‘I Ain’t Like You’ A Complex Network Model of Digital Narcissism

Abstract

Social media like Twitter or Instagram play a role of fertile ground for self-exhibition, which is used by various narcissists to share their frequent updates reflecting their narcissism. Their belief of saving and assisting others, make them vulnerable to the feedback of others, so their rage is as dangerous as their messiah complex. In this paper, we aim to analyse the behaviour of a narcissist when he is admired or receives negative critics. We designed a complex adaptive mental network model of the process of narcissism based on the theories of neuroscience and psychology including a Hebbian learning principle. The model was validated by analyzing Instagram data.

Full text

‘I ain’t like you’
A Complex Network Model of Digital Narcissism
Fakhra Jabeen1 , Charlotte Gerritsen 2 , and Jan Treur2
1,2 Vrije Universiteit Amsterdam, The Netherlands
fakhraikram@yahoo.com, [j.treur, cs.gerritsen] @vu.nl
Abstract. Social media like Twitter or Instagram play a role of fertile
ground for self-exhibition, which is used by various narcissists to share
their frequent updates reflecting their narcissism. Their belief of saving
and assisting others, make them vulnerable to the feedback of others, so
their rage is as dangerous as their messiah complex. In this paper, we
aim to analyse the behaviour of a narcissist when he is admired or re-
ceives negative critics. We designed a complex adaptive mental net-
work model of the process of narcissism based on the theories of neuro-
science and psychology including a Hebbian learning principal. The
model was validated by analyzing Instagram data.
Keywords: Narcissistic Rage, Narcissism, Complex mental network.
1 Introduction
My character has ever been celebrated for its sincerity and frankness,
and in a cause of such moment as this, I shall certainly not depart from
it. (Pride and Prejudice:56)
Narcissism was addressed by various fiction characters like Lady Catherine
(Pride and Prejudice), who always tried to be a savior and was overly con-
cerned about how others think of her. Now in the age of technology, they can
be observed by people who excessively use social media, like Twitter or Insta-
gram [1], to share their lifestyle updates. These online communities are target-
ed as fertile grounds for self-presentation and getting the appraisal, especially
by millennials, who tend to use whatever chrism and social skills to become
quantifiers or influencers [1]. The exponential growth of audience makes nar-
cissists more vulnerable to negative feedback, and usually make them unhappy
about others [2]. Narcissistic rage is a psychological construct that addresses a
negative reaction of a narcissist person, when he or she assumes that his self-
worth is in danger.
Narcissistic rage is a common outcome due to lack of empathy, and it is the
more related outcome to an ego-threatening action rather than self-esteem. It is
to be made clear, that this is not related to self-esteem, but to the desire of self-

2
admiration [3]. There are two types of narcissism, discussed in the literature:
‘overt’ or ‘covert’. They both share grandiose fantasies, feeling of superiority,
but covert narcissists, don’t reveal their true self, thus may show aggression
towards others [4]. In literature, different studies were conducted with respect
to the psychology and neuroscience of a narcissist in certain surrounding con-
ditions [3, 5]. Also, Artificial intelligence is used to identify a narcissist [6].
However, no study has been presented in the domain of Mental Network
Modeling, which can address: a) the mental organization of a narcissist, b)
how his mental processes learn from experience, c) how to relate a presumed
narcissist with his or her cognitive behavior by taking social feedback into
account, with respect to his social interaction.
In this paper, we present a complex-adaptive network model of a narcissist,
based on psychological and neurological studies. We address non-trivial inter-
action of a narcissist brain a) during a reward-seeking behavior, and b) a con-
sequence of an unwanted remark. We validated our model by case studies.
Section 2 follows related work while section 3 presents the model of a narcis-
sist. Section 4 discusses the simulation scenarios, while Section 5 validates
the model using public data through Instagram. Section 6 concludes the paper.
2 Related Work
This section explains the psychological and, neurological perspective of a
narcissist. On the one hand, it provides literature: how a narcissistic behaves
when a) admired or b) negatively criticized. On the other hand, it will address
the problem in relation to complex networks and artificial intelligence.
Psychologically, a narcissist exhibits a higher tendency for self-presentation
[2]. Many studies show an association between narcissism and reward-seeking
behavior [3]. A survey indicates that Instagram is widely used service among
other social networks, to exhibit grandiose narcissism. A narcissist who re-
ceives added appreciation often appears to be compassionate and happy [1]. A
survey showed that people, who are not satisfied with their appearance are
more vulnerable to anger, due to lack of empathy. As a result, narcissists are
prone to bullying or violence [3, 5].
In a cognitive and neurological aspect, we would like to discuss cognitive
parts of brain, hormones, along with the neurotransmitters, which process the
self-relevant stimuli. A narcissist seeks admiration, which activates brain re-
gions, like the Prefrontal Cortex, Anterior Cingulate Cortex (ACC), anterior
Insula and Temporal lobe, which is strengthened during self-enhancement and
mentalizing [7]. High activations in the anterior insula indicate focus on one-
self or representing selfishness [5]. It is indicated that ventral striatum is in-
volved with ACC, and get activated during reward-seeking behaviour. It de-
pends on pre-synaptic and post-synaptic activations along with dopamine re-
lease. The facial attractiveness strengthens the synaptic transmission due to
contingent feedback and dopaminergic projections [8]. ACC is also related to

3
negative emotional valence which may lead to aggression. Also, stress faced in
a social competitive environment makes the brain more vulnerable to experi-
ence anxiety [9].
Like in reward-seeking, hormones and neurotransmitters also play a role in
aggression. For instance, noticeable levels of progesterone, testosterone and
low levels of corticosterone help in mediating aggression along with γ-
aminobutyric acid (GABA) receptors activation, due to anxiety [10]. De-
creased levels in 5-serotonin (5-HT) leads to aggression. Further, vascular
endothelial growth factor (VEGF) is a signal protein in the hippocampal re-
gion, which is effected in psychological stress. Damage in its microstructure
and decrease in synaptic connections, influence the release of neurotransmit-
ters in the hippocampal region during stress. This decrease in synaptic plastici-
ty hinders the message transfer to the central nervous system along with the
changes in the brain structure, learning, and memory [10]. Long term effects
of stress can lead to long-term genetic and epigenetic metaplastic effects [11].
Also, a presynaptic receptor 5-HT1A, located in 5-HT has a greater density in
people with high aggression, along with brain regions that are related to im-
pulsive control [12].
A study was also presented, which incorporated machine learning tech-
niques to identify a narcissist [6]. A temporal causal model is `discussed in the
context of esteem [13]. However, no research was found, which could address
the vulnerability in a narcissist and his reaction over certain feedback.
3 Complex Adaptive Mental Network Model of a Narcissist
In this section, a complex adaptive mental network of a narcissist is presented
based upon studies addressed earlier, using a multilevel reified architecture[14,
15]. This is a layered architecture, where the temporal and adaptive dynamics
of a model are represented in layers, from the base model to the evolution of
the complex network by first and second order adaption principles, along with
its mathematical representation.
A base temporal-causal model refers to a conceptual representation of a
real-world scenario, depicted by states and connections where a connection
designates a causal relationship among states. For example, consider two
states X and Y, if Y is affected by X then XY is a causal relationship. More
specifically, the activation value of Y is the aggregated impact of all influenc-
ing states along with X. The influencing states have different activation levels
and connection weights to influence Y, that has a speed factor indicating the
timing of influence. Such a temporal-causal model is characterized by [16]:
Connection weight ωX,Y indicates how strong state X influences state Y. The magni-
tude varies between 0 and 1. A suppression effect is categorized by a negative con-
nection.
Speed factor ηY indicates that how fast a state Y changes its value upon a causal
impact; values range from [0-1].

4
Combination function cY(..) is chosen to compute the causal (aggregated) impact
of all incoming states (Xi : i = 1 to N ) for state Y. Certain standard combination
functions are defined already to compute the aggregated impact of Y.
In Fig. 1, layer I presents the base model with 38 states, depicting the mental
organization and psychology of a narcissist. It shows reactions of a narcissist
when he receives admiration or negative criticism. The extended structure of
the model is also explained by Layer II and III. A concise explanation of each
state in figures is specified in Table 1 and Table 2.
In layer I, a narcissist receives a feedback while using social media
(wss,sss), after sharing his picture or a status. Feedback can be a positive (wspf,
sspf, srspf) or a negative (wsnf, ssnf, srsnf) remark which can make him happy or
he can feel hurt. On receiving a compliment like ‘you are awsome’ his self-
belief (bs+) evaluates it as a positive remark (eval+), thus leads to a happy
reaction (eshappy: i.e. a gratitude). Brain related parts (striatum , pfc and insula)
get activated more than usual, along with feelings of self-reward (fsreward) and
self-love (fslove). This behavior increases by experiencing the same kind of
feedbacks over the time (adaption / learning).
I
III
II
wspf
striatum
sspf
wss
wsnf
srspf
sss
ssnf
eval+
srspf
bs+
fslove fsreward
PFC
insula
cs
eshappy
eval-
os
psact
wsang ssang srsang psang
fsang
esang
val
esact
wseff sseff srseff
fsemp psemp
wsanx ssanx srsanx psanx
fsanx
_
_
_
Wfslove,bs
Wfsang,psa
Wfsrew,sat
Wpsa,srseff
Wbs,fslove
Weval,psa Wsat,ins
M
H
Fig. 1. Reified Network Architecture for a Narcissist person.
Upon a negative critic (wsnf, ssnf, srsnf), a narcissist usually disagrees due to
high ego/self-belief, and gets angry, which also induces anxiety in him. To
explain it further, consider a poor remark ‘you are ugly’, activates state (eval-).
Self-belief (bs+) tries to suppress eval- through control state (cs). However,
eval- is too strong to be suppressed. It has dual influence: a) It stimulates an-

5
ger (psang) along with its body loop (wsang; ssang; srsang; fsang ; psang; esang), e.g
‘a raised eyebrow’ and, b) preparation state of action (psact) is also activated by
an aggregated impact of eval-, val and fsang. Here ‘val’ is the valuation state
which doesn’t gets activated if a person has empathy (fsemp; psemp), which nar-
cissist lacks. So, in turn psact activates the execution state (esact), i.e. an angry
reply. This reaction involves a thought process about predicted effect (wseff;
sseff; srseff), and eventually anxiety is induced (wsanx; ssanx; srsanx; psanx; fsanx).
However, anxiety still elevates reaction (esact).Please note, as social media is
not controllable, therefore esact doesn’t influence wsnf. Black horizontal arrows
(layer I) show non-adaptive causal relations, while green show the adaptive
ones. Purple arrows shows suppression from one state to another.
Table 1. Categorical Explanation of States of Base Model (Layer I).
Categories
References
Stimulus states:
wsi
World state. i = stimulus (s);
positive / negative feedback (pf/nf)
ssi
Sensory state. i = stimulus; pf / nf
srsi
Representation state j = pf / nf
Stimulus is sensed and leads to
representation: p51 [16]
Attribution / evaluation states:
eval+
Positive evaluation of feedback
eval-
Negative evaluation of feedback
Narcissism involves states for
self-enhancement and mental-
izing. [7]
Happiness related states:
bs+
Self-belief state
striatum
Ventral Striatum : brain part
PFC
Prefrontal Cortex: brain part
fsreward
Feeling state of reward (Amygdala)
fslove
Feeling state self-love (Amygdala)
eshappy
Execution state of happiness
insula
Anterior Insula : brain part
fMRI studies show activations
at or near dopaminergic mid-
brain nuclei and the VS that
correlate with both reward
expectation and reward pre-
diction errors…[8]
Anger related action states:
os
Ownership state
psact
Preparation state of action
esact
Execution state of action
… predictive and inferential
processes contribute to con-
scious awareness … action”
p212 [16]
Body Loops (Anger and Anxiety):
wsi
World state i=anger/anxiety (ang / anx)
ssi
Sensor state i=anger/anxiety (ang / anx)
psi
Preparation state of i = ang / anx
fsi
Feeling state i = ang / anx
esang
Execution state (Expression of anger)
Body loop via the expressed
emotion is used to generate a
felt emotion by sensing the
own body state… the emer-
gence of states [16].
Predicted Effect of Action:
wseff
World state of effect
sseff
Sensor state of effect
srseff
Representation state of effect
Sensory feedback provides
more precise evidence about
actions and their effects. p212
[16]
Control states:
cs
Control state
val
Valuation state
ACC become active in parallel
with the insula when we expe-
rience feelings”. p109 [16]
Layer II, presents the plasticity of the model, while layer III represents meta-
plasticity. Each layer is connected by upward (blue) and downward (red) ar-

6
rows. Layer II incorporates Hebbian principle [17] by upward (blue) and
downwards (red) arrows with states at layer I [14]. For instance, for a narcis-
sist receiving appreciation, feeling of reward (fsreward) stimulates ventral stria-
tum (striatum ) [8], is represented by connection fsreward  striatum. Its in-
creases over the time, this increase is due to pre-synaptic (fsreward) and post-
synaptic (striatum) states depend on dopamine based activations (dopamine
release) and this is represented through W(fsrew,str ).
Layer III represents meta-plasticity, through states M and H, that can con-
trol the learning of state Wfsang,psa at layer-II. Former indicates persistence,
while later specifies the learning rate of fsang  psact. Usually, every W state at
layer II has meta-plasticity, because of presynaptic and post synaptic states
involvement in gaining experience [14]. However, to keep the complex net-
work simple, we only meta-plasticized the angry reaction of a narcissist, which
is due to changes in the synaptic connections [10, 11].
Table 2. Explanation of States in Layer II and III.
States per Layer
References
Layer II (Plasticity / Omega states):
1.Wfslove,bs
For fslovebs
2.Wbs,fslove
For bs  fslove
3.Wsat,ins
For striatum  insula
4.Wfsrew,str
For fsreward  striatum
5.Weval,psa
For eval-  psact
6.Wpsa,srseff
For psact  srseff
7.Wfsang,psa
For fsang  psact
1 – 4: Potentiation in the striatum depends
not only on strong pre- and postsynaptic
activation … reward prediction … modify
behavior.[8]
5 – 7: Presynaptic somatodendritic 5-HT1…
people with a high level of aggression, there
is a greater density … with impulse con-
trol.[12]
Layer III (Meta-Plasticity):
H
Speed factor for Wfsang,psa
M
Persistance factor for Wfsang,psa
Damage to neurons in hippocampal CA3
area and microstructure of synapse indi-
cates that anger… harms plasticity .... [10]
For computation of impacts most speed factors η and connection weights ω at
layer I, have values between 0 and 1. Please note that for Wfsang,psact the speed
factor is adaptive, i.e. based on the reification state H, therefore we used adap-
tive combination function for computation. Here t is 0.5 . We used three type
of combination functions for the simulation of our model (Figure 1):
a) For 24 states (wss; sspf; ssnf; sss; srspf; pfc; eval+; eshappy ; eval-; os; esact;
ssang; srsang; psang; esang; fsang; fsemp; wseff; sseff; wsanx; ssanx; srsanx; psanx; fsanx),
we used Euclidian function, with order n > 0 and scaling factor λ as the sum of
connection weights of a particular state:
eucln,λ (V1,…,Vk) = 
    
 


b) For 14 states (srsnf; bs; striatum ; fslove; fsreward;insula; cs; psact; wsang; val;
psemp; srseff; H; M), we used the alogistic function with positive values of
steepness σ and threshold  less than 1:
alogistic,(V1, …,Vk) = [(1/(1+e–σ(V1+ … + Vk -))) – 1/(1+eσ)] (1+e–σ)

7
where V is the single impact computed by product of state values and its con-
nection weight i.e ωX,Y X(t).
c) Lastly, for 7 adaptation states (W(bs,flove); W(flove,bs);W(striatum ,insula); W(fsrew,str );
W(psact,srspe); W(fsAng,psact); W(eval,psact)) we used Hebbian learning:
hebbµ(V1,V2,W) = V1V2(1-W) + µW
Mathematically, a reified-architecture based model is represented as [14]:
1. At every time point t, the activation level of state Y at time t is repre-
sented by Y(t), with the values between [0,1].
2. Impact of state X on state Y at time t is represented by impactX,Y(t) =
ωX,Y X(t); where ωX,Y is the weight of connection XY.
3. Special states are used to model network adaptation based on the no-
tion of reification network architecture. For example, WX,Y represents
an adaptive connection weight ωX,Y(t) for the connection XY, while
HY represents an adaptive speed factor ηY(t) of state Y. Similarly, Ci,Y
and Pi,j,Y represent adaptive combination functions cY(.., t) over time
and its parameters respectively. Combination functions are built as a
weighted average from a number of basic combination functions
bcfi(..), which take parameters Pi,j,Y and values Vi as arguments. Uni-
versal combination function c*Y(..) for any state Y is defined as:
c*Y(S,C1,…,Cm,P1,1,P2,1,…,P1,m,P2,m,V1,…,Vk,W1,…,Wk,W)=W+S[C1bcf1(P1,1,P2,1,W1V1,
…,WkVk) + … +Cmbcfm(P1,m,P2,m,W1V1,…,WkVk)] / (C1+…+Cm) – W]
where at time t:
 variable S is used for the speed factor reification HY(t)
 variable Ci for the combination function weight reification Ci,Y(t)
 variable Pi,j for the combination function parameter reification Pi,j,Y(t)
 variable Vi for the state value Xi(t) of base state Xi
 variable Wi for the connection weight reification WXi,Y(t)
 variable W for the state value Y(t) of base state Y.
4. Based on the above universal combination function, the effect on any
state Y after time Δt is computed by the following universal differ-
ence equation as:
Y(t+Δt) = Y(t) + [c*Y(HY(t), C1,Y(t), …, Cm,Y(t), P1,1(t), P2,1(t), …., P1,m(t), P2,m(t),
X1(t), …, Xk(t), WX1,Y(t), …, WXk,Y(t), Y(t)) - Y(t)] Δt
which also can be written as a universal differential equation:
dY(t)/dt = c*Y(HY(t), C1,Y(t), …, Cm,Y(t), P1,1(t), P2,1(t), …., P1,m(t), P2,m(t), X1(t), …,
Xk(t), WX1,Y(t), …, WXk,Y(t), Y(t)) - Y(t)
Our Simulation environment was implemented in MATLAB, and receives
input of the characteristics of a network structure represented by role matrices.
A role matrix is a compact specification with the concept of the role played by
each state with specified information. Detailed information for the designed
model can be found online [15, 18].

8
4 Simulation Scenarios
Simulation scenarios are used to verify the dynamic properties of the model by
simulating real-world processes. Simulations of the model are addressed be-
low by two kinds of comment: a) appreciation or b) a negative critic.
4.1 Reaction of a narcissist when appreciated
Twitter, Facebook [2] or Instagram [1] are a few popular platforms used by
narcissists to gain recognition and prominence. Many scholars have been
studying “messiah complex” of Donald Trump [19]. To illustrate his behavior
related to reward-seeking and self-love, we can take his tweet as an example
of self-love and self-reward:
“…my two greatest assets ... mental stability and being, like, really smart … I went from
VERY successful businessman, to top T.V Star….” (Tweeted: 1:27 PM – Jan 6,2018)
Fig. 3 shows that when positive feedback arrives, eval+ (purple) is activated,
which activates PFC and reward seeking process through striatum. These acti-
vations along with self-belief (bs+) make feelings of self-love (fslove) and re-
ward (fsreward) high. As a result, he expresses his gratitude (eshappy). Here, insu-
la indicates the self-thinking process, which increases self-love and feeling of
reward more and more.
Fig. 2. Simulation of the Model when wspf = 1 and wsnf = 0.
4.2 Reaction of a narcissist on a negative feedback
A narcissist, observing ego-threatening feedback, doesn’t hesitate to share his
reactions. While studying Donald Trump, we can identify his overtly reactions
easily on any of stances, which are evaluated as threat to his ego. For example,
consider the tweet of Donald Trump as:
“… world class loser, Tim O`Brien, who I haven’t seen or spoken … knows NOTHING
about me … wrote a failed hit piece book…” (Tweeted: 6:20 AM – Aug 8,2019) [20]
Fig. 4 explains the behavior of a narcissist over negative feedback (shaded
region) in episodes: i.e. from 100 – 200; and 300 - 400. Reward related states
in the white region (e.g. fslove, fsreward ,..) are suppressed, eval- (purple) is
raised due to srsnf and cs. It is responsible for two more activations. Firstly, it
activates body loop of anger in red (wsang, ssang, srsang, fsang, psang and esang),

9
which in turn activates the effect related states in green (wseff,sseff, and srseff).
Secondly, it stimulates urge to respond by psact, os and esact, however this leads
him to experience anxiety in blue (wsanx, ssanx, srsanx). Narcissists lacks empa-
thy (fsemp and psemp), therefore valuation (val) is not suppressed. As a result he
replies to such feedback. Over the time (X-axis), it can be seen that such be-
havior continues to aggravate (by Hebbian learning) by similar experiences.
Detailed explanation of each curve can be found online [18].
Fig. 4. Simulation of the Model with alternative episodes of wsnf = 1 or wspf = 1.
5 Model Validation and Analysis
The model is validated by three public Instagram users, with presumably nar-
cissistic characteristics (names are not disclosed here). Reasons for choosing
Instagram are: a) users show more tendency towards narcissism [1] and, b)
this platform contain conversational elements, which can be helpful for our
study.
5.1 Extraction and Analysis of Data
We extracted data from Jan 2017-July 10,2019 and tested for three hypoth-
esis: A) Frequency of sharing posts increases over a period of time. B) Aver-
age number of likes increase with the number of shared posts. C) They get
happy when admired, but covert/overt behavior can be seen towards a critic.
To come up with these hypotheses, data was analyzed in two steps. First,
we analyzed a number of posts shared with the average number of likes per
month. Fig. 4-a shows that each profile user 1/2/3 tends to share more by the
passage of time. Similarly, Fig. 4-b indicates that the number of average likes
a profile receives increases. For example, Profile 1 (indicated by blue) 2 posts
with avg. likes (817.5) (in Jan 2017), raised by 25 posts with avg. likes 7045.5
(July 2019). Please note, we didn’t consider the number of followers the user
had in mentioned duration, which makes average number of likes a bit subjec-
tive. However, the trend line of Fig. 4-a, indicates that Profile1 finds a good
reason to share updates with increasing frequency of 25 posts per month.

10
(a)
(b)
Fig. 4. Number of a) Posts shared, b) Average Likes received during a period of time
Our third hypothesis is related to the temporal analysis of data. Each post is
deeply analyzed (Fig. 5). We selected conversations in each post, where pro-
file user participated. We used Vader Sentiment Analysis [21] to analyze the
first message in the conversation as positive, neutral or negative ones. Looking
closely, it was revealed that various elements in negative conversation were an
expression of positive assurance, but were misclassified (e.g. comment “🔥🔥
Fierce as fuck” score = -0.54). At this point, we used rule-based classification
using further characteristics of data. Algorithm can be found online [18].
Fig. 5. Steps taken to interpret the behavior of each user
A stacked graph of three categories of conversation for each profile was plot-
ted (Fig. 6). In general, it can be seen that most of the feedbacks received were
positive or neutral, which provides a good reason for frequent posting on In-
stagram (Fig. 4). A mixed ratio of conversations can be seen. Like Profile 1,
the users reaction was never negative in 2017. Similarly, Profile 2 has more in
2018 or 2019. On the contrary, Profile 3, had more negative conversations in
2017, a reason can be that he doesn’t indulge into conversations anymore.
Fig. 6: Conversation ratio per profile from Jan 2017 – July 2019
However, it can be concluded that narcissistic people do not hesitate to react
over a negative comment in an overtly or covertly manner. Anxious behavior
is not studied, as authors don’t know them personally. However, it would not
be wrong to say that learning from experience facilitates users to go further for

11
sharing in their lifestyle, or responding to different followers. Our model also
addresses the experience and reactions after experiences of the users.
5.2 Exhibition of learning experience in the model
While looking into Fig.4, we can see the complex learning behavior between
all three layers over the time (in episodes). As addressed in Section 4, that an
urge to respond (action: psact;esact;os) increases on the basis of predicted effect
(effect: wseff,sseff and srseff). Similarly, reward related states (striatum,fsreward,
fslove,insula) are also elevated than before when a narcissist is admired. This
can also be seen by the adaptive states (Layer II and III) shown in Fig. 7. For
example, considering Weval,psa (purple), we can see that it start increasing its
value (e.g. from 0.2) in every negative episode (to 0.7). Similarly, M (brown)
and H (blue) increase in every negative episode and suppressed otherwise.
However, it can be seen that due to meta-plasticity, Wfsang,psa was not much
raised (shaded region), which indicates that due to synaptic plasticity learning
is effected [10], the similar pattern is observed during analysis of data, that
action is followed if effect is assumed to be higher, i.e. if the feedback is really
hurting the esteem of a narcissist then he will go towards an angry reaction
and a happy gesture is observed upon admiration, the reason can be he wants
to show that he is loved by many individuals.
Fig. 7: Effects of plasticity (W states) and Metaplasticity for Wfsang,psa (M and H)
6 Conclusion
In this paper, we discussed the complex adaptive mental network model of the
processes of a narcissist, who reacts in different ways after having a positive,
or negative feedback, and how prior experiences play role in learning and
responding through different layers of reified network architecture. We tested
our network model on Instagram data which is assumed to be a fertile ground
for people with a narcissistic personality. Through the analysis of temporal
data obtained from three users of Instagram, it was concluded that our model
indeed depicts the behavior of a narcissist and reflects his/her joy or rage upon
feedback.
In future research, we aim to extend our work, to incorporate how to react
to a narcissistic person and explore his traits further. Moreover, we aim to
detect and model how to support these complex behaviors among narcissists.

12
References
1. Moon, J.H., Lee, E., Lee, J.-A., Choi, T.R., Sung, Y.: The role of narcissism in self-promotion on Insta-
gram. Personality and Individual Differences. 101, 22–25 (2016).
2. Wang, D.: A study of the relationship between narcissism, extraversion, drive for entertainment, and
narcissistic behavior on social networking sites. Computers in Human Behavior. 66, 138–148 (2017).
3. Bushman, B.J., Baumeister, R.F.: Threatened Egotism, Narcissism, Self-Esteem, and Direct and Displaced
Aggression: Does Self-Love or Self-Hate Lead to Violence? Journal of personality and social psychology.
75.1, 219. (1998).
4. Fan, C., Chu, X., Zhang, M., Zhou, Z.: Are Narcissists More Likely to Be Involved in Cyberbullying?
Examining the Mediating Role of Self-Esteem. Journal of Interpersonal Violence. 34.15, 3127-3150 (2019).
5. Fan, Y., Wonneberger, C., Enzi, B., de Greck, M., Ulrich, C., Tempelmann, C., Bogerts, B., Doering, S.,
Northoff, G.: The narcissistic self and its psychological and neural correlates: an exploratory fMRI study.
Psychological Medicine. 41, 1641–1650 (2011).
6. Neuman, Y.: Computational personality analysis: Introduction, practical applications and novel directions.
Springer (2016).
7. Olsson, J., Berglund, S., Annett, J.: Narcissism – Brain and Behavior: Self-Views and Empathy in the
Narcissistic Brain term, University of Skovde (2014).
8. Daniel, R., Pollmann, S.: A universal role of the ventral striatum in reward-based learning: Evidence from
human studies. Neurobiology of Learning and Memory. 114, 90–100 (2014).
9. Weger, M., Sandi, C.: High anxiety trait: A vulnerable phenotype for stress-induced depression. Neurosci-
ence & Biobehavioral Reviews. 87, 27–37 (2018).
10. Sun, P., Wei, S., Wei, X., Wang, J., Zhang, Y., Qiao, M., Wu, J.: Anger Emotional Stress Influences
VEGF/VEGFR2 and Its Induced PI3K/AKT/mTOR Signaling Pathway. Neural Plasticity. 2016, 1–12
(2016).
11. Schmidt, M.V., Abraham, W.C., Maroun, M., Stork, O., Richter-Levin, G.: Stress-induced metaplasticity:
From synapses to behavior. Neuroscience. 250, 112–120 (2013).
12. de Almeida, R.M.M., Cabral, J.C.C., Narvaes, R.: Behavioural, hormonal and neurobiological mecha-
nisms of aggressive behaviour in human and nonhuman primates. Physiology & Behavior. 143, 121–135
(2015).
13. Jabeen, F.: How Happy You Are: A Computational Study of Social Impact on Self-esteem. In: Interna-
tional Conference on Social Informatics, Social Informatics, Springer Cham, 11186, 108-117 (2018).
14. Treur, J.: Multilevel Network Reification: Representing Higher Order Adaptivity in a Network. In:
Complex Networks and Their Applications VII. pp. 635–651. Springer International Publishing (2019).
15. Treur, J.: Modeling Higher-Order Adaptivity of a Network by Multilevel Network Reification. Network
Science journal. to appear (2019).
16. Treur, J.: Network-Oriented Modeling. Springer International Publishing, Cham (2016).
17. Hebb, D.: The Organization of Behavior. Wiley. (1949).
18. URL, “https://github.com/MsFakhra/Narcissism_TemporalSpecifications,” (2019).
19. Nai, A.: Disagreeable narcissists, extroverted psychopaths, and elections: a new dataset to measure the
personality of candidates worldwide. European Political Science. 18, 309–334 (2019).
20. Folley, A.: Trump rips MSNBC, CNN for inviting guests who “have no idea what I am all about,” (2019).
21. Hutto, C.J., Gilbert, E.: Vader: A parsimonious rule-based model for sentiment analysis of social media
text. AAAI conference on weblogs and social media. ICWSM (2014).