Temporal Causal Network Model for Appraisal Process in Emotion

Fawad Taj and Michel C. A. Klein

Behavioural Informatics Group, VU University Amsterdam, The Netherlands

Keywords: Emotion Modelling, Temporal Causal Model, Appraisal, CPM, Component Process Model.

Abstract: Most of the emotion theories consider appraisal as the major component in an emotional episode. The

appraisal theories legitimately try to explain the actual process of appraisal. Number of computational

architecture for emotional and cognitive agents exists, which try to incorporate the major cognitive appraisal

theories, but they compromise on a certain aspect of the theories due to its complexity. In this paper, a

temporal causal network model approach is used to address the dynamics and temporal processing of different

evaluation checks in the appraisal component. The checks included in the model are inspired by the

Component Process Model and other neuro and cognitive science literature. Simulations have been done to

show the temporal causality between different evaluation checks.

1 INTRODUCTION

Appraisal theories of emotion define emotion as a

process, not state (Moors et al., 2013). The term

“appraisal” was first introduced by (Arnold, 1960) as

a counter argument against William James (1884)

famous bear example, where he claims that emotions

are the reaction or interpretation of physical arousal,

after stimulus onset. In contrast, Arnold claims that,

before any emotional experience, human thoroughly

evaluate the event/situation according to their well-

being. So, these are those thoughts that make our

perception and generate emotion. (Kemper and

Lazarus, 1992) claim that “emotions are organized

psycho-physiological reactions to news about on-

going relationships with the environment.” (Arnold,

1960), (Lazarus, 1966), (Kemper and Lazarus, 1992)

(Moors et al., 2013), (Moors et al., 2013), (Frijda,

1986) and (Scherer, 1984) are the major adherents

and formalizer of this type of theories. Appraisal

theories consider emotion as a componential process

because so many sub-systems work together to

coordinate and synchronize the process.

Furthermore, a recent and well explained

appraisal theory is proposed by (Scherer, 2001), who

tries to answer all the major questions related to

dynamic design feature of emotion through

Component Process Model (CPM) (Scherer, 2001,

2004, 2009). The appraisal component in CPM has

defined clear criteria or checks for the elicitation of

the event, Scherer calls them as Stimulus Evaluation

Checks (SEC’s). In a recent version of CPM (Scherer,

2013) the SEC’s are categorised into four major

appraisal intents. The ordering and output of these

SEC’s are thoroughly explained and scientifically

proved through numerous studies.

There are a number of computational models that

used Scherer theory as the main component for

cognitive appraisal processing within their models,

e.g. WASABI (Becker-Asano, 2008), PEACTIDM

(Marinier, 2008), etc. Most of the computational

models ignore the causal and temporal dimension of

CPM and are mostly designed as the rule-based

systems (Sander, Grandjean and Scherer, 2005).

Scherer proposed a network-based representation for

the computational model in which one node

represents single evaluation check (see Fig 1.). In

parallel fashion and through some sort of appraisal

derivation model, each node will always be updated

through best estimated value about the event (Sander,

Grandjean and Scherer, 2005). Moreover, he also

suggest to adopt non liner dynamics system for

emotional modelling rather than linear function or

statistical methods (e.g. regression analysis) (Sander,

Grandjean and Scherer, 2005).

The focus of this article is to design a temporal

causal network model for appraisal component of

emotion, mainly inspired by the CPM, whereas

keeping the component view in mind, for

computational appraisal model suggested in

(Marsella et al., 2010) (see Fig 2.). This article uses

CPM variables as a base for affect derivation with its

intensity and its consequences on action preparation

and execution. The temporal dimension and the

Taj, F. and Klein, M.

Temporal Causal Network Model for Appraisal Process in Emotion.

DOI: 10.5220/0006867403470356

In Proceedings of 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2018), pages 347-356

ISBN: 978-989-758-323-0

347

causality between the checks have been critically

analysed and simulated.

The paper is structured as follows. Section 2

consists of detail discussion about all the SEC’s in

appraisal component of CPM. Section 3 presents the

temporal causal network modelling approach. In

Section 4 the implementation of the computational

model for appraisal is discussed. Section 5 describes

simulations that illustrate the working of the model.

Finally, Section 6 contains the conclusion.

2 THEORITICAL BACKGROUND

In the CPM architecture emotion is defined as “an

episode of interrelating, synchronized changes in the

states of all or most of the five organismic subsystems

in response to the evaluation of an external or internal

stimulus even as relevant to major concerns of the

organism” (Sander, Grandjean and Scherer, 2005).

So, from the definition, it can be concluded the

storyline of emotion starts with the appraisal.

Appraisal is, therefore considered as one of the major

components of CPM.

Figure 2: Component model view of computational

appraisal models, adopted from (Marsella et al., 2010).

Recent studies of electrophotography (EEG) and

event-related potentials (ERPs) on neural response of

the brain to emotional events recorded this response

within 200 ms after stimulus presentation (Hillyard,

Teder-Sälejärvi and Münte, 1998). The earliest

processing of emotional episode starts with cognitive

appraisal, which ranges from detecting the novelty to

more complex check of causality and evaluating

coping potential (Folkman and Lazarus, 1985;

Scherer, 2001). The responses from all the sub-

systems are collectively labelled to certain emotion in

the language spoken in the respective culture. The

main components that uncover the whole emotional

episode are; appraisal results, action tendencies,

motor expression, somatovisceral changes and

feeling component with subjective experience

(Scherer, 2001, 2005, 2013). Some of the main issues

that are still under discussion in appraisal theories are

the number of appraisal criteria’s and the ordering of

the checks. In this paper only those checks are used

for which we have found conclusive solid evidence

for its validity and ordering from neuroscience,

psychology or cognitive sciences literature.

2.1 CPM Variables

Stimulus Evaluation checks(SEC’s), categorized in

four groups (Relevance, Implication, Coping

potential, Normative significance), listed in order of

processing proposed in CPM architecture (Scherer,

2013).

2.1.1 Relevance

Almost all the appraisal theorist agree on the fact that

the first evaluation check in the appraisal process is

Figure 1: Architecture of appraisal process in CPM produced from (Scherer, 2005).

SIMULTECH 2018 - 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

348

relevance. It is processed at very low unconscious

level and at very automatic fashion, due to this

property, some of the critics are against using this as

an appraisal check. The stored schemata have been

matched to event/stimulus features and if they match

closely then attention is deployed.

In CPM the relevance detection has been further

explored and it has been found that the following

parameters determine the relevance of the event:

Novelty occurrence, Intrinsic pleasantness, Concern

relevance.

Novelty occurrence check: At this check the

question is how familiar or sudden the event is? It is

considered that novelty has been checked at sensory–

motor level and it is considered as first step toward

emotional episode. Different responses have been

recorded in the peripheral cortex and hippocampus as

soon as 100 ms after stimulus onset (Brown and

Bashir, 2002) and an increased response to novel

stimulus (Blackford et al., 2010).

Intrinsic pleasantness is the innate quality of an

event/stimulus and mostly the processing of this

check is free from appraiser motivational factors

(current needs, goals, taste etc.), but some of the

theorists define this check as concerned relevance of

event, given all of his motivational factors. Some of

the famous intrinsic un/pleasantness stimuli groups

exists e.g evolutionary prepared (snake, anger,

expression), reproduction (sex), taste(sweetness), etc

(Scherer, 2013). At the end, most of the researchers

agree with the fact that intrinsic pleasantness

increases amygdala response and is recorded at 140

ms after stimulus onset (Pourtois et al., 2010).

Concern pertinence is the relevance check where

current motivational factors matter. For example, less

attention will be paid to food cues in a state of satiety

(Sacharin, Sander and Scherer, 2012). So, the concern

relevance covers a large array of motivational states.

This is why it is the most complex part of relevance

check.

2.1.2 Implication

To further process the relevant emotional stimulus,

the next target is to find its consequences for one’s

well-being. The checks define Implication that are:

Goal Conduciveness checks the compatibility of

the event with the current goal, i.e. whether the event

is facilitating or blocking progress toward goal-

attainment. The neuroscience literature shows that

conflict processing during goal quest has been

recorded in cingulate cortex (AAC) and dorsolateral

prefrontal cortex (DLPFC) (Brosch and Sander,

2013). The activity of ACC has been detected at 340-

380 ms after the conflicting stimulus onset (Van

Overwalle, 2009).

Agency: The core theme of this check is to find

causation of the event: caused by me, someone else

or nature? Neurosciences research shows that

different regions are involved in both internal and

external agency (Sperduti et al., 2011).

Causal motive check tries to find the reason for

stimulus. Why it happened and determines whether it

is intentional or negligence.

Discrepancy from expectation, calculate the

difference between the expectation and actual action

at the point of time. So, the higher the difference, the

higher the value will be.

Outcome probability checks the likelihood of the

event in form of probability. This check is also

important in calculating the intensity of the appraisal

frame.

Urgency Check determines the strength of action

when something important is on stake. In

neuroscience perspective, its effect is an immediate

increase in action in autonomic nervous system

(ANS).

2.1.3 Coping Potential

An earlier appraisal model of (Folkman and Lazarus,

1985) used the term secondary appraisal, which finds

the resource options available to deal with the current

situation. Scherer in his CPM model discussed three

further aspects of coping potential appraisal.

Control determines the coping potential by

investigating how much control one has over the

situation. For example, in a natural disaster one has

very low control, whereas events can be controllable

when humans are involved.

Power aspect of coping potential appraisal checks

for physical strength, money social support,

information etc.

Adjustment determines one’s ability to

accommodate the effect of an event.

2.1.4 Norm Significance

Humans live in society and have certain social norms

due to which individuals are always curious about the

views of others on certain actions. There are two

further sub-aspects of normative significance.

Internal standard checks how much the action

and outcome is compatible with one’s internal

standard and moral standards.

External standard checks how much the action

and outcome are compatible with society norms and

standards.

Temporal Causal Network Model for Appraisal Process in Emotion

349

2.2 Affect Derivation Component

After detail discussion about all the checks for

stimulation evaluation, now we move to the outcome

of these variable as a whole and what should we

expect out of the model. The feeling component

within CPM has a very important role of monitoring

and regulation of the component process model.

Every component within CPM expresses its

emotional experience through the feeling component

and this state also serve the purpose of a

communication link between the components. All

information that becomes conscious in central

representation state, is called “feeling” or qualia. The

integrated information from all the components to

feeling states determines its quality, intensity and

duration (Scherer, 2001). Now a days, the only

measure we have to count conscious feeling is

through verbal self-report. Therefore, to

accommodate this model into other

emotional or

cognitive agent architecture we propose a separate

affect derivation block where we can find intensity

and label of the current appraisal profile.

Moreover, some of the modeller calculated

intensity of the appraisal profile and is used to map

the appraisal profile to two or three-dimensional

space. This intensity can be used to differentiate

between number of close affective states, for example

cold anger vs hot anger. We also calculated intensity

which will be discussed in detail in section 4.

The feeling actually serves a kind of monitoring and

helps in choosing the best possible options for an

action. We tried to

incorporate the feeling component

suggested by (Damasio, 1998). In our model before

performing an action, feeling state is affected by

predictive as-if body loop, which give a sense of

preview and valuing the action before it has actually

been performed. This feeling state can be later used

for emotion regulation and integration of other

cognitive states.

3 TEMPORAL CAUSAL

NETWORK MODELING

The temporal causal-network modelling approach

explained in (Treur, 2016) has been used to model the

proposed model. It is generic approach to model any

dynamic process with causal relations. The temporal

dimension enables the modelling of cyclic causal

relations with exact timing. In broader terms, there

are some similarities between artificial neural

networks and this approach, for example in case of

continuous time and recurrent, but there are important

differences as well. For example, no hidden layer

exists that do not represent any real-world

phenomena; each state within this approach should be

clearly defined with exact causal and temporal

dimensions.

The models in temporal causal network modelling

approach can be represented in two ways: a)

conceptual representation and numerical represent-

tation. Both types of representation can be easily

transformed into each other in a systematic manner.

Conceptual representation can be done through

graphs or matrices. A graphical representation

involves states which represent some real wold

phenomena and the arrows show the causal relation

between the two states. Some additional information

is given below:

• Value of connection(ωX,Y) representing

strength of causality and it value ranges between

[-1, 1].

• How fast a state Y can change upon casual

impact. Speed factor is denoted by ηY, and value

ranges between [0,1].

• For multiple impacts on state Y, combination

function cY(...) is used to combine the effect.

There are a number of combination functions

defined, varying from simple sum function to

advance logistics function.

The conceptual representation of model can be

translated into numerical representation as follow.

• For any state Y at any time point t, Y(t) denots

the activation value of Y.

• The causal impact of state X on Y at time point t,

can be defined by

Impact

X,Y

= ω

X,Y

X(t).

• Total aggregated impact of the multiple impact

on state Y at time t combined by combination

function cY(...) can be defined by

aggimpact

(t) = c

(impact

, impact

impact

, …)

= c

(ω

(t), ω

(t), …)

the aggimpact

(t) will have upward or downward

effect at time point t, but how fast this change

takes place depends on the speed factor

• Y(t+Δt) = Y(t) + η

[aggimpact

(t) – Y(t)] Δt

• The following difference and differential

equation can be obtained for state

Y(t+Δt) = Y(t) + η

(ω

(t), ω

(t),

(t), …) – Y(t)]Δt

dY(t)/dt = ηY [c

(ω

(t), ω

(t), …) – Y(t)]

SIMULTECH 2018 - 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

350

Figure 3: The graphical conceptual representation of proposed model. Circles represent the states and arrows show the

connections. The orange arrows are used to calculate the intensity. Abbreviations: SR(s), sensory representation of stimulus;

Rel, Relevance; Impl, Implication; C.P, coping potential; N.S, normative significance; PA(a), Preparation for action (a);

SR(b), sensory representation of bodily state(b); FS(b), Feeling for action; EA(a), Execution of Action(a).

4 THE COMPUTATIONAL

MODEL

The proposed model is designed at the conceptual

level, keeping in mind the temporal and causal

attributes of CPM and also the component view of

computational model from Marsella (see Fig. 2). The

insights are already discussed in Section 2. In most of

the architectures the appraisal variables derivation

and affect derivation process are separated. We

mainly focused on affect derivation and its effect on

action selection.

4.1 Graphical Representation of the

Model

An overview of the model is depicted in Figure 3. It

shows the conceptual representation of the temporal

causal model of the cognitive appraisal process. The

parameter defining each checks are separately

marked in doted box, e.g. for relevance check is

defined by novelty, intrinsic pleasantness, concern

relevance. The other checks in second third and fourth

place are implication, coping potential and normative

significance respectively. Scherer theory claims that

the order of causality is caused by economical and

logical reasons and argue that it is uneconomical to

process the stimulus if it is not relevant. He further

argues that once the stimulus is considered relevant,

the attention has been developed toward the stimulus

and further checks have been performed. This

casualty has been adopted in the model by applying

advance logistic function in which a certain threshold

can be defined, if the values goes

above that threshold

then further checks would be performed.

Furthermore, Control state has been used which

check the value of the relevance, once it get high from

a given threshold, it would activate the effecting

states for example in our model when relevance is

high, control state for relevance will activate

attention. The other control state is used for

implication check which also activate action state.

There is a causality between appraisal and action

formation which has been modelled by combining

Damasio’s as-if body loop (Damasio, 1999). Damasio

argues that before taking any action there is an

internal simulation of the action prior to the actual

action. This simulation is then compared with the

feeling associated with each option available and

gives a sort of action selection process, a GO signal.

The as-if body loop conceptually looks like:

Sensoryreprsentation(Srs)→

preparationbodilychanges(PA(b))→ActionExecution

(EA(a))

Temporal Causal Network Model for Appraisal Process in Emotion

351

4.2 Formalization of the Parameters

Valence is considered as simple continuous one-

dimensional value e.g. bad vs good, tall vs. short,

positive vs. negative etc.). Whereas, in CPM and

many other appraisal models different type of valence

appraisals are used for example intrinsic pleasantness

vs intrinsic unpleasantness and goal conduciveness

vs. goal obstructiveness. For this model, we have

scaled all the type of valence appraisal to the values,

ranges between [-1,0]. The range of scaled values are:

Novelty [0,1], Intrinsic Pleasantness [-1,1], Concern

Pertinence [0,1], Goal Conduciveness [-1,1], Agency

[0,0.5,1], Causal Motive [0, 0.5, 1], Outcome

probability [0,1], Discrepancy from Expectation (DE)

Table 1: Initial values, Sped factor and combination

function used in model.

States Initial

value

Speed

factor

Combination

function

Stimulus(s) 0 0.3 Identity

Sr(s) 0 0.3

Alogistic

with σ=30

τ=0.3

Novelty 1 0.4 Identity

Int. Pleas. 1 0.4 identity

Con. Perten. 1 0.4 Identity

Rel 0 0.4

Alogistic

with σ=30

τ=0.3

Attention 0 0.4 Identity

Cs_R 0 0.3

Alogistic

with σ=8 τ=0.3

Agency 0.9 0.4 Identity

Cause Motive 0.9 0.4 Identity

Out C. Prob. 0.9 0.4 Identity

Disc. From

Expec

0.9 0.4 Identity

Goal Cond. 0.9 0.4 Identity

Urgency 0.9 0.4 Identity

Imp 0 0.3

Alogistic

with σ=0.4

τ=0.4

Cs_Imp 0 0.3

Alogistic

with σ=5.5

τ=0.2

Control 0.9 0.3 Identity

Power 0.9 0.3 Identity

Adjustment 0.9 0.3 Identity

CP 0 0.3

Alogistic

with σ=0.5

τ=0.4

Internal

standard

0.9 0.3 Identity

External

Standard

0.9 0.3 Identity

NS 0 0.3

Alogistic

with σ=0.5

τ=0.3

PA(a) 0 0.2

Alogistic

with σ=0.7

τ=0.2

SR(b) 0 0.2 Identity

FS(b) 0 0.2 Identity

EA(a) 0 0.09

Alogistic

with σ=20

τ=0.3

[0,1], Urgency [0,1], Control [0, 1], Power [0,1],

Adjustment [0,1]

The combination function, initial value and speed

factor for each state is given in Table 1, whereas

Table 2 defines connection between states.

Table 2: Connection values between states.

From To (connection value(s))

Sr(s)

All the connection value from Sr(s)

to other states is 0.9.

Novelty, Int Pleas, conc.

Pert.

Rel(1,0.7,0.5)

Agency, Cause Motive,

Out C. Prob, Disc From

Expec, Goal Cond,

Urgency

Connection value is same for all

these connections, Impl(1).

Control, Power,

adjustment

Connection value For all these

connection to C.P value is 1.

Internal standard,

external standard

N.S(1)

Rel Cs_

(1)

Cs_Rel Rel(-0.15), Attention(1)

Attention Sr(s)(0.5)

Impl Cs_Impl(1), C.P(1)

Cs_Impl Imp(-0.15), PA(a)(1)

C.P N.S(1)

N.S PA(0.4)

PA(a) Sr(b)(1),EA(a)(1)

Sr(b) FS(b)(1)

FS(b) PA(a)(1)

4.3 Affect Derivations

As discussed in section 2, the affect derivation is the

key component in any appraisal model. In affect

derivation process, we are calculating Intensity and

would label the appraisal profile that is currently

under consideration.

4.3.1 Intensity

One of the important aspects which determines the

effect of appraisal on behaviour is intensity. It is also

important for mapping appraisal into a

multidimensional space. Currently there are no

standard theories or rules for producing intensity but

(Marinier, Laird and Lewis, 2009) defined three

general criteria for an intensity function:

1. Intensity should in limited range of [0,1].

2. Value of intensity should not be influenced by

single appraisal value, each appraisal check

should contribute

3. The value of intensity for expected stimulus

should be less than unexpected one.

Keeping in mind all the above conditions, the

function that will calculate the intensity is the

combination of Outcome probability(OP) and

SIMULTECH 2018 - 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

352

Discrepancy from expectation (DE). If both OP and

DE are low or high, intensity will be high because

either check doesn’t meet, if they are having opposite

values then intensity would be low. They call it a

surprise factor:

I = (1 - OP) (1 - DE) + (OP. DE) ... (1)

To include other checks and to meet first and

second conditions above, normalized values of the

checks having range between [-1,1] are used, the

overall equation comes out like:

=

[

1 − OP



1 − DE



+



OP. DE

]

.[Nvl

+

+CP+

+CTRL

+ PWR + ADJ + IntS

+ ES]/num_dimen’s (2)

4.3.2 Appraisal Profile and Emotion Label

In appraisal theories, it is usually assumed that there

is no direct relation between situation and specific

emotion. But somehow few of the appraisal theorists

manage to show the appraisal profile regarding some

basic emotions e.g. (Nezlek et al., 2008). Animate

organism have this evolutionary adaption process

which produce frequently recurring patterns of

environmental evaluation, which Scherer (1984,

1994) called as modal emotions. These modal

emotions result from specific SEC outcome, are

labelled in a single word according to certain social

and cultural similarities

The results of the appraisal process will not only

determine the type of emotion or blend of emotions

but also the intensity. So, the verbal reporting of the

feeling relies on language and the certain emotion

categories through different pragmatic devices cannot

produce the whole story. We can calculate

the label

by using any classification technique for example we

can find the Manhattan distance or K-Mean clustering

algorithm to find the nearest

modal emotion to the

appraisal profile, this part will be done in future wok.

Furthermore, CPM assumes emotion in continues

space of emotion as opposite to categorical emotions

(e.g happy, sad, disgust etc.). Scherer also provided

mapping between appraisal profile and emotion

labels, he called it as modal emotions .

5 SIMULATION AND RESULTS

A number of simulations are performed to prove the

below mentioned hypotheses.

H1: If the relevance is high then attention will be

devoted toward stimulus and the rest of the

checks will be processed.

H2: All the checks will be executed sequentially

according to CPM model. The order will be

relevance, implication, coping potential and

normative significance.

H3: If the relevance value is low no other check

will be processed.

H4: when Implication is low, lower activity is

shown at action preparation state.

H5: Low coping potential and normative

significance do not disturb the causality

among the appraisal checks and has not much

effect on action preparation.

Every simulation is performed for 120 time steps

with ∆t=0.1. The initial value of the check defines

the valance or strength of the check.

Table 3: States initial values for hypotheses 1 to 5.

Criterion

H1/H2 H3 H4 H5

Relevance

Novelty 1 0.1 1 1

Intrinsic

Pleasantness

1 0.1 1 1

Concern

Pertinence

1 0.1 1 1

Implication

Agency 1 1 0 1

Cause Motive 1 1 0 1

Outcome

Probability

0.9 0.9 0 0.9

Discrepancy from

Expectation

0.9 0.9 0 0.9

Conduciveness 0.9 0.9 -1 0.9

Urgency 0.9 0.9 0 0.9

Coping Potential

Control 0.9 0.9 0.9 0

Power 0.9 0.9 0.9 0

Adjustment 0.9 0.9 0.9 0

Normative

Significance

Internal Standards 0.9 0.9 0.9 0

External Standards 0.9 0.9 0.9 0

5.1 H1: High Relevance

The following values have been used as initial values

to determine the high relevance (see Table 3). Note

that all the other values are also kept high to show the

impact of control state. The control state acts a

monitoring and regulatory state which is used to

monitor the value of relevance.

The Fig. 4(a) shows the simulation with high

novelty, intrinsic pleasantness and concern pertinence

values, which in turns define high relevance. The

Temporal Causal Network Model for Appraisal Process in Emotion

353

simulation clearly shows that the relevance(red line)

is high at the start, but it takes a while to activate

attention (yellow). Once the attention is developed it

start processing the current stimulus, that’s why the

sensory representation state (dark blue) value starts

increasing after attention development. All the other

values checks are also so high but they are not

processed until the relevance get high.

5.2 H2: Ordering between Checks

The most important point is that the values of all the

checks are not processed until and unless the sensory

state value gets high and there is also the casualty

ordering among the checks; Relevance, Implication,

Comping potential, normative significance. In Fig.

4(a) it can be seen that there is so high values for all

the states at start of the simulation, but the casualty

between the states are intact.

Figure 4(a): The simulation of model with parameter for

hypotheses H1 and H2.

Figure 4(b): The simulation of model with parameter for

hypotheses H3.

5.3 H3: Low Relevance

To prove third hypotheses, we assigned very low

values to relevance checks and left other values

unchanged. The Fig. 4(b) shows that no states value

is executed because the relevance is so low and

according to Scherer it’s illogical and un economical

to process further state, if relevance is low.

5.4 H4: Low Implication

Implication check plays an important role in

behaviour preparation after any stimulus onset.

According to Scherer, any action taken is depended

initially on the value of implication appraisal check.

For higher value of implication in Fig 4(a) you can

see the preparation of action state get higher when the

value of implication get higher but with low value it

gets down. Simulation of low implication with values

given in Table 3 is shown in fig.5.

5.5 H5: Low Coping Potential and

Normative Significance

The sequence assumption, made in CPM is still valid

even when the coping potential and normative

significance gets low. The lower coping potential

value delay the process of normative significance.

The effect of coping potential and normative

significance on action preparation has not been

discussed in detail in CPM model, but the ordering

and the causality of the checks are elaborated. These

ordering can be seen in the fig. 6 below, which still

intact what’s so ever the values are. The values of the

state’s defining coping potential and normative

significance are set to zero.

Figure 5: The low implication value lowers the action

preparation state. Some line are made invisible to see the

clear effect.

Figure 6: The low coping potential and normative

significance keep casualty ordering intact. Some line are

made invisible to see the clear effect.

SIMULTECH 2018 - 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

354

5.6 Intensity

For differentiation between different affective states

or distinguishing mood and emotion, Intensity plays

an important role. Fig. 4 show the intensity for both

the appraisal profile defined in Table 3. These

intensities are calculated based on equation given at

section 4.3.1

Table 4: States initial values for hypotheses 1,2,3,4 & 5.

H1/H2 H3 H4 H5

Intensity

0.3918 0.1640 0.3918 0.2460

6 CONCLUSIONS

In this paper, a temporal causal network model has

been presented, which simulates the dynamics and

causality claimed in component process model of

appraisal by Scherer. This computational model is

designed in such away that, it can be embedded in any

cognitive or emotional architectures for agents. The

simulation clearly represents the causal relation

between the evaluation checks, high relevance of the

stimulus will leads to further processing of stimulus.

The high implication value will activate behaviour

responses state. The Damasio feeling for action has

been embedded which can be used for emotion

regulation in future. The intensity graph is separately

represented because it is not calculated over time. The

label states can be assigned to the given appraisal

profile by simple classification techniques.

In future, we will try some of emotion regulation

techniques proposed by (Gross, 1998), through

cognitive reappraisal. The feeling state will be used to

control the different emotion regulation strategies.

REFERENCES

Arnold, M. B. (1960) ‘Emotion and personality.’ Columbia

University Press.

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

agents with believable interactivity’, 319, p. 186.

Available at: https://becker-asano.de/Becker-Asano_

WASABI_Thesis.pdf.

Blackford, J. U. et al. (2010) ‘A unique role for the human

amygdala in novelty detection’, Neuroimage. Elsevier,

50(3), pp. 1188–1193.

Brosch, T. and Sander, D. (2013) ‘Comment: The

appraising brain: Towards a neuro-cognitive model of

appraisal processes in emotion’, Emotion Review, 5(2),

pp. 163–168. doi: 10.1177/1754073912468298.

Brown, M. W. and Bashir, Z. I. (2002) ‘Evidence

concerning how neurons of the perirhinal cortex may

effect familiarity discrimination’, Philosophical Trans-

actions of the Royal Society B: Biological Sciences.

The Royal Society, 357(1424), pp. 1083–1095.

Cunningham, W. A. and Brosch, T. (2012) ‘Motivational

salience: Amygdala tuning from traits, needs, values,

and goals’, Current Directions in Psychological

Science, 21(1), pp. 54–59. doi: 10.1177/0963721411

430832.

Damasio, A. R. (1998) ‘Emotion in the perspective of an

integrated nervous system1’, Brain research reviews.

Elsevier, 26(2–3), pp. 83–86.

Damasio, A. R. (1999) ‘The feeling of what happens: body

and emotion in the making of consciousness’. Harcourt

Brace.

Ellsworth, P. C. (2013) ‘Appraisal theory: Old and new

questions’, Emotion Review, 5(2), pp. 125–131. doi:

10.1177/1754073912463617.

Folkman, S. and Lazarus, R. S. (1985) ‘If it changes it must

be a process: study of emotion and coping during three

stages of a college examination.’, Journal of personality

and social psychology. American Psychological

Association, 48(1), p. 150.

Freese, J. L. and Amaral, D. G. (2009) Neuroanatomy of

the primate amygdala. Guilford Press.

Frijda, N. H. (1986) The emotions. Cambridge University

Press.

Gross, J. J. (1998) ‘The emerging field of emotion

regulation: an integrative review.’, Review of general

psychology. Educational Publishing Foundation, 2(3),

p. 271.

Hillyard, S. A., Teder-Sälejärvi, W. A. and Münte, T. F.

(1998) ‘Temporal dynamics of early perceptual

processing’, Current opinion in neurobiology. Elsevier,

8(2), pp. 202–210.

Kemper, T. D. and Lazarus, R. S. (1992) ‘Emotion and

Adaptation.’, Contemporary Sociology, 21(4), p. 522.

doi: 10.2307/2075902.

Lazarus, R. S. (1966) ‘Psychological stress and the coping

process.’ McGraw-Hill.

Marinier, I. I. I. R. P. (2008) A computational unification of

cognitive control, emotion, and learning. University of

Michigan.

Marinier, R. P., Laird, J. E. and Lewis, R. L. (2009) ‘A

computational unification of cognitive behavior and

emotion’, Cognitive Systems Research. Elsevier B.V.,

10(1), pp. 48–69. doi: 10.1016/j.cogsys.2008.03.004.

Marsella, S. et al. (2010) ‘Computational models of

emotion’, A Blueprint for Affective Computing-A

sourcebook and manual. Oxford University Press New

York, NY, 11(1), pp. 21–46.

Moors, A. et al. (2013) ‘Appraisal theories of emotion:

State of the art and future development’, Emotion

Review, 5(2), pp. 119–124. doi: 10.1177/1754073

912468165.

Nezlek, J. B. et al. (2008) ‘Appraisal-Emotion

Relationships in Daily Life’, Emotion, 8(1), pp. 145–

150. doi: 10.1037/1528-3542.8.1.145.

Temporal Causal Network Model for Appraisal Process in Emotion

355

Van Overwalle, F. (2009) ‘Social cognition and the brain: a

meta-analysis’, Human brain mapping. Wiley Online

Library, 30(3), pp. 829–858.

Pourtois, G. et al. (2010) ‘Temporal precedence of emotion

over attention modulations in the lateral amygdala:

Intracranial ERP evidence from a patient with temporal

lobe epilepsy’, Cognitive, Affective, & Behavioral

Neuroscience. Springer, 10(1), pp. 83–93.

Sacharin, V., Sander, D. and Scherer, K. R. (2012) ‘Levels

of valence’, Manuscript submitted for publication.

Sander, D., Grandjean, D. and Scherer, K. R. (2005) ‘A

systems approach to appraisal mechanisms in emotion’,

Neural Networks, 18(4), pp. 317–352. doi: 10.1016/

j.neunet.2005.03.001.

Schachter, S. and Singer, J. (1962) ‘Cognitive, social, and

physiological determinants of emotional state.’,

Psychological review. American Psychological

Association, 69(5), p. 379.

Scherer, K. (2004) ‘Feelings integrate the central

representation of appraisal-driven response

organization in emotion’, Feelings and emotions: The

Amsterdam…, pp. 136–157. Available at: http://books.

google.com/books?hl=en&lr=&id=t191L1qxYc8C&oi

=fnd&pg=PA136&dq=Feelings+Integrate+the+Centra

l+Representation+of+Appraisal-

driven+Response+Organization+in+Emotion&ots=OX

_1f_lMyA&sig=ql7Ymq6C-tyeocBHbN9JIXneBx8.

Scherer, K. R. (1984) ‘On the nature and function of

emotion: A component process approach’, Approaches

to emotion, 2293, p. 317.

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

multilevel sequential checking’, Appraisal processes in

emotion: Theory, methods, research. New York, NY,

92(120), p. 57.

Scherer, K. R. (2005) ‘What are emotions? and how can

they be measured?’, Social Science Information, 44(4),

pp. 695–729. doi: 10.1177/0539018405058216.

Scherer, K. R. (2009) ‘Emotions are emergent processes:

they require a dynamic computational architecture’,

Philosophical Transactions of the Royal Society B:

Biological Sciences, 364(1535), pp. 3459–3474. doi:

10.1098/rstb.2009.0141.

Scherer, K. R. (2013) ‘The nature and dynamics of

relevance and valence appraisals: Theoretical advances

and recent evidence’, Emotion Review, 5(2), pp. 150–

162. doi: 10.1177/1754073912468166.

Sperduti, M. et al. (2011) ‘Different brain structures related

to self-and external-agency attribution: a brief review

and meta-analysis’, Brain Structure and Function.

Springer, 216(2), pp. 151–157.

Treur, J. (2016) ‘Dynamic modeling based on a temporal-

causal network modeling approach’, Biologically

Inspired Cognitive Architectures. Elsevier B.V., 16, pp.

131–168. doi: 10.1016/j.bica.2016.02.002.

SIMULTECH 2018 - 8th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

356