A computational model for flexibility in emotion regulation

Abstract

Emotion regulation is a vital psychological process that allows people to manage their own emotional states. Recent psychological research has highlighted the importance of flexibility in emotion regulation, such that people can alternative between different emotion regulation strategies a strategy is chosen depending upon the demands of the situation. This means that healthy emotion regulation is context-sensitive. This paper presents a computational model which models this form of flexible adaptation in emotion regulation in a simplified scenario in which the person has to switch between expressive suppression and attention modulation in managing anger in different work situations, . Simulation results are reported that illustrate the capacity of the model to display adaptivity in emotion regulation across different contexts.

Full text

A Computational Model for Flexibility
in Emotion Regulation
Nimat Ullah1, Jan Treur1 and Sander L. Koole2
1Behavioural Informatics Group, Department of Computer Science,
Vrije Universiteit Amsterdam, The Netherlands.
2Amsterdam Emotion Regulation Lab, Department of Clinical Psychology,
Vrije Universiteit Amsterdam, The Netherlands.
nimat.ullah@vu.nl, j.treur@vu.nl, s.l.koole@vu.nl
Abstract
Emotion regulation is a vital psychological process that allows people to manage their own emotional
states. Recent psychological research has highlighted the importance of flexibility in emotion regulation,
such that people can alternative between different emotion regulation strategies. A strategy is chosen
depending upon the demands of the situation. This means that healthy emotion regulation is context-
sensitive. This paper presents a computational model which models this form of flexible adaptation in
emotion regulation in a simplified scenario in which the person has to switch between expressive
suppression and attention modulation in managing anger in different work situations. Simulation results
are reported that illustrate the capacity of the model to display adaptivity in emotion regulation across
different contexts.
Keywords: Emotion regulation, Strategies, Context, Flexibility, Adaptivity, Psychopathology
1 Introduction
Emotion regulation refers to the process whereby people control their emotions to achieve some
long or short-term desirable goals. Because emotions are pervasive in everyday life, emotion regulation
is a vital capacity for psychological health and wellbeing (Koole, 2009). The dysregulation of emotion
disrupts people’s thoughts, actions, and goal achievements, and impairs social functioning (Zeman et
al., 2006). Indeed, emotional dysregulation is associated with a wide array of psychological problems,
including substance abuse, eating pathology, depression and anxiety (Aldao et al., 2010; Gross, 2012).
It is well-documented that people use many different strategies in emotion regulation (Gross, 1998;
Koole, 2009; Parkinson & Totterdell, 1999). Some examples are distraction (diverting attention away
from the negative emotion triggering entity), suppression (preventing the expression of emotion),
venting (freely expressing emotions), and cognitive reappraisal (reappraising belief about the emotion

triggering stimuli). Each of these strategies may be useful under a specific set of conditions, Thus, an
important question becomes how people are able to flexibly adapt their use of emotion regulation
strategies to shifting situational demands (see also Aldao, Gross, & Sheppes, 2015).
This paper addresses this question and proposes a computational model of flexibility in emotion
regulation. The rest of the paper is organized as follows. Section 2 of the paper discusses background
knowledge and the example strategies used later in simulations, Section 3 presents the computational
model, Section 4 elaborates the scenarios and the simulation results and finally Section 5 concludes the
paper.
2 Background
Different studies in the field have shown that the types of emotion regulation strategies chosen have
different types of consequences on the personality of an individual (Gross, 1998, 2001, 2014; Koole,
2009; Tamir, 2009, 2011; Webb, Miles, & Sheeran, 2012). Effects of these strategies can show up
immediate as well as over a longer time span. Based on the varying results of the studies conducted in
(Richards & Gross, 2000; Butler et al., 2003; Gross & Thompson, 2007; Sheppes & Meiran, 2007)
wherein various emotion regulation strategies have been compared for the possible outcomes in different
situations, Suri et al. (2018) conclude that “these differences suggest that decisions about which emotion
regulation strategies to use in a particular situation can be profoundly consequential, and it is therefore
important to understand the drivers of these decisions in different emotional contexts”.
Similarly, (Aldao et al., 2014) qualifies rigid emotion regulation responses as maladaptive due to
the varied environmental situations, thereby suggesting contextual and adaptive emotion regulation
instead. Koole and Veenstra, (2015) also acknowledge (Aldao et al., 2014) by “criticizing traditional
models of emotion regulation as being too much based on decontextualized mental representations. The
latter authors propose a new theoretical approach in which emotion regulation is continually and flexibly
adjusted to the contingencies of the situation”.
Various studies support the notion that good mental health tends to go hand in hand with emotion
regulation flexibility (Kashdan and Rottenberg 2010; Hollenstein et al. 2013; Bonanno and Burton
2014). This literature has emphasized the importance of emotion regulation flexibility as it is considered
crucial to be properly understood for better treatment of a variety of mental disorders (Aldao, 2013;
Bonanno and Burton, 2014). To gain more depth in the analysis of this phenomenon by applying
computational techniques, this paper presents a computational model for context-dependent adaptive
emotion regulation. It will be shown how it can adaptively switch from one strategy to another and vice
versa as the context changes over time. It will be illustrated for two important and widely studied
regulation strategies: expressive suppression and attention deployment.
As a subcategory of response modulation, expressive suppression refers to the act of suppressing
emotional expression, specifically, facial expression experienced as a result of an emotion eliciting
stimulus. Initial work suggested that expressive suppression is generally an ineffective emotion
regulation strategy (Gross, 2001) with negative social consequences (Butler et al., 2003). However,
subsequent work indicated that the expressive suppression may have fewer negative consequences in
non-Western cultures (Butler, Lee, & Gross, 2007). Moreover, some authors have suggested that
expressive suppression may also be adaptive in some specific contexts (Tamir, 2009).
Attention deployment (Gross, 1998) can further be classified into distraction (Sheppes and Gross, 2011),
rumination, worry and thought suppression (Campbell-Sills & Barlow, 2007). The most widely studied
form of attention deployment is distraction, in which attention is diverted away from the emotion-
triggering stimulus, for example, by performing a cognitive task (Van Dillen & Koole, 2007). However,
there are also more subtle forms of attention deployment, for instance, when people focusing on
remembering what is being said rather than who said it.

3 The Computational Model
This section discusses the proposed computational model. As discussed earlier, context and
adaptivity play an important role in emotion regulation and through this model anger is adaptively
regulated in different contexts. First the Network-Oriented Modeling approach used from (Treur, 2016)
is briefly explained. For this modelling perspective, networks are used in which the nodes are interpreted
as states (or state variables) that vary over time, and the connections are interpreted as causal relations
that define how each state can affect other states over time. This type of network has been called a
temporal-causal network. A conceptual representation of a temporal-causal network model as a labelled
graph provides a fundamental basis, and includes the following labels for such a graph:
• a notion of strength of a connection is used as a label for connections, to distinguish between
different types of connections (connection weight X,Y)
• some way to aggregate multiple causal impacts on a state is used, to distinguish between the
differences in how different causal impacts work together (combination function cY(..))
• a notion of speed of change of a state is used for timing of the processes (speed factor Y)
These three notions connection weight X,Y, combination function cY(..), and speed factor Y, define the
network structure of a temporal-causal network model by a conceptual representation; see the upper part
(first 5 rows) of Table 1.
concept
conceptual representation
Explanation
States and
connections
X, Y, X→Y
Describes the nodes and links of a network structure
(e.g., in graphical or matrix format)
Connection weight
X,Y
The connection weight X,Y  [-1, 1] represents the
strength of the causal impact of state X on state Y
through connection X→Y
Aggregating
multiple impacts on
a state
cY(..)
For each state Y (a reference to) a combination function
cY(..) is chosen to combine the causal impacts of other
states on state Y
Timing of the effect
of causal impact
Y
For each state Y a speed factor Y  0 is used to
represent how fast a state is changing upon causal
impact
concept
numerical representation
explanation
State values over
time t
Y(t)
At each time point t each state Y in the model has a real
number value in [0, 1]
Single causal impact
impactX,Y(t)
= X,Y X(t)
At t state X with a connection to state Y has an impact
on Y, using connection weight X,Y
Aggregating
multiple causal
impacts
aggimpactY(t)
= cY(impactX1,Y(t),…, impactXk,Y(t))
= cY(X1,YX1(t), …, Xk,YXk(t))
The aggregated causal impact of multiple states Xi on Y
at t, is determind using combination function cY(..)
Timing of the causal
effect
Y(t+t) = Y(t) + Y [aggimpactY(t) - Y(t)] t
= Y(t) + Y [cY(X1,YX1(t), …, Xk,YXk(t)) - Y(t)] t
The causal impact on Y is exerted over time gradually,
using speed factor Y; here the Xi are all states with
outgoing connections to state Y
Table 1. Conceptual and numerical representations of a temporal-causal network model
There are many different approaches possible to address the issue of combining multiple impacts. To
provide sufficient flexibility, for each state its combination function can be chosen that specifies how
multiple causal impacts on this state are aggregated. For this, a library with a number of standard
combination functions are available as options, but also own-defined functions can be added.
In the lower part of Table 1, the numerical representation describing the dynamics of a temporal-
causal network is displayed, and it is shown how it is defined based on the characteristics (labels) of the
conceptual network structure specification. In the lower part of Table 1, this interpretation is expressed
in a formal-numerical way, thus associating semantics to any temporal-causal network specification in

a detailed numerical-mathematically defined manner; see also (Treur, 2016b), Ch. 2. The obtained
difference equations in the last row in Table 2 constituting the numerical representation of the network
dynamics are just defined in terms of the network structure parameters X,Y, cY(..), and Y.
For the specific temporal-causal network model introduced here the conceptual representation is
given in Fig. 1. In Table 2 the different states are explained.
Figure 1. Conceptual representation of the computational model as a temporal-causal network
Formal
Informal Name
Description
wss
World state for stimulus s
The situation in the real world that
triggers anger
srss
Sensory representation state for
stimulus s
Internal representation of the anger
triggering situation
srsbp
Sensory representation state for bp
Internal representation of the context in
the real world
fsb
Feeling state for body state b
Feeling associated to body state b
psb
Preparation for body state b
Preparation state for body state b
psad
Preparation state for attention
deployment
Preparation for the Attention
deployment action
cssup
Control state for suppression
Control (and monitoring) state for anger
Suppression of Expression
csad
Control state for attention deployment
Control (and monitoring) state for
Attention Deployment
esb
Execution state for body state b
Execution state for body state b
esad
Execution state for attention
deployment
Execution state for the Attentional
Deployment action
Table 2. Overview of the states of the proposed model (in connection to Fig. 1)
The computational model presented above, regulates emotion over time. The positive and
negative links of srsbp shows the presence and absence of a specific context factor (in the example
simulations the presence of the boss) which has some effect on the way emotions are regulated. In case
of presence of this context factor, expressive suppression is employed while in case of absence of it

attentional deployment is chosen for downregulating negative emotions. In both contexts, the primary
stimulus srss triggering anger is present. Moreover, when the context changes over time, the emotion
regulation strategies also change, as emphasized by various scholars; this model shows indeed a change
in the choice of strategy over time; this will be demonstrated in the next section through graphs of
simulation results. In the model in Fig. 1 it can been seen the control state cssup for suppression reduces
fsb, psb and esb by the negative connections to them. On the other hand, in attentional deployment, the
attention is moved away from the emotional eliciting situation. It can be seen in Fig. 1 that csad works
via psad and esad on srss to reduce attention for the stimulus.
4 Scenarios and Simulation Results
Using the computational model introduced above, the following scenario has been simulated, which
is loosely based on an example discussed by Aldoa et al. (2015):
“An employee feels quite angry every time when a particularly obnoxious coworker starts
speaking. This anger might be accompanied by an intense urge to roll his eyes.
Fortunately, the employee can prevent his anger by deploying his attention and focusing
on remembering what was said in the meeting rather than who said it. However, when the
employee’s manager is present, the employee becomes concerned that she will still pick
up on some of his hostile feelings toward the coworker. Therefore, when the manager is
present, he attempts to hide his facial expressions of anger, and therefore switches from
attention deployment to expressive suppression”.
To simplify the scene, let’s say person A gets angry as person B talks. Four different contexts
are possible with this situation. The first is the boss is present and B talks. Person A would go for
suppression of his expression. If the boss is not present and B talks, A would go for attention deployment
and focus on remembering to what B says. It is also possible that the boss is initially present and B talks.
A goes for suppression but in the meanwhile boss leaves the room. So, A switches to attention
deployment from expressive suppression and vice versa. In this way this model adaptively models
emotion dependent on the context which changes over time.
In the simulations presented here for all states for the combination function the advanced logistic sum
combination function alogisticσ,τ (…) is used (Treur, 2016); here σ is a steepness parameter and τ a
threshold parameter.
cY(V1, …Vk) = alogistic,(V1, …, Vk) = ( 
 - 
 ) (   )
The parameter values used are shown in Table 3 and 4.
State



State



wss
0
0
0
fsb
0.1
5
0.2
srsbp
0
0
0
psad
0.4
4
0.2
srss
0.4
5
0.2
psb
0.4
4
0.2
csad
1
15
0.1
esad
0.5
4
0.2
cssup
2
15
0.02
esb
0.5
4
0.2
Table 3. Values of threshold and steepness parameters for the logistic sum combination functions and the
speed factors used

Connection
Weight
Connection
Weight
wss, srss
1
fsb, cssup
0.1
srsbp, csad
-1
fsb, psb
0.4
srsbp, cssup
1
psad, csad
0.8
srss, csad
1
psad, esad
1
srss, cssup
1
psb, csad
0.8
srss, psb
1
psb, cssup
0.8
csad, psad
1
psb, fsb
0.8
cssup, fsb
-0.2
psb, esb
1
cssup, psb
-0.7
esad, srss
-1
cssup, esb
-0.5
esb, cssup
0.8
fsb, csad
0.1
Table 4. Values of connection weights used
Figure 2. Simulation results when the boss is not present (Attentional Deployment is applied)
Figure 3. Simulation results when the boss is initially not present but joins later on (switch from Attentional
Deployment to Expressive Suppression)

Figure 4. Simulation results when the boss is present (Suppression of Expression is applied)
Figure 5. Simulation results when the boss is present initially but leaves later on (switch from Expressive
Suppression to Attentional Deployment)
Fig. 2 illustrates attentional deployment; the boss has not joined the meeting yet and B talks.
So srss goes up, triggering anger of A, but it starts decreasing as csad starts going up, i.e., the attention
deployment strategy gets activated and hence the anger comes down. A case in which the boss still joins
the meeting at a later point in time is depicted in Fig. 3. A has already lost his temper due to B and is
busy with attention deployment when the boss joins. After the boss has joined A switches from attention
deployment to expressive suppression. This transition in strategy occurs with the change in context,
which is referred to as adaptivity in terms of strategies: expressive suppression gets activated and
attentional deployment goes down as per demand of the context.
Fig. 4, represents expressive suppression, where A gets angry as B talks but as the boss is
present right from the start, A suppresses the emotion. It can be seen that the stimulus representation
srss still remains high but suppression gets higher with time and doesn’t let A express the anger.
In another scenario, as soon as A is about to get some control over his expression, the boss gets a call
and leaves the meeting. Fig. 5 demonstrates this context dynamics where initially the boss is present and
A gets teased as B talks. A suppresses the emotion but in the meantime, the boss leaves the room. It can
be observed that when the boss was present, suppression is active but as soon as the boss leaves, attention
deployment gets activated and the stimulus representation srss also falls down, due the attention
deployment. Note that it can also be seen in the graphs srss remains up in case of expressive suppression,
depicted in the respective graphs.

5 Conclusion
The computational model introduced here was inspired by literature regarding the adaptive
contextual choice of emotion regulation strategies. As elaborated by various researchers in the area,
emotion regulation strategies cannot be termed adaptive or maladaptive without taking, into account, a
specific context. These strategies can only be termed adaptive or maladaptive pertaining to a specific
context. The model described in this paper elaborates this computationally and highlights the adaptivity
of the strategies in different contexts. The results of the simulation experiments for the different
scenarios are in accordance with what is described in this literature.
In future work, as an extension, a more complex model may be designed for simulating the
adaptivity of more emotion regulation strategies in various contexts relative to each other. In the current
model each context triggers a strategy in a reactive manner. Another extension of this work may be
obtained by incorporating a more explicit decision-making process for the choice of strategy, for
example, by including elements from (Manzoor, Abro, and Treur, 2017).
References
Aldao, A. (2013). The Future of Emotion Regulation Research: Capturing Context. Perspectives on
Psychological Science, 8(2), 155–172. https://doi.org/10.1177/1745691612459518
Aldao, A., Jazaieri, H., Goldin, P., & Gross, J.J. (2014). Adaptive and maladaptive emotion regulation
strategies: Interactive effects during CBT for social anxiety disorder. Journal of Anxiety Disorders,
28, 382–389.
Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotion-regulation stratgies across
psychopathology: a meta-analytic review. Clinical Psychology Review 30, 217e237.
Aldao, A., Sheppes, G., & Gross, J. J. (2015). Emotion Regulation Flexibility. Cognitive Therapy and
Research, 39(3), 263–278. https://doi.org/10.1007/s10608-014-9662-4
Bonanno, G.A., & Burton, C.L. (2014). Regulatory flexibility: An individual differences perspective on
coping and emotion regulation. Perspectives on Psychological Science, 8, 591–612.
Butler, E.A.; Egloff, B.; Wlhelm, F.H.; Smith, N.C.; Erickson, E.A.; Gross, J.J. (2003). The social
consequences of expressive suppression. Emotion. 3 (1): 48–67. doi:10.1037/1528-3542.3.1.48.
Butler, Emily A.,Lee, Tiane L.,Gross, James J. (2007). Emotion regulation and culture: Are the social
consequences of emotion suppression culture-specific? Emotion, Vol 7(1), Feb 2007, 30-48.
PMID: 17352561 DOI:10.1037/1528-3542.7.1.30
Campbell-Sills, L. & Barlow, D. H. (2007). Incorporating emotion regulation into conceptualizations
and treatments of anxiety and mood disorders. In J. J. Gross (Ed.), Handbook of Emotion
Regulation (pp. 542-559). New York: Guilford Press.
Dan-Glauser, E.S.; Gross, J. J. (2011). The temporal dynamics of two response-focused forms of
emotion regulation: Experiential, expressive, and autonomic consequences.
Psychophysiology.48:1309–1322. doi:10.1111/j.1469-8986.2011.01191.x. PMC 3136552.
Gross, J.J. (1998). The emerging field of emotion regulation: An integrative review. Review of General
Psychology. 2: 271–299. doi:10.1037/1089-2680.2.3.271.
Gross, J.J. (2001). Emotion regulation in adulthood: Timing is everything. Current Directions in
Psychological Science, 10, 214–219.

Gross, J.J., & Thompson, R. A. (2007). Emotion regulation: Conceptual foundations. In J. J. Gross (Ed.),
Handbook of emotion regulation (pp. 3–24). New York, NY: Guilford Press.
Gross, J. J. (Ed.). (2012). Handbook of emotion regulation (2nd edition). New York: Guilford.
Gross, J.J. (2014). Emotion regulation: Conceptual and empirical foundations. In J. J. Gross (Ed.),
Handbook of emotion regulation (2nd ed., pp. 3–20). New York, NY: Guilford.
Hollenstein, T., Lichtwarck-Aschoff, A., & Potworowski, G. (2013). A model of socioemotonal
flexibility at three time scales. Emotion Review, 5, 397–405.
Kashdan, T.B., & Rottenberg, J. (2010). Psychological flexibility as a fundamental aspect of health.
Clinical Psychology Review, 30, 865–878.
Koole, S.L. (2009). The psychology of emotion regulation: An integrative review. (23 ed., Vol. 1, pp.
4-41). Psychology Press. Retrieved
from http://www.tandfonline.com/doi/abs/10.1080/02699930802619031
Koole, S.L., & Veenstra, L. (2015). Does Emotion Regulation Occur Only Inside People’s Heads?
Toward a Situated Cognition Analysis of Emotion-Regulatory Dynamics. Psychological Inquiry,
26(1), 61–68. https://doi.org/10.1080/1047840X.2015.964657
Larsen, R.J., & Prizmic, Z. (2004). Affect regulation. In R. F. Baumeister & K. Vohs (Eds.), Handbook
of self-regulation research (pp. 40–60). New York: Guilford Press.
Manzoor, A., Abro, A.H., Treur, J. (2017). Monitoring the Impact of Negative Events and Deciding
About Emotion Regulation Strategies. In: Proc. of the 14th European Conference on Multi-Agent
Systems, EUMAS'16. Lecture Notes in AI, vol. 10207, pp 350-363. Springer Publishers.
Parkinson, B., & Totterdell, P. (1999). Classifying Affect-regulation Strategies. Cognition and Emotion,
13(3), 227–232.
Richards, J.M., & Gross, J. J. (2000). Emotion regulation and memory: The cognitive costs of keeping
one’s cool. Journal of Personality and Social Psychology, 79, 410–424.
Sheppes, G., & Meiran, N. (2007). Better late than never? On the dynamics of online regulation of
sadness using distraction and cognitive reappraisal. Personality and Social Psychology Bulletin,
33, 1518–1532.
Sheppes, G.; Gross, J. J. (2011). Is timing everything? Temporal considerations in emotion
regulation. Personality and Social Psychology Review. 15(4): 319–
331. doi:10.1177/1088868310395778.
Suri, G. Gal Sheppes, Gerald Young, Damon Abraham, Kateri McRae & James J. Gross (2018).
Emotion regulation choice: the role of environmental affordances, Cognition and Emotion. 32:5,
963-971, DOI: 10.1080/02699931.2017.1371003.
Tamir, M. (2009). "What do people want to feel and why? Pleasure and utility in emotion
regulation". Current Directions in Psychological Science. 18 (2): 101–105. doi:10.1111/j.1467-
8721.2009.01617.x
Tamir, M. (2011). The maturing field of emotion regulation. Emotion Review, 3(1), 3–7.
https://doi.org/10.1177/1754073910388685
Treur, J. (2016). Network-Oriented Modeling: Addressing Complexity of Cognitive, Affective and
Social Interactions. Springer Publishers.

Van Dillen, L. F., & Koole, S. L. (2007). Clearing the mind: A working memory model of distraction
from negative emotion. Emotion, 7, 715723
Webb, T.L., Miles, E., & Sheeran, P. (2012). Dealing with feeling: A meta-analysis of the effectiveness
of strategies derived from the process model of emotion regulation. Psychological Bulletin, 138(4),
775–808.
Zeman, J.; Cassano, M.; Perry-Parrish, C.; Stegall, S. (2006). Emotion regulation in children and
adolescents. Journal of Developmental and Behavioral Pediatrics. 27: 155–
168. doi:10.1097/00004703-200604000-00014.