TELKOM
NIKA
, Vol. 13, No. 4, Dece
mb
er 201
5, pp. 1162
~1
169
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i4.2737
1162
Re
cei
v
ed Au
gust 14, 20
14
; Revi
sed
No
vem
ber 1
4
, 2015; Accepte
d
No
vem
ber
25, 2015
Correlation of Student’s Precursor Emotion towards
Learning Science Interest using EEG
Norz
aliz
a Md Nor
1,2
, Abdu
l Wahab Bar
1
, Sheikh Hus
sain Shaikh Salleh
3
1
Departme
n
t of Computer Sci
ence, Kul
l
i
y
ya
h
of
Information & Commun
i
cati
on T
e
chnol
og
y (IIUM), 53100,
Kual
a Lum
pur, Mala
ysi
a
2
F
a
cult
y
of Bio
scienc
e & Medi
cal Eng
i
ne
eri
n
g, Univer
s
i
ti T
e
knol
ogi Ma
la
ys
ia (UT
M), 81310 Johor Ba
hru,
Johor, Mal
a
ysi
a
3
Center for Bio
m
edic
a
l Eng
i
n
eeri
ng, Univ
ers
i
ti T
e
k
nologi M
a
la
ysi
a
(UT
M),
813
10 Jo
hor B
ahru, Joh
o
r,
Malay
s
ia
e-mail: n
o
rzal
iz
a@ii
um.ed
u
.my, ab
dul
w
aha
b@
i
i
u
m.ed
u
.
my
, h
u
ssa
in
@
f
ke
.u
tm.my
A
b
st
r
a
ct
Mathe
m
atics
and
scie
n
ce
are tw
o
impo
rtant sub
j
ects
for stu
dents
to d
o
w
e
l
l
i
n
sch
ool.
Unfortun
ately ma
jority of
the
stude
nts ar
e
havi
ng
difficu
lti
e
s i
n
co
pin
g
w
i
th these s
u
b
j
ects. Mal
a
ysia
i
s
ranke
d
third l
o
w
e
st in the Pr
ogra
m
for Inte
rnatio
nal Stu
d
ent
Assessment
(PISA) fo
r
m
a
them
a
t
i
cs an
d
scienc
e. An e
m
oti
o
n
a
lly
dist
urbe
d stude
nt see
m
s to h
a
ve
probl
e
m
co
pin
g
w
i
th the le
ar
nin
g
of mathe
m
atic
s
and
scie
n
ce
thus
it is
i
m
por
tant to i
d
e
n
tify the
de
motiv
a
t
i
ng
factors aff
e
cting
the
perf
o
rmanc
e of s
u
ch
students. In th
i
s
pa
per, ita
nal
y
z
e
the
c
o
rrel
a
tion
of prec
urs
o
r e
m
oti
on t
o
w
a
rds stu
dent
in
terest in
le
arni
ng
math
e
m
atics a
nd
scie
n
ce usi
ng
el
ectroe
nce
pha
logr
a
m
(EE
G
) device. T
h
is
correlati
on a
n
d
their res
pecti
ve
emotio
n can b
e
ana
ly
z
e
d
ba
sed on th
e 2-
D Affective
Space Mod
e
l (A
SM) using four
basic e
m
oti
o
n
s
of
hap
pin
e
ss, cal
m
n
e
ss, fear an
d sad
ness as r
e
ferenc
e
stimul
i. EEG device
w
a
s used to ex
tract brain w
a
v
e
s
sign
al w
h
i
l
e
an
sw
ering th
e
mathe
m
atics
a
n
d
sci
ence
q
u
e
s
tions. T
h
e
EE
G sign
als w
e
r
e
ca
pture
d
o
n
the
scalp
of the st
ude
nt an
d fe
atures
extract
e
d
usin
g Me
l F
r
equ
ency
Ceps
tral Co
efficie
n
t (MF
CC). Neu
r
a
l
netw
o
rk classifi
er of Multilayer
Perceptron (M
LP) w
a
s us
ed to classify the vale
nce an
d aro
u
sal ax
es for the
ASM.Pre
l
i
m
i
nary re
su
l
t
s sh
ow th
e
re
la
ti
onsh
i
p o
f
p
r
ecur
sor e
m
otion
a
nd th
e
dyna
mi
c e
m
oti
ons
of
the
student w
h
il
e takin
g
the
mat
h
e
m
atics a
nd
scienc
e tes
t. W
e
hope th
at these resu
lts can he
lp us furt
her
relate the b
e
h
a
v
ior an
d intere
st of students tow
a
rds the le
a
r
nin
g
of math
e
m
atics a
nd sci
ence.
Ke
y
w
ords
: pre
c
ursor e
m
oti
o
n
,
student, MF
CC, neura
l
netw
o
rk, MLP
Copy
right
©
2015 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Although n
u
m
ber
of studi
es h
ad a
naly
z
ed th
e deve
l
opmental t
r
e
nd toward
s
student’s
ability-rel
a
ted
beliefs in ma
thematics but
few had
fo
cuse
d on
stud
ent’s inte
re
st toward lea
r
ni
ng
in mathem
atics. Ko
¨lle
r et
al. [5] did a lo
ngitudi
n
a
l stu
d
y by analy
z
ing stu
dent’
s
i
n
tere
st toward
s
mathemati
cs
in a German
high-ability track scho
ol
s
(Gymna
sium)
at three time
points (end
of
grad
e 7,
en
d of
grade
10,
and middle of
grade 12
). Student intere
st in mathematics wa
s foun
d
to be cru
c
ial
in education
field as a fou
ndation to
other future co
urses. Mo
st of the studen
ts
were having
difficulties
in learni
ng mat
hematics.
Answering mathematics
questions is a risky
decision
ma
king process and pu
ts th
e studen
t u
nder stress. Each a
n
sw
er mad
e
b
y
the
studen
t w
ill
affec
t
his
or
her resul
t
s
in the
ma
the
m
a
t
ics tes
t
and
their
precursor emo
t
ion
seems pl
ays
a critical role.
If such
precursor emo
t
ion c
o
n
t
inues d
u
rin
g
the l
earni
ng
exp
e
rience i
n
class this
ma
y affect the st
uden
ts learning e
x
pe
rience thus resulting in p
oor
perfor
m
ance
.
Although the
chan
ge in
student’s inte
rest in lea
r
nin
g
may not d
epen
d sol
e
ly on the
stude
nt’s cu
rrent emotional
stat
e but also it also affected on the en
vironme
n
t and the student’
s
prio
r em
otion
.
Acco
rdin
g t
o
Cla
r
k [1], e
m
otions
are
central to h
u
m
an motivati
on, whi
c
h i
n
clude
the precurso
r emotion. In f
a
ct
p
r
e
c
urso
r emotion wa
s also kn
own
to
affect stu
den
t behaviou
r
[7
].
Thus stu
dent
pre
c
u
r
sor
em
otion may infl
uen
ce th
e
stu
dent’s
dynam
ic em
otion a
n
d
their interest
towards le
arn
i
ng mathem
atics.
There have
been m
any rese
arche
s
co
nce
p
tual
i
z
ing
emotion al
o
ng two
dime
nsio
ns
of
valence, which descri
b
e
s
the ext
ent of
plea
sure or sadne
ss and a
r
ou
sal de
scri
bes the exten
t
of
c
a
lmness
or
exc
i
tation [2,
3].
Yet
tremendo
us wo
rk
has
b
een a
c
compli
she
d
based
on ne
ural
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 116
2 – 1169
1163
respon
se
s u
nder me
mory formation
of emotio
nal
events,
ho
wever the
st
udy on
p
r
e
c
ursor
emotion effe
cts by usi
n
g
Electroe
nce
phalo
g
ra
m
(EEG) is sca
r
ce [7]. In this pap
er, a n
o
ve
l
approa
ch
ha
s be
en
pro
p
o
se
d to a
nal
yze an
d u
n
d
e
rsta
nd th
e student be
hav
ior b
a
sed o
n
the
affective sp
a
c
e m
odel
(A
SM) that en
a
b
les
emotion
s
to be
viewed in two different
axes
of the
valence (V) a
nd aro
u
sal (A
) as the e
s
se
nt
ial basi
s
fun
c
tion a
s
sh
own in Figure 1.
Figure 1. The
Affective Sp
ace Mo
del wit
h
the Diffe
ren
t
Position of Basic Emotio
ns with Emoti
on
Primitives Axis x for Valence, and y for Arou
sal [6]
The val
e
n
c
e
arou
sal
(V
A) ap
proa
ch
ca
n
also h
e
lp to
an
alyze
and
visu
alize
the
pre
c
u
r
sor an
d dynamic e
m
otions of th
e student
in
answe
ring th
e mathemati
c
s que
stion
s
. It can
also reveal the co
rrel
a
tio
n
s between
pre
c
u
r
so
r e
m
otion and the dynami
c
emotion
s
of the
stude
nt’s inte
rest. In
additi
on, the
study
of the
stu
d
e
n
t’s dyna
mic
emotion
s
whil
e an
sweri
ng t
he
variou
s math
ematics qu
estions
can p
r
ovide a
bett
e
r un
derstan
ding in
a
naly
z
ing the
stud
ent
intere
st. Hen
c
e the effect
on pre
c
u
r
so
r emotion to
wards e
a
rly
detectio
n
of highly emotio
nal
agitated stu
d
ent can b
e
id
entified to detect t
he s
t
udent interes
t
towards
mathematic
s
.
2. Experimental Setup
To e
n
sure
a
prop
er conn
e
c
tion
s a
n
d
pl
acem
ent
of the el
ect
r
od
es on
the
scalp
of the
partici
pant
s, Nup
r
ep el
ect
r
o-gel was u
s
ed to cl
ean
the scal
p
surface. In order to ma
ke
the
sen
s
o
r
s
stick well onto the
scal
p
, the Ten 20 TM
co
ndu
ctive gel wa
s used. Th
e visco
sity paste
is need
ed to ensure that the gel will no
t flow
easily and yet it can be easily remove. Besi
des,
Ten20
TM co
ndu
ctive also
helpe
d to further
redu
ce th
e impeda
nce on the se
nsors.
Data were
collecte
d
and
divided into two pa
rts. Firstly in orde
r to derive the
ASM for
each in
dividu
al ba
si
c e
m
otion
stimuli
we
re u
s
ed
while
EEG sig
nal
s
reco
rde
d
. Se
condly
while
th
e
stude
nt wa
s
answe
ring
ea
ch qu
estio
n
the EEG si
g
n
als were also
record
ed. T
he expe
rime
ntal
desi
gn flow a
nd proto
c
ol i
s
sho
w
n in Fig
u
re 2 a
s
a ge
nerali
z
e exp
e
riment
s.
Notice from
Figure 2 both t
he eyes
open and eyes
close
w
ill provide initial inf
o
rmation
about
the em
otional state of
the
stu
den
t. To ensure
a prope
r initi
a
lizatio
n of th
e brain a
c
tivity
durin
g eyes o
pen task, stu
dent will be looki
ng at
a blank white screen. The four basic em
otio
ns
movie clip
wil
l
then be di
splayed for
on
e and h
a
lf minute pe
r mov
i
e clip
rep
r
e
s
enting em
otio
n
happi
ne
ss, fe
ar, calmne
ss
and
sad
n
e
ss.
After ea
ch
movie cli
p
, the stu
dent
wa
s requi
red
to
fill
up the Self
-Assessme
n
t M
aniki
n (SAM
). Finally
stu
d
e
n
t wa
s requ
e
s
ted to
acco
mplish
two ty
pes
of tests co
nsi
s
ting of math
ematics and
sci
en
ce qu
est
i
ons.
In this pape
r, student wa
s re
quired to
answe
r 12
que
stion
s
of mathemati
c
s and 10
que
stion
s
of sci
en
ce. Thre
e levels of q
uestio
n
s rang
ing from ea
sy, medium and difficult were
desi
gne
d. Student was
req
u
ired to
an
swer the e
a
sy
q
uestio
n
s i
n
1
0
se
co
nd
s, medium q
u
e
s
tio
n
s
in 20 second
s and
difficult
questio
n
s i
n
30 se
co
nd
s for mathe
m
ati
cs te
st. Whe
r
eas fo
r scien
c
e
Arousal
Excit
e
d
HAPP
Y
Deli
ght
e
d
Ple
a
s
e
d
CAL
M
Rel
a
xed
Tense
FEAR
An
no
yed
Depresse
d
SAD
Tired
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Correl
ation of
Student’
s
Precu
r
sor Em
otion towa
rd
s L
earni
ng Sci
e
n
c
e …
(N
or
zali
za Md
Nor
)
1164
test, easy qu
estion
s in 15 se
con
d
s, me
diums q
u
e
s
tions in 30 second
s and difficult que
stion
s
in
40 se
co
nd
s. All question
s
were display
ed on the
p
e
rson
al co
mput
er screen a
n
d
stude
nt ne
eds
to verbally an
swer the qu
e
s
tion.
Figure 2. Experime
n
tal set
up and p
r
oto
c
ol
2.1. Stimuli
The em
otion’
s movie
clip
s were u
s
ed
to obtain
emo
t
ional re
sp
on
se
s an
d on
e
set of
mathemati
c
s test an
d
sci
ence test to
identify th
e
stud
ent inte
rest. F
our b
a
si
c em
otion
s
of
happi
ness, fear, calmness
and
sadness from the Internationa
l Affective Picture (IAPS)
were
use
d
to ge
ne
rate the ASM
referen
c
e
s
o
f
each
stud
en
t. Bernard Bo
uch
a
rd’
s
synt
hesi
z
e
d
mu
si
cal
clip
s an
d G
r
o
s
s an
d Leve
n
s
on’
s m
o
vie
clips
we
re
use
d
to eli
c
it em
otional
re
spo
n
se
s [4] p
r
io
r to
stude
nt doing
the mathema
t
ics an
d scien
c
e test
s.
2.2. Participants
15 healthy st
udent
s (8 fe
male and 7
male)
were recruited from
Sekolah Ke
bang
sa
an
Taman
Unive
r
siti 1. Students we
re cho
s
en fro
m
fi
rst
,
seco
nd and
third grad
e. These stud
e
n
ts
must be
withi
n
the age
of 10 to 11, si
n
c
e the ta
rget
subj
ect mu
st
be familiar
with the p
r
im
ary
school’
s
sylla
bus tau
ght.
3. Correla
tio
n
of Stude
nt’
s
Precurso
r
Emotion and
Learning Interes
t
Figure 3 sh
ows the blo
ck di
agram
of the experiments in
o
r
der toe an
al
yse the
correl
ation of pre
c
u
r
sor em
otion to stude
nt’s l
earning i
n
tere
st. In th
e pre-processing stage n
o
i
s
e
and oth
e
r
art
e
facts will
be
remove
d fro
m
the raw E
E
G sig
nal
s u
s
ing th
e ellip
ord filter. F
e
a
t
ure
s
will then
be
e
x
tracted
usi
n
g the M
e
l-fre
quen
cy Ce
p
s
tral Coefficie
n
t (MF
C
C) m
e
thod a
nd fin
a
lly
usin
g the Mul
t
ilayer perce
p
t
ron (M
LP) to
cla
ssify the valen
c
e an
d arousal.
Brief from
the experimenter
Electrode’s
placement
Answering Math
em
atics qu
estion
Emotion movie
cli
p
10
minutes
per
t
ask
1
m
inutes
e
ach
Ey
e
s
cl
ose
Ey
e
s
o
p
e
ne
d
1
m
inutes
e
ach
SAM
Rest
Ans
w
ering S
c
ie
nce qu
es
tion
10
minutes
per
t
ask
Ha
ppy
Sad
Fea
r
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 116
2 – 1169
1165
Figure 3. Block
Diag
ram o
f
the experim
ent
3.1. Featur
e Extrac
tion (MFCC)
Based
on
ou
r previo
us stu
d
y [7], in this
pape
r featu
r
e
s
were al
so
e
x
tracted
usi
n
g MF
CC
to extract the low freq
ue
ncy EEG sig
nals. In
ste
a
d
of using 40
MFCC
coeff
i
cient
s like m
o
st
feature
extra
c
tion fo
r
spe
e
ch,
here onl
y 10 MF
CC
coefficient
s were
extra
c
ted
and fo
und
to
be
sufficient. Thus
every i
n
st
ance
will have 160 feat
ure from
16 channel
s based
on 256
nfft points
at 83Hz sam
p
ling freq
uen
cy and 20% o
v
erlap.
Table 1. Training pa
ram
e
ters fo
r MLP
No. of hidden la
yer
1
No. of neu
ron in t
he hidden la
y
e
r
10
Activation functio
n
for hidden la
ye
r
tan-sig
Activation functio
n
for outpu
t la
y
e
r
purelin
Learning r
a
te
0.01
Mean-squar
e er
r
o
r goal
0.1
3.2. Classific
a
tion (MLP
)
MLP wa
s a
dopted a
s
the cla
s
sifier in orde
r
to classify the extracted feature
s
to
investigate t
he pre
c
u
r
sor emotion an
d its dynam
i
c
. A
feed forward artifici
al
neural n
e
twork
RE
S
U
L
T
Precursor Emotion (M
L
P
)
Testing
(E
y
e
s Closed Data
)
Classif
i
cation (MLP
)
Training
Emotion Dat
a
(
VA)
Bas
i
c
emotion
stimuli
(I
A
PS)
Math &
Sc
ie
nc
e
Q
uestion
Feat
ur
e Ex
t
r
act
i
on
(
M
F
CC)
EEG Sig
n
als
(P
re-P
roc
e
ssi
ng
)
EEG Sig
n
als
(R
aw Dat
a
)
Data Collection
Noise
f
ilte
r
(
E
llipor
d F
ilte
r
)
Ey
e
s
O
p
e
n
and E
y
e
s
Close
d
Dat
a
Colle
c
t
ion
Pre-P
r
ocessin
g
F
e
a
t
ure Ext
r
action
Classif
i
cation
RE
S
U
L
T
Student B
e
haviour(M
L
P
)
Testing
(Math Question)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
9
30
Correl
ation of
Student’
s
Precu
r
sor Em
otion towa
rd
s L
earni
ng Sci
e
n
c
e …
(N
or
zali
za Md
Nor
)
1166
model, multi
-
l
a
yer p
e
rcept
ron ma
ps
set
s
of input
d
a
ta
onto a
set
of app
r
op
riate
output. Optim
a
l
model
sel
e
cti
on for the
nu
mber of laye
r and
the
nu
mber of n
euron n
e
ed
ed fo
r the
be
st M
L
P
architectu
re i
s
req
u
ired to ensure o
p
tim
u
m perfo
rma
n
ce.
3.3. Questio
ns for th
e Te
st
In this re
se
arch, we have
been
req
u
e
s
ting the
subje
c
t to answer
8
sci
en
ce q
u
est
i
ons fo
r
the test.Table
2, depicts th
e que
stion
s
that has b
een
prep
ared for the su
bje
c
t.
Table 2. Que
s
tion for the
subje
c
ts
No. Question
1
X,
Y
and
Z a
r
e th
e parts involved in breathing.
X : nose
Y
:
w
i
ndpipe
Z : lungs
Which of the follow
i
ng is the corre
c
t order of th
e pa
ssage of air
w
h
e
n
we exhale?
A X
Y
Z
C Z
Y
X
B X
Z
Y
D
Y
X
Z
2
What is the nameof thiscomponent?
Asw
i
tchCBattery
BResistorsDWires
3
What isthe sy
mb
olofthe Figure b
e
l
ow
?
AlightCResistor
BBatteryDs
w
itch
4
Among theseries circuitanda parallel circui
t, w
h
ichw
illsw
i
tch on theli
ghtsw
i
thb
righter?
ACircuitSerialCBothcircuits
BCircuitParallelDNotbothcircuits
5
Living thingsneed energ
y
to
ABreathingCRu
n
n
ing
BWalkingDAll oft
he above
6
Lighted candlesproduce ene
rg
y
Alight energ
y
a
n
d
thermal energ
y
BElectricity
andth
ermal energ
y
Ckinetic energ
y
a
ndlight energ
y
Energ
y
Dsounda
ndlight energ
y
7 The
phon
erang
g
enerateen
erg
y
AKinetic Energ
y
CElectricity
Energ
y
BSoundEnerg
y
D
Light Energ
y
8
The Figu
re sho
w
sapartofthe hu
m
an bod
y
Whatpartma
r
ked
X
?
ANoseCStomach
BLungDAir Duct
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 116
2 – 1169
1167
4. Results a
nd Discu
ssi
on
In this
pap
er
only on
e stu
d
ent data
were
analy
s
ed as a
prelimina
r
y results.
Eye
s
clo
s
ed
data we
re first analyze
d
to identify the precurso
r e
m
otion. Precursor em
otio
n for stude
nt 1 is
sho
w
n in Fi
g
u
re 4
with co
mbination
of calm
a
nd sad
.
Student see
m
s to have b
o
th positivea
nd
negative val
ence an
d ne
gative aro
u
sal while th
e
eyes
were cl
ose
d
. This
could me
an t
hat
stude
nt wa
s not sure wh
at to feel and
the brain i
s
still having lo
w activation.
Figure 4. Pre
c
urso
r emotio
n of student 1
In Figure 5,
it sho
w
s that
the stu
dent
s s
eem to
b
e
cal
m
emoti
on at the
be
ginnin
g
,
having sad e
m
otion in the
middle and
calm em
otio
n again to
wa
rds th
e end.
It has positi
v
e
valence
an
d negative
a
r
o
u
sal whi
c
h showi
ng calm
emotion. It al
so h
a
s
neg
ative valence a
nd
negative
a
r
o
u
sal whi
c
h
showi
ng sad
emotion.
T
h
is
indi
cate
s th
at the p
r
e
c
ursor emotio
n
cal
m
and sad ha
s i
n
fluen
ce the
subj
ect while
answe
ring th
e que
stion.
Figure 5. Student’s dyn
a
m
i
c emotion
s
while an
swerin
g first que
stio
n
In Figure 6,
it depict
s n
egative vale
nc
e and neg
ative
arou
sal
whi
c
h sho
w
ing
sad
emotion. The
subje
c
t may not have intere
st in
answeri
ng the seco
nd que
sti
on due to difficult
que
stion. Thi
s
re
sult al
so
arise in q
uestion
numb
e
r seven whe
n
the
subje
c
t
al
so sho
w
ing sad
emotion which has b
een
shown in Figu
re 7.
Figure 6. Student’s dyn
a
m
i
c emotion
s
while an
swerin
g se
con
d
que
stion
calm
cal
m
sad
sad
sad
sad
sad
cal
m
cal
m
cal
m
cal
m
cal
m
sad
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Correl
ation of
Student’
s
Precu
r
sor Em
otion towa
rd
s L
earni
ng Sci
e
n
c
e …
(N
or
zali
za Md
Nor
)
1168
Figure 7. Student’s dyn
a
m
i
c emotion
s
while an
swerin
g third que
sti
o
n
In Figu
re
8
a
nd Fi
gure 9,
the subje
c
t i
s
having
ne
ga
tive valence
and
po
sitive
arou
sal
whi
c
h sh
owi
n
g fear emoti
on. This is al
so reve
aling
that the subj
ect may not have intere
st
in
answe
ring th
e fourth and fi
fth question.
Figure 8. Student’s dyn
a
m
i
c emotion
s
while an
swerin
g fourth que
st
ion
Figure 9. Student’s dyn
a
m
i
c emotion
s
while an
swerin
g fifth questio
n
In Figure 1
0
, 11 and 1
2
, the subj
ect is
showi
ng ne
gat
ive valence a
nd neg
ative arou
sal
whi
c
h in
dicate sa
d em
otio
n. The
su
bje
c
t seem
s to
be sho
w
ing
a
negative
em
otion towards the
end of th
e q
uestio
n
s. Pe
rhap
s the
su
b
j
ect do
es not
unde
rstand t
he qu
estio
n
or m
a
y not h
a
ve
intere
st in an
swerin
g scien
c
e qu
estio
n
.
Figure 10. Student’
s
dyna
mic emotio
ns
while an
swering sixth que
stion
sad
fear
fear
sad
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 116
2 – 1169
1169
Figure 11. Student’
s
dyna
mic emotio
ns
while an
swering seve
nth q
uestio
n
Figure 12. Student’
s
dyna
mic emotio
ns
while an
swering eight qu
e
s
tion
5. Conclusio
n
In con
c
lu
sion
, the student
’s dynami
c
e
m
otions
sh
o
w
a domi
n
a
n
t negative
emotion
towards an
sweri
ng the
scien
c
e te
st. This
co
uld
b
e
due to th
e
inca
pability
of the stud
en
ts to
answe
r corre
c
tly by which
difficult que
stion
s
se
em
s to indicate
negative em
otions. Stude
nt’s
pre
c
u
r
sor
em
otion of
sad i
s
reflecte
d th
roug
h o
u
t experim
ents i
n
dicatin
g
an
i
m
porta
nt rol
e
of
pre
c
u
r
sor em
otion. Con
s
e
quently, the pre
c
u
r
sor
em
otions h
a
ve
a major i
n
flu
ence in hum
an
emotion
sin
c
e it ha
s exi
s
ted in
ou
r me
mory for a l
o
ng time. Alth
ough
only o
n
e
stu
dent EE
G date
wa
s a
nalyze
d
it sh
ows the
potential
of correlating pre
c
urso
r emotio
n to lea
r
ni
ng i
n
tere
st. Re
su
lt
from our ex
perim
ent sh
owe
d
that it is possi
ble
to identify
the student
intere
st towa
rds
mathemati
cs according
to pre
c
u
r
sor em
otion
an
d
stu
dent’s
dynam
ic em
otion
s
. In future,
we
are
intere
sted to cla
ssify the e
m
otion and
student
be
havi
o
r by usin
g fuzzy ne
ural n
e
t
work.
Ackn
o
w
l
e
dg
ement
This work is
suppo
rted by the Mi
nist
ry of Education, M
a
laysia.
Referen
ces
[1]
Clark DE. T
he Affective R
easo
ner: A P
r
oce
ss Mod
e
l
of Emotions
in a Mu
lti-A
gent S
y
stem.
Unp
ubl
ishe
d d
o
ctoral thes
is, North
w
e
s
tern
Univers
i
t
y
, C
h
i
c
ago.1
9
9
2
.
[2]
Mehra
b
ia
n A & Russel
l
JA. An appr
oac
h to envir
onme
n
t
a
l ps
ych
o
lo
g
y
,
Cambri
dge,
MA, USA: MIT
Press.187
4.
[3]
Lan
g PJ. T
he three s
y
stem
a
ppro
a
ch to
em
oti
on. In B
i
rba
u
mer N & O
h
man A. T
he Organ
izatio
n of
Emotion (Ed.). T
o
ronto: H
ogre
f
e-Hub
e
r, 199
3
;
18-30.
[4
]
C
h
an
e
l
G, Kron
e
g
g
J, Gra
n
j
ea
n
D
& Pun
T
.
Emo
t
i
o
n
a
sse
ssme
n
t: Aro
u
s
al e
v
a
l
u
a
t
i
o
n
u
s
in
g
EEG’
s
and
peri
pher
al ph
ysi
o
lo
gic
a
l sig
nals. Proce
e
d
i
ngs
Intern
ati
ona
l W
o
rksho
p
on Multim
edi
a Conte
n
t
Repr
esentati
o
n
,
Classificati
on
and Sec
u
rit
y
, I
s
tanbu
l, 200
6: 530-
537.
[5]
Ko¨ller O, Baumert J, & Sch
nabel K. Does
in
terest matter? T
he relati
onsh
i
p b
e
t
w
ee
n acad
em
i
c
interest an
d ac
hiev
em
ent i
n
mathematics.
J
ourn
a
l for Res
earch i
n
Mathe
m
atics E
ducati
o
n
. 200
1; 32
:
448
–4
70
[6]
Russell JA. A
circum
pl
e
x
mo
del
of affect.
Journ
a
l
of Pers
ona
lity a
nd S
o
cial Psyc
hol
og
y
. 198
0; 39
:
116
1-11
78.
[7]
W
ahab
A, Ka
marud
d
in
N,
M Nor
N, Abu
t
H.
“Pre- a
n
d
Postacci
dent
Emotion
Ana
l
ysis on
Driv
in
g
Behav
ior”.
Smart Mobil
e
In-Vehicl
e Syste
m
s
. 2014; 22
5-23
9
sad
sad
Evaluation Warning : The document was created with Spire.PDF for Python.