A formal model of procrastination

Ruurdje Procee a Bart Kamphorst a Arlette van Wissen b

John-Jules Meyer c

a
Utrecht University, Janskerkhof 13A, 3512 BL Utrecht
b
VU University Amsterdam, De Boelelaan 1081a, 1081 HV Amsterdam
c
Utrecht University, Princetonplein 5, De Uithof, 3584 CC Utrecht
Abstract
Procrastination is widespread self-undermining behaviour that negatively impacts individual performance
and well-being. Psychological research has identified a variety of factors that influence this complex
behaviour. However, how these factors interact and influence procrastination is poorly understood; there
is no uniform theory that integrates all these factors. This paper presents an initial conceptual model of
procrastination built from factors that are found in the psychological literature. In addition, it presents a
formalization of the model that will serve as the basis for a computational model. The aim of the project
is to develop model-based e-coaching software that can offer personalized support to individuals in their
struggle with procrastination.

1 Introduction
Procrastination is voluntarily delaying what one intended to do despite expecting negative consequences.
This self-undermining behaviour is widely spread among students [13], but also chronically affects 15% –
20% of the general population [8]. Research shows that procrastination has a negative influence on an indi-
vidual’s performance [17] and well-being [18]. The majority of the procrastinators (95%) see this behaviour
as problematic and want to reduce it [17].
There are many definitions of procrastination but most include the existence of an irrational and unnec-
essary delay and negative consequences. The definition by Klingsieck combines all these features, stating
that procrastination is “the voluntary delay of an intended and necessary and/or [personally] important
activity, despite expecting potential negative consequences that outweigh the positive consequences of the
delay” [9, p. 26]. Notice that not performing an intended action is not necessarily procrastination. Someone
may justifiably shift his or her priorities if something more important comes along.
Procrastination is complex behaviour influenced by many factors such as low conscientiousness [17],
task aversiveness [4] and self-efficacy [17]. However, how these factors interact and influence procrastina-
tion is poorly understood. There is no uniform theory that integrates all these factors. This paper presents an
initial conceptual model of procrastination built from factors that were found in the psychological literature.
Using this conceptual model, a formalization is made that captures the different factors and defines an agent
that can choose to perform activities that are either task-related or not. The goal of the formal model is
to gain a better understanding of the factors influencing procrastination and the connections between them.
In future work, the formal model will be used to make a computational model to run simulations, train
the model on empirical data and test it. It will simulate what decisions an agent will make when it has to
choose between different tasks and indicate whether the selected activity classifies as procrastination or not.
The aim is that in future work the trained model will serve as the basis for e-coaching software to support
individuals in decreasing procrastination.
The paper is organized as follows. Section 2 discusses other approaches to modeling procrastination.
Section 3 presents a conceptual model of procrastination that is built from factors prominently featured in
psychological literature. A formalization of the model is presented in Section 4. Section 5 outlines an
evaluation plan and, lastly, Section 6 discusses conclusions and ideas for future work.
2 Related Work
There are only a few other models on procrastination and the majority of them are economic models. The
idea of the economic models on procrastination (e.g. [11], [19]) is that an agent can choose between some
tasks. Each task is associated with costs that are incurred when the task is executed, and benefits that are
received when the task is completed. Traditional economic models assume time-consistency: an agent’s
preference for a task remains the same over time. Most people are however time-inconsistent: they prefer
well-being at an earlier date over a later date [6]. Economic models on procrastination differ from traditional
economic models in that time-inconsistency is taken into account.
In such models agents can have a self-control problem and are either fully aware of this (‘sophisticated’)
or unaware (‘naive’). A sophisticated agent will estimate the utility of a task the same for every day, so it
will perform the task immediately. A naive agent believes that it is better off to perform the task later instead
of now, so it will do the task the next day. However, when this continues, the agent may end up spending
too little time on a task. The main determinants of utility are time preference, costs, and benefits.
These economic models give some insight into the dynamics of procrastination. However, it is often
not clear how these models relate to relevant psychological concepts. To get a better understanding of
procrastination with the use of a model, it is important to specify how the model components are mapped to
concepts in psychological literature.
An example of a model that also integrates psychological concepts is the temporal motivation theory
(TMT). It is based on the economic models, but it also incorporates expectancy and need theory, and aims
to integrate different theories on motivation. The model is applied to procrastination, using the following
formula [17]:
E×V
U tility =
ΓD
with expectancy E, value V (the costs and benefits received from performing a task), sensitivity to delay Γ
(representing to what extent delay influences utility) and the time to the deadline D. Each element consists
of different task-related and personality-related factors. This theory integrates a number of psychological
notions connected to procrastination. However, it is not specified how the theory should be implemented;
only some small, non-formal examples are given. Implementation into a formal model can validate the
theory and provide new insights into the underlying mechanisms of procrastination instead of only listing
the factors that influence it.

3 Conceptual model
In this section, prominent factors influencing procrastination found in the literature will be discussed. The
factors can be divided into different groups: task-related, personality-related and other factors like ego
depletion, mood, temptations and coping strategies. All these factors and their influence on procrastination
can be found in Figure 1. The + and − of the arrows represent respectively the positive and negative
influence from that factor on procrastination or another factor. This conceptual model was reviewed at an
early stage by a self-regulation expert.
The nature of the task and how the agent perceives it influence how likely someone is to procrastinate:
Boredom When a person finds a task really boring, it is hard to sustain it as an intentional activity when
there are less boring alternatives present [3]. The more boring a task, the more procrastination [4].
Frustration Being frustrated makes it harder to focus on a task. This lack of focus makes someone more
likely to procrastinate by engaging in other activities [4].
Personal Meaning The extent to which individuals feel their projects are worthwhile pursuits [10]. This
includes factors such as fun, pleasure, passion and other’s benefit. The lower this is, the less desirable
it is to do a task, which leads to more procrastination [4].
Autonomous Motivation A combination of intrinsic motivation, where there is an inner drive to reach
a certain goal, and extrinsic motivation, where people have identified with an activity’s value [5].
Autonomous motivation for a task leads to less procrastination [20].
Task Delay Events that are further away in time have less impact on people’s decisions. So when the
deadline or the consequences are far away, there is more procrastination [14].
Self-efficacy A person’s belief in his/her own competence. When someone’s self-efficacy for a certain task
is high, one is more likely to put effort in the task and set appropriate goals for oneself [1]. Self-
efficacy varies across different domains and tasks and is thus grouped with the task-related causes.
 When self-efficacy is high and someone believes that a certain goal is attainable, he/she is less likely
to procrastinate [17].
Task Structure When a task is less structured and more difficult to coordinate, there is more procrastina-
tion. The reason for this is that when the task is less structured, there are many points in time at which
a decision has to be made about what to do next, and this gives more opportunities for procrastination
[4].
Personal meaning Autonomous motivation
−
-
+
Boredom Task aversiveness + Frustration Mood

Task delay
−
+
−
Task Structure − −
+
−

Self-efficacy −
+/− Procrastination +
Ego depletion
+/− + +
+

Fear of failure Temptations

Sensation Seeking +
Legend:
Task related +

Personality related
Other factors −
Self esteem Trait procrastination Impulsiveness
−→ Influence on another factor +

or procrastination (positive (+) −

or negative (−)). Weights may be
influenced by coping strategies. Conscientiousness

Figure 1: Overview of causes and relations of procrastination

The personality of the agent is also important:

Impulsiveness Impulsive people are more likely to act on the desires of the moment. People that score high
on impulsiveness are more likely to procrastinate [17].
Sensation Seeking Sensation seeking is the tendency to find exciting activities, take risks and avoid bore-
dom. The higher someone scores on this trait, the more likely he/she is to procrastinate [17].
Conscientiousness Someone who scores high on this factor can be described in terms such as orderly, neat
and organized. Highly conscientious people are less likely to procrastinate [15].
Self-esteem How someone evaluates oneself. When someone has low self-esteem, one is more likely to
procrastinate [17].
Fear of Failure The fear of the negative consequences of not reaching one’s goals. How it affects procras-
tination is dependent on self-efficacy: individuals with fear of failure and high perceived competence
believe they can avoid failure by working very hard, thus reducing procrastination. When the per-
ceived competence is low, individuals feel that the probability of failing is very high and they will
avoid this by procrastinating [7].
Other factors that are taken into account are listed below. These are either external or internal variables that
change.
Mood The influence of mood on procrastination is not clear. Some research (e.g. [18]) suggests that a bad
mood can increase procrastination and that procrastination will have a negative influence on mood in
return, which will increase procrastination again. Yet others say there is no relation between mood
and procrastination [17]. We include it to be able to investigate the influence.
Ego Depletion The strength needed for self-regulation is a limited resource and when this resource is de-
pleted, it is much harder to control oneself [2]. The lower the ego resources, the more procrastination.
There is also a link between mood and ego-depletion. The reason for this is that when mood is low, it
makes a person more depleted, and a more depleted person has more difficulties maintaining a positive
mood.
Temptations Other activities that can be done instead of the current activity, but don’t contribute towards
reaching any of agent’s goals. The more temptations are present, the more likely that someone will
procrastinate [17].
Coping Strategies Strategies that can be applied to reduce procrastination [16]. Good coping strategies
help to reduce the effects that certain factors have on procrastination. For example, a coping strategy
like ‘gamifying a task’ can decrease procrastination by decreasing task aversiveness, while another
strategy may affect the influence of a particular temptation. Coping strategies are not directly included
in the figure. Instead, they are considered co-determinants for the weights of the connections.

4 Formalisation of the model

In this section a formalisation of the conceptual model is presented. In this model an individual agent has
various goals. These goals are the personal projects [10] of the agent and can be virtually anything. A
goal can be implemented by tasks. These tasks can have subtasks or a single activity. The tasks are on
a list of intended tasks and the place on this list is determined by the priority of the task. The agent can
perform activities belonging to tasks (one at the time) to reach given goals or perform other activities that
do not belong to a task. The activities that don’t belong to a task are generated by events. Figure 2 gives
an illustration of the model. Note that the events e1 and e2 are triggers that add a5 and a6 to the agent’s
options. When, for example, one of the activities in the block ‘Procrastination’ is performed, the agent is
procrastinating. Structured procrastination occurs when an agent is doing something useful that is on the list
of intended tasks, but not the most important thing to be done [12]. (A typical example of this is the increase
in cleanliness of student housing by the end of the semester because students don’t feel like studying but do
want to do something useful.) This is the case when a2 , a3 or a4 are performed.
In the following sections, first all the definitions will be discussed, followed by an explanation of how
the model works. All concepts and their formal notations can be found in Table 1.

g1 g2 g3 List of intended tasks:

1: τ1
2: τ2
τ1 τ2 τ5 e1 e2 3: τ3
4: τ4
5: τ5
τ3 τ4

a1 a2 a3 a4 a5 a6

Structured procrastination Procrastination

Options of the agent

Figure 2: An illustration of the formal model

4.1 Definitions
The agent p, goal g, task τ and activity a are defined as tuples:

p =<Mood, EgoResource, SensationSeeking, SelfEsteem, Conscientiousness, Impulsiveness, I, Gp , Cur-

rentActivity, O >

where I is an ordered set containing all intended tasks ordered by their priority, Gp the set of personal
goals of the agent, CurrentActivity the activity the agent is performing and O the set of all possible activities
an agent can choose from (i.e. the agent’s options): O = AE ∪ AT (where AE is the set of all activities
that belong to any event and AT the set of all activities that belong to any task). The other elements are the
personality traits found in the literature (see Section 3).

g =<Description, Tg , Importance, EstimatedTime, Deadline>

where Tg is the set of tasks belonging to a goal. This set can be empty when the agent hasn’t imple-
mented the goal yet with some tasks. It is assumed that each goal has a known importance value for the
agent. EstimatedTime and Deadline can be 0 when the agent doesn’t know these values.
 Concept Formal notation Definition
Agent p An individual agent in the model.
Goals gp ∈ Gp The personal projects of the agent. Given as an input in the model. Gp
is the set of goals of an agent p.
Task τ ∈ Tg Implementation of the goal g. A task consists of subtasks or an activity.
Activity a ∈ AE ∪ AT Description of what an agent can do. An activity belongs either to an
event (ae ) or a task (aτ ), and is thus either part of the set of all activities
belonging to an event (AE ) or to a task (AT ).
Intended tasks List I Every task is on this list, but ordered by priority. The priority is a func-
tion of the importance of the goal and other task characteristics. I is a
list containing all elements in TG .
Options o∈O Options are all possible activities an agent can do. O = AE ∪ AT .
Event e∈E Something that triggers the agent to reconsider his current activity.
Adds an activity to O.
Utility U (a, p) Representation of the desirability of an activity a for a person p.

Table 1: List of definitions

τ =<Description, Components, Priority, EstimatedTime, Deadline, Preconditions, Structure, Boredom,

PersonalMeaning, AutonomousMotivation, Frustration, FearOfFailure, SelfEfficacy>

where Components = aτ | Subtasks. Subtasks is a set of tasks (which can have subtasks as well). So a
task either has an activity, other tasks or nothing as its component. Preconditions are the tasks that have to
be performed before τ can be performed. Initially, we assume there are no preconditions to any tasks. The
priority determines the order in the list of intended tasks. It can be calculated with the following formula:
Eτ,p · Vτ,p
P riority(τ, p) = · Importance(gτ )
Γ·D
where Eτ,p is the expectancy of agent p succeeding in the task, Vτ,p the value, which are the costs and
benefits of performing τ , Γ the sensitivity to delay, D the delay and gτ is the goal associated with task τ .
EstimatedTime is a summation over all subtasks
X or the underlying activity:
EstimatedTimeτ = EstimatedTimeaτ + EstimatedTimeSubtasks .
Subtasks∈τ
τ and a (described next) both include task-related factors found in literature (see Section 3).

a =<Description, EstimatedTime, Preconditions, Boredom, PersonalMeaning, AutonomousMotivation,

Frustration, FearOfFailure, SelfEfficacy>

In this model, an event is a trigger for the agent to reconsider the current activity. The reason for this is
that when an agent is performing an activity, it is most likely that the agent continues with this activity. An
agent will consider switching to another activity only when a) the current activity is finished, b) the deadline
of the current activity passed or c) an event occurs.
Finally, an event e is defined as:

e =< Kind, Activity, Saliency >

where Kind can be internal or external, Activity is an activity that will be added to the options of the
agent (and, when applicable, connected with a task) and Saliency is how difficult it is to resist the activity
associated with the event. Temptations can be modeled with events. An example of this is a friend who calls
to have a drink. This is an external event, with the activity ‘Go out for a drink’ and the strength depends
on how difficult it is to resist this temptation. Internal events are governed by rules like ‘when a value (e.g.,
EgoResource) gets below a certain threshold, a specific event will happen’. External events can also be
governed by rules, but will initially be added manually to model a given situation.
The definitions of procrastination and structured procrastination are as follows:

An agent p is procrastinating when (CurrentActivityp ∈ AE or CurrentActivityp = ∅) and I 6= ∅.

An agent p is performing structured procrastination when CurrentActivityp ∈ Aτ and I 6=< τ, . . . , n >.
4.2 Process overview
This section contains an overview of the process of the model. It explains how an agent chooses activities
and what happens when events occur. Algorithm 1 gives a more formal specification.

Algorithm 1 Process overview

1: for All tasks τ in I do
2: Calculate priority
3: end for
4: Sort I
5: for each activity a do
6: Calculate U (a, p)
7: end for
8: for Each timestep t do
9: EgoResource = EgoResource · updatespeed
10: for Each activity a do
11: if aτ is finished then # Remove finished activities and their tasks
12: O = O \ aτ
13: TG = TG \ τ
14: Go back to line 5
15: end if
16: end for
17: for Each task τ do
18: if t ≥ Deadlineτ then # Remove tasks with passed deadlines and their activities
19: O = O \ aτ
20: TG = TG \ τ
21: Go back to line 5
22: end if
23: if an event e occurs then
24: if activity a is associated with a task τg then
25: #Connect activity with associated task and add to set of task-related activities
26: τg = τg ∪ τ
27: AT = AT ∪ a
28: else#Add to set of event-related activities
29: AE = AE ∪ a
30: end if
31: EgoResource = EgoResource - Saliencye
32: Go to line 1
33: end if
34: CurrentActivityp = MaxUtility(a)
35: end for
36: end for

First, the priority of all the tasks in the model is calculated to determine the order in the list of intended
tasks. Next, for each activity the utility is calculated with one of the following formulas (depending on
whether it’s a task-related activity or event-related activity):
Eaτ ,p · Vaτ ,p Eae ,p · Vae ,p
U tility(aτ , p) = · Bτ U tility(ae , p) =
Γp · Daτ t − EstimatedTimeae
where Eaτ ,p is the expectancy, Vaτ ,p the value, Γp the sensitivity to delay, Dτ the delay, Bτ a small bonus
when an activity is associated with a task in the list of intended tasks and t the current time step. The for-
mulas to calculate these variables are given below. The weights w determine how strong a factor influences
procrastination. The formulas include all factors discussed in the conceptual model. Some factors have a
positive effect, others a negative, and this is dependent on the influence of this factor on procrastination.

Eaτ ,p = Selfefficacyaτ · wselfefficacy + SelfEsteemp · wselfesteem + AdjustedFearOfFailurep · wfearOfFailure

 
FearOfFailurep when SelfEfficacyaτ > 0, 5
AdjustedFearOfFailurep =
−FearOfFailurep else
Vaτ ,p = −Boredomaτ · wboredom −Frustrationaτ · wfrustration −AutonomousMotivationaτ · wAutonomousMotivation
−SensationSeekingp · wSensationSeeking +Conscientiousnessp · wconscientiousness +EgoResourcep · wEgoResource
+PersonalMeaningp · wPersonalmeaning +Moodp · wMood

Γp = Impulsivenessp · wImpulsiveness & Daτ = Deadlineτ − t − EstimatedTimeaτ

 position of τ in I
Bt = 1 −
number of items in I

In each time step, EgoResource is updated. Next, it is checked whether activities are finished, and if so,
they are removed from the options. Tasks with deadlines that are passed are also removed. After this, the
utility of all activities is calculated again. When an event occurs, it is checked whether they belong to a task
that doesn’t have an activity yet. If this is the case, the activity is connected to this task and added to the
task-related activities Aτ . Otherwise, the activity is added to the set of event-related activities Ae . Because it
requires ego resources to resist the event, this value is also updated (depending on the saliency of the event).
After this, the utility of all activities is calculated again. If none of the previously described conditions
are the case, the activity with the highest utility is performed. When this activity is in Ae , the agent is
procrastinating because the agent is performing an activity that does not belong to a task therefore is not part
of the agent’s goals. When an agent is performing an activity that does not belong to the task on top of the
list of intended tasks, it is structured procrastination.

5 Evaluation plan
The proposed model described in the previous sections should provide insights into the mechanisms of
procrastination. We will evaluate the model in three stages. First, we will implement the model in Matlab
and run some initial simulations. The weights will be determined in consultation with a procrastination
expert and will be the same for all scenarios. A few different scenarios with different kinds of agents (e.g.,
with high or low conscientiousness scores) will be tested to see if the expected behaviour will follow from the
model. Second, we will use an existing data set from a previous psychological experiment on procrastination
as training data to determine the appropriate weights for the model. Supervised learning can be used to train
the model. The trained model will be evaluated by using another data set as a test set. Third, the trained
model will be used to make predictions about behaviour under certain conditions. These predictions will be
tested empirically in collaboration with psychologists.

6 Conclusions & Future Work

In this paper, a conceptual model of procrastination is presented that includes prominent factors from psy-
chological literature that influence procrastination. In addition, it presents a formal specification of this
model that will serve as the basis of a computational model, which in turn will be used to validate the
conceptual model. In future work, the stages presented in the evaluation plan (Section 5) will be executed
to validate the model. The complexity of the proposed model mirrors the complexity of the phenomenon.
Depending on the outcome of the evaluation the model may be pruned, but we did not want to dismiss any
empirically identified factors out of hand.
Procrastination is problematic for many people because it negatively influences people’s performance
and well-being. There are a number of existing coping strategies (e.g., self-forgiveness [22] and making
very concrete plans [21]), but because procrastination is complex behaviour, some strategies may have been
unexplored. The validated model can be used to explore whether there are points of influence that the model
predicts will decrease procrastination, that as of yet are unused in interventions. This will lead to hypotheses
for further empirical studies. Another application of the model is to develop e-coaching software that can
support people on an individual basis (by identifying the factors that have the most influence for this indi-
vidual) in their struggle against procrastination.
Acknowledgments. This research was supported by Philips and Technology Foundation STW, Nationaal
Initiatief Hersenen en Cognitie NIHC under the Partnership programme Healthy Lifestyle Solutions. We
thank Floor Kroese from the Self-Regulation Lab for reviewing and discussing the conceptual model. We
thank Saskia van Dantzig (Philips) for extensive comments on an earlier version of this paper.

References
[1] Bandura, A. et al. (1989). Human agency in social cognitive theory. American psychologist 44(9),
1175–1184.
[2] Baumeister, R. F. (2003). Ego depletion and self-regulation failure: A resource model of self-control.
Alcoholism: Clinical and Experimental Research 27(2), 281–284.
[3] Blunt, A. and T. A. Pychyl (1998). Volitional action and inaction in the lives of undergraduate stu-
dents: State orientation, procrastination and proneness to boredom. Personality and Individual Differ-
ences 24(6), 837–846.
[4] Blunt, A. K. and T. A. Pychyl (2000). Task aversiveness and procrastination: a multi-dimensional
approach to task aversiveness across stages of personal projects. Personality and Individual Differ-
ences 28(1), 153–167.
[5] Deci, E. L. and R. M. Ryan (2008). Self-determination theory: A macrotheory of human motivation,
development, and health. Canadian Psychology 49(3), 182–185.
[6] Frederick, S., G. Loewenstein, and T. O’donoghue (2002). Time discounting and time preference: A
critical review. Journal of economic literature 40(2), 351–401.
[7] Haghbin, M., A. McCaffrey, and T. A. Pychyl (2012). The complexity of the relation between fear of
failure and procrastination. Journal of Rational-Emotive & Cognitive-Behavior Therapy 30, 1–15.
[8] Harriott, J. and J. R. Ferrari (1996). Prevalence of procrastination among samples of adults. Psycholog-
ical Reports 78(2), 611–616.
[9] Klingsieck, K. B. (2013). Procrastination: When good things don’t come to those who wait. European
Psychologist 18(1), 24.
[10] Little, B. R. (1983). Personal projects a rationale and method for investigation. Environment and
behavior 15(3), 273–309.
[11] O’Donoghue, T. and M. Rabin (2001). Choice and procrastination. The Quarterly Journal of Eco-
nomics 116(1), 121–160.
[12] Perry, J. (2012). The Art of Procrastination: A Guide to Effective Dawdling, Lollygagging and Post-
poning. Workman Publishing Company.
[13] Pychyl, T. A., J. M. Lee, R. Thibodeau, and A. Blunt (2000). Five days of emotion: An experience
sampling study of undergraduate student procrastination. Journal of Social Behavior & Personality.
[14] Schouwenburg, H. C. and J. Groenewoud (2001). Study motivation under social temptation; effects of
trait procrastination. Personality and Individual Differences 30(2), 229–240.
[15] Schouwenburg, H. C. and C. H. Lay (1995). Trait procrastination and the big-five factors of personality.
Personality and Individual Differences 18(4), 481–490.
[16] Schraw, G., T. Wadkins, and L. Olafson (2007). Doing the things we do: A grounded theory of
academic procrastination. Journal of Educational Psychology 99(1), 12.
[17] Steel, P. (2007). The nature of procrastination: A meta-analytic and theoretical review of quintessential
self-regulatory failure. Psychological Bulletin 133(1), 65–94.
[18] Tice, D. M. and R. F. Baumeister (1997). Longitudinal study of procrastination, performance, stress,
and health: The costs and benefits of dawdling. Psychological Science, 454–458.
[19] Van Broekhuizen, H. (2010). Write your paper now: Procrastination, conscientiousness and welfare.
Stellenbosch University, Department of Economics Working Papers.
[20] Vansteenkiste, M., E. Sierens, B. Soenens, K. Luyckx, and W. Lens (2009). Motivational profiles from
a self-determination perspective: The quality of motivation matters. Journal of Educational Psychol-
ogy 101(3), 671.
[21] Wieber, F. and P. Gollwitzer (2010). Overcoming procrastination through planning. In The Thief of
Time: Philosophical Essays on Procrastination, pp. 185–205. Oxford University Press.
[22] Wohl, M. J., T. A. Pychyl, and S. H. Bennett (2010). I forgive myself, now I can study: How
self-forgiveness for procrastinating can reduce future procrastination. Personality and Individual Dif-
ferences 48(7), 803–808.