selection of affective

plans:

personalityProﬁles (pro f ile

, pro f ile

, ...,

pro f ile

)

This new annotation allows to modify the process of

selection of plans based on the agent’s personality

proﬁle. Therefore, this annotation allows the user to

deﬁne different behaviors for each personality proﬁle

in a simple way. The introduction of this new anno-

tation improves the usability of GenIA

, allowing to

easily implement studies like (Gebhard, 2005; Santos

et al., 2011).

The default design of GenIA

has an evaluation

process in which the emotions are selected and the

mood is updated. The default design includes six dif-

ferent emotions: surprise, hope, joy, fear, sadness, and

anger, but can be easily extended by the user. In or-

der to facilitate the use of emotions, our proposal in-

corporate the intensity of emotions. The intensity of

the emotions allows to differentiate the way in which

the cognitive processes work and allows to determine

which emotions affect the individual in function of

his personality (Santos et al., 2011). The intensity of

positive and negative emotions also have an impact

on decision-making processes (Ristvedt and Trinkaus,

2005). In addition, our design allows to deﬁne how

the personality will affect the emotions according to

their intensity.

Our new default design allows the personality to

affect the mood. Initially GenIA

updated the mood

based on the model proposed in ALMA (Gebhard,

2005). According to that model the mood was moved

towards a theoretical point of the PAD space ob-

tained by the emotions selected in the appraisal pro-

cess. They called this point the virtual emotion center

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

240

(VEC). The mood was updated following a ﬁxed dis-

placement without taking into account the agent per-

sonality. But, personality should affect mood (Wat-

son and Naragon-Gainey, 2014; John and Srivas-

tava, 1999; Derryberry and Reed, 1994): people with

high levels of extraversion tend to have more posi-

tive mood than people with high levels of neuroti-

cism, while people with high levels of neuroticism

and lows level of extraversion tend to more negative

mood. Therefore, to simulate this natural human be-

havior it is necessary to consider the personality to up-

date the mood, as for example is proposed in (Gomes

et al., 2014). We propose to modify the current dis-

placement of the mood according to personality traits

following the Formula 2 in function of the neuroti-

cism and extraversion traits. This is deﬁned as

personalityDisplacement =



npe

∗ e −

nne

∗ n



(2)

were pe is the number of active positive emotions (i.e.

the positive emotions that have been calculated in the

appraisal process), ne is the number of negative active

emotions, and npe and nne represent the total number

of positive and negative emotions respectively. Fi-

nally e and n are the levels of extraversion and neu-

roticism. This equation is bounded between zero and

one because the personality traits are also bounded

between zero and one. By this formula we simulate

the effect of the personality in the mood according to

different theories (Derryberry and Reed, 1994; John

and Srivastava, 1999; Ditto et al., 2006). This al-

lows to adjust the mood displacement according to

different personality traits getting closer to the way

it happens in humans (Zelenski, 2007; Watson and

Naragon-Gainey, 2014). For example, if the number

of positive emotions is zero, the number of negative

emotions is greater than zero, and the level of neu-

roticism is high, the mood displacement will be nega-

tive and greater than the mood displacement when the

level of neuroticism is low.

4 RESULTS

In order to test our proposal, we have developed an

experiment with sixty agents playing the Blackjack

game. These agents have different personalities, and

are induced with positive and negative emotions to an-

alyze the evolution of the mood. As we saw in Sec-

tion 3.1, the default design of GenIA

uses the FFM to

represent the personality and PAD to represent mood.

We have created two personality proﬁles: Proﬁle One

consists of agents with low level of extraversion and

high level of neuroticism, while Proﬁle Two repre-

sents agents with high level of extraversion and low

Figure 4: Average evolution of pleasure for Proﬁle One and

Two.

Figure 5: Average evolution of arousal for Proﬁle One and

Two.

Figure 6: Average evolution of dominance for Proﬁle One

and Two.

level of neuroticism. We have established the equi-

librium mood at the point [0.0, 0.0, 0.0] so that all

agents start with the same mood. This decision al-

lows to easily compare the evolution of the mood in

both personality proﬁles.

To simplify the experiment we have chosen only

two emotions: joy and sadness. The main concern

Modeling Personality in the Affective Agent Architecture GenIA

241

of the agents is to win. Therefore, winning a game

causes the joy emotion in the agent while losing a

game causes sadness. Agents play eighteen rounds:

seven have a winning result and eleven have a losing

result. Therefore, we have alternated the two emo-

tions over time and checked the mood evolution for

the two different proﬁles. As Figures 4, 5, and 6 show,

the mood of agents classiﬁed in Proﬁle One tends to

decrease, while mood of agents classiﬁed in Proﬁle

Two tends to increase. We can also see that the ﬁnal

mood is different for each proﬁle in all dimensions

of the PAD. This is consistent with the psychologi-

cal theories that determine the individual differences

produced by the personality when dealing with a par-

ticular emotion (Zelenski, 2007; Tong, 2010).

We can also see that negative emotions have a

greater impact on Proﬁle One, and positive emotions

have a greater impact on Proﬁle Two. For example,

the ﬁrst iteration produces a negative emotion and the

effect is greater in Proﬁle One than in Proﬁle Two.

5 CONCLUSIONS

Personality is a crucial factor in understanding the in-

dividual differences that affect the way human beings

perceive the environment and emotions, and that has

an impact on mood and on cognitive processes. We

have seen different ways of representing personality.

We have also seen how different personality traits can

directly inﬂuence certain emotions and behaviors. In

this paper we have used the FFM focusing on neuroti-

cism and extraversion as the most important factors

that inﬂuence the affective and cognitive processes.

Our proposal is based on different psychological the-

ories that argue that extraversion and neuroticism are

the factors with greatest relevance in affective and

cognitive processes.

We have presented an extension for the architec-

ture of affective agents GenIA

. Our extension en-

large and generalizes the use of the personality within

this architecture extending its adaptability employing

personality theories based on types and traits. We

have modiﬁed the cognitive processes of the agents to

allow the personality to inﬂuence the process of rea-

soning and decision making. We have also proposed

a formula for updating the mood according to the dif-

ferent parameters of the personality, based on differ-

ent theories that relate neuroticism to negative emo-

tions and mood and extraversion to positive emotions

and mood. Our proposal allows the use of personality

proﬁles that are very useful when modeling different

behaviors grouping individuals with a similar person-

ality. Therefore, personality proﬁles allow the user

to abstract from the different personality traits when

modeling different behaviors. We have tested our pro-

posal by an example where sixty agents classiﬁed into

two personality proﬁles have been induced to negative

and positive emotions. The obtained results are con-

sistent with the psychological theories mentioned in

this paper.

One of the future extensions of this work has its

origin on the effect of the temporal dynamics pro-

cess. Currently, this process calculates the decay of

the mood towards the equilibrium state using a ﬁxed

and constant value for all the agents. One possible im-

provement would be to investigate how the personal-

ity traits affect this process and to propose a formula,

such as the one we have proposed for the updating of

the mood, that allows to calculate the way in which

mood must be updated in function of personality.

ACKNOWLEDGEMENTS

This work was supported by the Spanish Government

project Pesedia (TIN2014-55206-R) and the General-

itat Valenciana project Humbace: Human and Social

Behaviour Models for Agent-Based Computational

Economics (PROMETEOII/2013/019).

REFERENCES

Alfonso, B. (2017). Agents with Affective Traits for

Decision-Making in Complex Environments. PhD the-

sis, Universitat Polit

ecnica de Val

encia.

Alfonso, B., Vivancos, E., and Botti, V. (2014). An open

arquitecture for affective traits in a BDI agent. Proc.

of the 6th ECTA 2014. Part of the 6th IJCCI 2014,

pages 320–325.

Alfonso, B., Vivancos, E., and Botti, V. (2017). Toward

formal modeling of affective agents in a BDI archi-

tecture. ACM Transactions on Internet Technology

(TOIT), 17(1):5.

Alfonso, B., Vivancos, E., Botti, V., and Hern

andez, P.

(2015). Building emotional agents for strategic de-

cision making. In Proceedings of the International

Conference on Agents and Artiﬁcial Intelligence, vol-

ume 2, pages 390–397. ICCART’15.

Allen, M. S., Laborde, S., and Walter, E. E. (2017).

Health-related behavior mediates the association be-

tween personality and memory performance in older

adults. Journal of Applied Gerontology, First publica-

tion March 2017.

Allport, G. W. (1937). Personality. Holt New York.

Becker, P. (2001). Structural and relational analyses of emo-

tions and personality traits. Zeitschrift fur Differen-

tielle und Diagnostische Psychologie, 22(3):155–172.

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

242

Becker-Asano, C. (2008). WASABI: Affect simulation for

agents with believable interactivity. IOS Press.

Bordini, R. H., H

ubner, J. F., and Wooldridge, M. (2007).

Programming multi-agent systems in AgentSpeak us-

ing Jason, volume 8. John Wiley & Sons.

Broome, J. (2002). Practical reasoning. Reason and nature:

Essays in the theory of rationality, pages 85–111.

Clonninger, S. (1993). Theories of personality: Under-

standing persons. Englewood Cliffs, NJ: Prentice

Hall.

Damasio, A. and Sutherland, S. (1994). Descartes’ er-

ror: Emotion, reason and the human brain. Nature,

372(6503):287–287.

Derryberry, D. and Reed, M. A. (1994). Temperament and

attention: Orienting toward and away from positive

and negative signals. Journal of personality and social

psychology, 66(6):1128–1139.

Ditto, P. H., Pizarro, D. A., Epstein, E. B., Jacobson, J. A.,

and MacDonald, T. K. (2006). Visceral inﬂuences on

risk-taking behavior. Journal of Behavioral Decision

Making, 19(2):99–113.

Frijda, N. H. (1986). The emotions: Studies in emotion

and social interaction. Paris: Maison de Sciences de

l’Homme.

Gebhard, P. (2005). Alma: a layered model of affect.

In Proc. of the fourth international joint conference

on Autonomous agents and multiagent systems, pages

29–36.

Gomes, M., Oliveira, T., Silva, F., Carneiro, D., and Novais,

P. (2014). Establishing the relationship between per-

sonality traits and stress in an intelligent environment.

In Int Conf on Industrial, Engineering and Other Ap-

plications of Applied Intelligent Systems, pages 378–

387.

John, O. P. and Srivastava, S. (1999). The big ﬁve trait

taxonomy: History, measurement, and theoretical per-

spectives. Handbook of personality: Theory and re-

search, 2(1999):102–138.

Lazarus, R. S. (1991). Emotion and adaptation. Oxford

University Press.

Marsella, S. C. and Gratch, J. (2009). EMA: A process

model of appraisal dynamics. Cognitive Systems Re-

search, 10(1):70–90.

Matthews, G., Deary, I. J., and Whiteman, M. C. (2003).

Personality traits. Cambridge University Press.

McCrae, R. R. and John, O. P. (1992). An introduction to

the ﬁve-factor model and its applications. Journal of

personality, 60(2):175–215.

Mehrabian, A. (1996). Pleasure-arousal-dominance: A gen-

eral framework for describing and measuring individ-

ual differences in temperament. Current Psychology,

14(4):261–292.

Myers, I. B. (1962). The Myers-Briggs Type Indicator:

Manual. Palo Alto, CA: Consulting Psychologist

Press.

Ortony, A., Clore, G. L., and Collins, A. (1990). The cog-

nitive structure of emotions. Cambridge University

Press.

Picard, R. W. and Picard, R. (1997). Affective computing,

volume 252. MIT press Cambridge.

Ristvedt, S. L. and Trinkaus, K. M. (2005). Psychologi-

cal factors related to delay in consultation for cancer

symptoms. Psycho-Oncology, 14(5):339–350.

Roseman, I. J. (1996). Appraisal determinants of emotions:

Constructing a more accurate and comprehensive the-

ory. Cognition & Emotion, 10(3):241–278.

Santos, R., Marreiros, G., Ramos, C., Neves, J., and Bulas-

Cruz, J. (2011). Personality, emotion, and mood in

agent-based group decision making. IEEE Intelligent

Systems, 26(6):58–66.

Scherer, K. R. (2001). Appraisal considered as a process of

multilevel sequential checking. Appraisal processes

in emotion: Theory, methods, research, 92(120):57.

Smith, C. A., Lazarus, R. S., et al. (1990). Emotion and

adaptation. Handbook of personality: Theory and re-

search, pages 609–637.

Taverner, J., Alfonso, B., Vivancos, E., and Botti, V. J.

(2016). Integrating expectations into Jason for ap-

praisal in emotion modeling. In Proc. of the 8th Inter-

national Joint Conference on Computational Intelli-

gence, IJCCI 2016, Volume 1: ECTA, pages 231–238.

Tong, E. M. (2010). Personality inﬂuences in appraisal–

emotion relationships: The role of neuroticism. Jour-

nal of Personality, 78(2):393–417.

Watson, D. and Naragon-Gainey, K. (2014). Personality,

emotions, and the emotional disorders. Clinical Psy-

chological Science, 2(4):422–442.

Wooldridge, M. and Jennings, N. R. (1994). Agent the-

ories, architectures, and languages: a survey. In Int

Workshop on Agent Theories, Architectures, and Lan-

guages, pages 1–39.

Zelenski, J. M. (2007). The role of personality in emotion,

judgment and decision making. In Vohs K, Baumeis-

ter R, Loewenstein G, Do Emotions Help or Hurt De-

cision Making, pages 117–132.

Modeling Personality in the Affective Agent Architecture GenIA

243