Original article

doi: 10.1111/j.1365-2729.2008.00279.x

Serious games for higher education: a

framework for reducing design complexity
W. Westera, R.J. Nadolski, H.G.K. Hummel & I.G.J.H. Wopereis
Educational Technology Expertise Centre, Open University of the Netherlands, Heerlen, The Netherlands

Abstract Serious games open up many new opportunities for complex skills learning in higher education.
The inherent complexity of such games, though, requires large efforts for their development.
This paper presents a framework for serious game design, which aims to reduce the design com-
plexity at conceptual, technical and practical levels. The approach focuses on a relevant subset
of serious games labelled as scenario-based games. At the conceptual level, it identifies the
basic elements that make up the static game configuration; it also describes the game dynamics,
i.e. the state changes of the various game components in the course of time. At the technical
level, it presents a basic system architecture, which comprises various building tools. Various
building tools will be explained and illustrated with technical implementations that are part of
the Emergo toolkit for scenario-based game development. At the practical level, a set of design
principles are presented for controlling and reducing game design complexity. The principles
cover the topics of game structure, feedback and game representation, respectively. Practical
application of the framework and the associated toolkit is briefly reported and evaluated.

Keywords serious games, complex skills, design framework, higher education, scenario-based games.

widely adopted by new generations of learners, the

Introduction
so-called digital natives, who have grown up immersed
Various authors anticipate the great opportunities of in new communication technologies (Prensky 2006).
games (and simulations) in education because of posi- Although the games industry has grown rapidly over
tive effects on learning outcomes (see, e.g. Kiili 2005; the last decades, the use of games in education is still
Quinn 2005; De Freitas 2006; Amory 2007). Games limited. Despite the great opportunities for synergy,
have demonstrated to provoke active learner involve- the game industry and education largely act as well-
ment through exploration, experimentation, competi- separated sectors driven by their own missions and
tion and co-operation. They support learning because of objectives. Education and games share the idea that
increased visualization and challenged creativity. They participants have to achieve some goal, but game
also address the changing competences needed in the objectives do not necessarily match pursued learning
information age: self-regulation, information skills, objectives. Squire (2002) reports that attitudes to com-
networked co-operation, problem solving strategies puter and video games are likely to support hypercom-
and critical thinking. Importantly, games have become petitiveness and warped sexual values rather than
well-considered learning goals. Garris et al. (2002) and
Ma et al. (2007) note that game-based learning often
Accepted: 3 February 2008 concerns repetition of cyclic content. This provokes per-
Correspondence: Wim Westera, Head of Educational Implementa- sistent re-engagement, which tends to address lower
tion, Educational Technology Expertise Centre, Open University of
the Netherlands, Valkenburgerweg 177, NL-PO Box 2960, 6401 DL level learning goals rather than higher level goals. Espe-
Heerlen, The Netherlands. Email: [email protected] cially in higher education, the mental mode of learning

420 © 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd Journal of Computer Assisted Learning (2008), 24, 420–432
Serious games design framework 421

which reflects profundity, reflection, concentration and conceptual framework (Nadolski et al. 2007). At the
perseverance seems to conflict with the mental mode of practical level, we will present and explain principles for
gaming which is commonly associated with amuse- reducing game-design complexity. In conclusion, we
ment, fun and relaxation. In response to the one-sided will discuss the outcomes of this synthesis.
entertainment image of games, educators increasingly
use the term ‘serious games’ to indicate that fun and its
Literature review
positive effects on motivation are not the main motives
for using games in education (Serious Game Initiative, The domain of games covers a great variety of game
http://www.seriousgames.org/; Michael & Chen 2006). genres and modes of play (Gredler 1992, 2004; Rieber
In higher education, the main objective of applying 2005). Also, very different disciplines are involved like
games is to engage learners in complex problem spaces text writing, software development, artificial intelli-
that mimic real world situations, without importing gence, graphics design, video production, marketing,
unwanted constraints and risks of the real world. Learn- sales and many others. This inherent diversity produces
ers are challenged to develop relevant knowledge repre- a greatly fragmented domain, both from the perspective
sentations and the associated reasoning and problem- of design methodology and the underlying theories. The
solving strategies. To this end, learners are supposed to domain of game design is considered semi-formalized,
be confronted with ill-defined problems that often allow fuzzy and incoherent (Björk & Holopainen 2005) and it
multiple solutions and require the application of neces- eclectically combines various approaches that cannot
sary methodologies or tools and collaboration with be fully covered by prescriptive or even descriptive theo-
fellow learners. An important impediment for such ries. There have been various efforts to produce
games, though, is the large efforts needed for their devel- co-ordinating frameworks for explaining and designing
opment. The development of serious higher education complex games (Hunicke et al. 2004; Salen & Zimmer-
games has proven to be complex, time consuming and man 2004; Björk & Holopainen 2005; Kiili 2005;
costly (Westera et al. 2003; Nadolski et al. 2007). Exist- Amory 2007). But as De Freitas and Oliver (2006)
ing approaches and toolkits for developing leisure signal, a dedicated framework for educational games is
games cannot simply be transferred to education lacking, which causes a significant impediment for their
because of their very different characteristics. The main application. Many frameworks choose a technical per-
objective of this paper is to provide a framework for spective by providing authoring environments and other
serious game design, which serves these pedagogical tools for game development (i.e. Java, .Net, mobile
goals and which simultaneously helps in reducing the learning). Hunicke et al. (2004) have developed the
efforts needed for design, development and exploitation. Mechanics, Dynamics, Aesthetics (MDA)-framework,
The framework offers a coherent description of the which attempts to bridge the gap between game design,
games’ basic components and their interrelationships. It game criticism and technical game research by itera-
also provides a shared vocabulary for game design and tively switching between three different levels of
directions for efficient game development. We will first abstraction during game design. Unfortunately, the link
review existing game design frameworks and strategies with education is weak. Salen and Zimmerman (2004)
and explain their limited applicability for educational have produced a well-elaborated conceptual framework
game design. Next, we will go into the topic of creating for the analysis of games.Again, little attention is paid to
complex and challenging contexts for learning and education and learning. De Freitas and Oliver (2006)
narrow down our scope to a relevant subset of serious review various frameworks that can help teachers evalu-
games labelled as scenario-based games. The frame- ate the appropriateness of educational games and simu-
work will be presented in three steps. At the conceptual lations for a particular learning context. The paper
level, we will describe the separate components that presents a four-dimensional framework that extends the
make up the game. At the technical level, we will present existing ones. The framework focuses on evaluation
the basic system architecture for scenario-based game rather than design and thus, it does not present strategies
development. For reasons of illustration, we will explain or methods for game design. In order to strengthen the
its application by referring to the existing Emergo toolkit link between game design and educational design,
(http://www.emergo.cc/), which has been based on the Burgos et al. (2007) have suggested to develop a frame-

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

422 W. Westera et al.

work based on the IMS Learning Design technology learning. Theoretically, there are two opposite ways for
specification (IMS 2003). Practical application of the generating complexity. First, complex structures can
framework is hampered because it would require emerge from the self-organized nature of a collection of
complex tooling. Amory (2007) presents a game object interacting agents. Here, examples would be plants, ant-
model for the development and analysis of computer hills, climate, ecosystems, etc. Such systems are charac-
video games. The model identifies the key concepts of terized by large numbers of agents interacting according
games and it has been recently updated to include educa- to simple generic rules. As in common multiplayer
tional goals. Due to its general and theoretical level of games like football or baseball, very complex situations
reasoning, it is of little direct significance for practical, can emerge from the simple rules of play that regulate
educational game design (Kiili 2007). Because of the the behaviours of the participants (agents). Each run of
insurmountable diversity of games, Björk & Holopainen such game will be greatly different and unpredictable,
(2005) have suggested to work with game design pat- which indeed makes these games quite exciting.
terns, which describe well-identified re-occurring prob- Exactly such uncertainty of the system is a direct threat
lems in game design and offer reusable solutions. The for education because it is unclear whether the emerging
approach neatly addresses the diversity of the domain: it events match the required learning goals. Second, com-
provides a shared vocabulary of game elements, allows plexity can be created by design. The great advantage
structured comparison between games and supports of such an approach would be that the complexity
component-based tooling. Over 200 game design pat- is largely controlled, as in many human artefacts like
terns have been identified. However, the framework pro- microcircuit design, the production of medicines or the
vides only very little support for the integration of construction of aeroplanes. Because education is about
patterns into an appropriate game. Dickey’s (2005) well-tailored learning rather than carefree playing, edu-
research suggests that various game design strategies for cation is prompted to create complexity by design. Even
achieving player engagement (like narrative, viewpoint, when the growing interest in Web 2.0 and social net-
setting and roles) can be applied for the design of engag- working increasingly demonstrate the possibilities of
ing learning environments. Kiili (2007) focuses on self-organized structures, the behavioural complexity of
games for experiential learning and looks into the con- higher order competences cannot be covered in all
ditions that contribute to achieve experiential flow. extent by simple generic rules. It is very unlikely that
Although the research of Kiili explicitly links educa- agent-based approaches combined with simple behav-
tional theory with game design, it sincerely reports not to ioural rules – as displayed in football or baseball –
be able to address or improve game design. Various produce learning environments that are sufficiently
approaches have been developed to cover specific types tuned to support the effective acquisition of predeter-
of games, for instance, classroom games or massive mined academic learning objectives. An agent-based
multi-user online games (Steinkuehler 2004). These game design would require a large set of behavioural
approaches have only little significance for serious game rules in order to stay tuned to its objectives; it thus
design. For audio-visual design and narrative, various would drift toward a complexity by design approach.
frameworks are available, but these are not necessarily Although emergent complexity can be useful to a
tuned to audio-visuals in serious game design (Westera certain extent – as with random noise – serious games
1999; Koumi 2006). Altogether, quite some valuable cannot do without a pre-structured complexity that is
frameworks are available that enhance our understand- based on a narrative or scenario, covering the dynamics
ing of games. But they generally provide only little of the game. Indeed, Murray (1997) and Crawford
design directions, often lack a pedagogical perspective (2005) stress the importance of narrative content in
and fail to suggest how to deal with game complexity. computer-based learning. Lindley (2005) also stresses
the importance of pre-designed narrative content
because it allows structured experiences in time, but still
Method
enables the players to exert their personal preferences.
Although complexity is a keyword for serious games, Such scenarios allow education to emphasize functional
education does not require complexity per se; it complexity and to reduce types of complexity that are
demands controlled complexity in support of effective irrelevant to learning.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

Serious games design framework 423

So the starting point of our approach is designed specification of the relevant elements while taking into
complexity rather than emergent complexity. It thus account the two fundamental dimensions of space and
identifies a subset of serious games which can be time. The space dimension covers the static configura-
labelled scenario-based serious games or case-based tion of the (virtual) game locations, including the
serious games (Galarneau 2005; Nadolski et al. 2007). associated objects, attributes and relationships. Its
Here, predefined rules are used rather than individual development over time covers the dynamics of the
agents to arrange programme flow, required achieve- game, viz. the evolution and the state changes of the
ments and access to new features. There are many game elements. At the technical level, the framework
examples of such games (occasionally labelled virtual describes the basic system architecture of scenario-
practicals) that conform to the demands of higher edu- based game development; it characterizes the required
cation (Nadolski et al. 2007). They all share the ten- tooling system and its separate tools to deal with the
dency to designed complexity and the associated great development of game locations, objects, actor roles and
development efforts. Before elaborating our approach, game scenarios. At the practical level, the framework
we summarize the main characteristics of the subset of provides principles for reducing design complexity.
pursued scenario-based serious games. These principles refer to the way the game is structured,
to the feedback that is offered to learners and to the
• The environment multimedia representation of the game environment,
Learners are exposed to a challenging context for respectively. A glossary of relevant terms that are used
learning that mimics real world situations, displays in the framework is available at the end of this paper.
ambiguity and conflicting information and offers
large degrees of freedom.
• The learning activities Framework description part 1: conceptual level
Learning activities involve complex decision making, We will start with the static game configuration. The
problem solving strategies, intelligent reasoning and educational game environment is assumed to comprise
other higher order skills. They are based on profes- four distinct subsystems: the world of game play, the
sional or academic role adoption and modeled after learner world, the teacher world, the game management
expert behaviour. Learners are in charge to deal with world (Fig 1).
complex problems according to professional or scien- In the world of game play, learners carry out their
tific standards. learning tasks by accessing the locations of the game
• Multi-user and the game objects. The learner world is a separate
Complex real world problems are likely to involve environment that extends and complements the context
several participants. The games should thus allow of the game play. It contains the assignments and the
multi-user scenarios. Single user solutions are pos- players, which may be individuals as well as groups. It
sible when some of the actor roles are covered by the allows players to cross the so-called magic circle (Huiz-
computer. inga 1938/1955), which is the social ‘contract’ agreed
• Methodology upon by the players, that defines the physical and tem-
Expert behaviour is framed by one or more method- poral boundaries of the game. Likewise, the tutor world
ologies or strategies, which can be used to counter
trial and error behaviours, to control complexity and
to act as a reference for generating relevant feedback Game management world
during the game.

Learner world Teacher world

The proposed game design framework applies essen-
tially to the subset of scenario-based games. It combines Game play world
three different levels of description. At the conceptual
level, a game is considered a system, a set of interrelated
or interacting elements that constitutes a complex
aggregate (Littlejohn 1983). Designing a game involves Fig 1 Subsystems of the educational game environment.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

424 W. Westera et al.

provides a metachannel, transcending the game play, The world of game play need not be a closed, self-
which allows tutors to assess progress, provide feed- contained environment. In addition to the predefined
back and make interventions. In pervasive games, the locations and objects, players may also extend the game
magic circle is blurred because game play and ordinary universe by utilizing external channels and objects in
life become greatly intertwined (Montola 2005). This order to optimize their performances. Some of these
blurring may also occur in education when the learning external channels can be incorporated in the game logic.
is greatly connected with ordinary life, as in various For instance, the development of share prices or topical
modes of informal learning, apprenticeship learning weather reports can easily be fed into the scenario
and workplace learning (Westera et al. 2000). The engine to enhance the dynamics and the authenticity of
fourth subsystem is the world of game management. the game. External channels also allow the incorpora-
Here, the different game runs are arranged and tion of expert consultation, external forums, external
co-ordinated, users are registered and new game sce- tools, external archives, which are not necessarily
narios are added. Below, we will focus on the world of linked with the game logic.
game play. Indeed, as Kiili (2005) notices, game play is Next, we will go into the dynamics of the game. The
the most critical feature of game design. dynamics of the game include the state changes of the
Objects are the building blocks of the game environ- various game components. According to Björk and
ment. When applying a spatial metaphor, the principal Holopainen (2005), playing a game is just making
objects of the game environment are the game’s loca- changes in game states. The overall game state is the
tions. The game world may comprise different loca- aggregate of the states of all components of the game,
tions, which can be visited and which may provide including the progressive states of learners. State
access to different tools, resources or fellow players. changes are also referred to as events. In order to make
Locations may also cover abstract entities like pages, events take place, any status information has to be
table cells or other representations that define a particu- tracked and stored for all objects and players of the
lar environment for specific actions. Navigation to the game. State changes are triggered when certain condi-
various locations may be regulated by admission and tions within the game are met. The collection of trig-
prohibition rules. For instance, one might state that gers and the underlying rules acting on the game world
location A can only be accessed from location B or and its users is referred to as the game logic: it com-
location C, being its neighbour locations, but not from prises the constitutive and operational rules of the game
location D. Also, learner performance conditions may (Salen & Zimmerman 2004). Naturally, many state
be used to regulate location access (A is accessible only changes will be triggered by the players’ actions, for
if learner X has visited location Y, or has completed instance, moving an object, selecting a document or
task Z, etc.). All too often, games display many scenery completing a multiple choice question, etc. But state
objects that have aesthetic or atmospheric value as changes may also occur by internal triggers that regu-
such, but which are irrelevant to the problem solving late the autonomous evolution of an object’s state
process. Unless such extensions are necessary to effect attributes over time. Attributes may concern any object
a realistic, cluttered context, they are likely to distract quality: position, angle, speed, colour, size, age, vis-
the learning rather than support it (Koumi 2006). Loca- ibility, etc. Computational models may be used for
tions are assumed to contain various types of objects, incorporating gravity, kinematics, the daily cycle of
be it tools like machines or telephones, or knowledge light and dark, for birth and death of objects or, in
resources like videos or documents. Naturally, the general, population dynamics and system evolution.
objects should have some relevance for the learner’s In many games, complexity and unpredictability are
assignment, that is, they should display relevant created by random triggers, often in conjunction with
features or allow appropriate interaction. Locations some probability distribution function. For critical
may also contain subjects (which formally are a events, however, users may find it difficult to anticipate
subclass of objects). These are fellow participants these unexpected, arbitrary changes. Indeed, from
who make up the social dimension of the game by film theory, it is known that viewers should be
enabling communication, co-operation, opposition or prepared to radical changes of state by providing
competition. explicit of implicit clues in order to avoid frustration

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

Serious games design framework 425

Fig 2 A decision tree as the abstract lay-

out of subsequent learner options.

or annoyance (Koumi 2006). For the larger part, Scenario builder

state changes in scenario-based games are controlled

by predefined rules, scripted as Boolean expressions. Location builder
Defining such Booleans may be very laborious and
complex because numerous rules are needed and Object builder Game play world
each rule may contain any state parameter of any
object. This is an important impediment for the devel- Role builder
opment of scenario-based games. An essential condi-
tion for the feasibility of scenario-based games is the
State logging
restriction of the overall number of system states. Prac-
tically, this means that the game scenario aims to
Fig 3 Basic architecture of scenario-based game development.
reduce the option space for learners, that is, at any loca-
tion or point in time, learners may choose only from a
limited number of alternative actions. The basic layout
Framework description part 2: technical level
of such focused scenario reflects a decision tree,
which incorporates all possible routes learners can take At the technical level, a tooling system would be
(cf. Fig 2). needed, which allows game development and
Note that the basic pattern of Fig 2 only depicts the implementation. Essential features of such tooling
principle of restricted option space. It is about deci- system follow directly from the conceptual framework.
sions, not about content selection or content testing, and Its basic layout is displayed in Fig 3.
therefore, it should not be mistaken for a simple collec- From the perspective of game play, the game devel-
tion of successive multiple choice questions. Likewise, opment architecture comprises four types of tools. First,
the structure does not constrain the games to closed a location builder is necessary to define the game
scenarios. For instance, when at a certain point in time locations. Second, an object builder is needed to specify
learners are asked to consult a subject matter specialist objects and the associated objects’ attributes. Various
of their choice or to produce an original text, musical types of objects will be needed to build a sensible game.
score or video clip, the outcome is uncertain, but still, it Therefore, the object builder will comprise a series of
is one option. different building tools. Third, the role builder allows to

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

426 W. Westera et al.

specify different player roles, associated with different

tasks and different powers. Fourth, a scenario builder
helps in specifying the logical conditions that make up
the game play scenario. Each object or each class of
objects should exhibit a range of well-defined state
attributes that can be addressed and influenced by the
scenario builder. Naturally, during game play, the
execution of the game logic requires monitoring and
logging of all state data of the system. These include the
states of locations, the states of objects and the states of
the players.
Technically, the architecture of Fig 3 may be imple-
mented with any programming platform. To demon-
strate the power of the approach, we refer to the Emergo
Fig 4 Video interview in one of the games. The learner can
methodology and toolkit (Nadolski et al. 2007), which
consult non-playing characters by selecting questions from the list.
have been developed for the efficient creation of
complex scenario-based games. The Emergo project
has identified a number of essential objects and interac- tasks to be completed first, and what learner output is
tion modes for scenario-based game play. The Emergo required for the task to be completed. The attributes
toolkit contains various authoring and scripting tools to ‘accessed/not accessed’ and ‘completed/not completed’
define and assemble scenario-based games. It uses a indicate important state values for tracking learner
layered architecture according to the model-view- progress and providing feedback.
controller model and it applies web services for the
interfaces with external systems. It is fully web-based The e-mail builder
and uses the J2EE/Java development platform. The This tool allows developers to generate pseudo e-mail
toolkit has been made available under a General Public messages that will become available for the players
License (http://www.emergo.cc/). Below, we will when certain conditions are met. The e-mail messages
briefly describe a relevant subset of the Emergo tools. may contain clues within the scenario that induce the
learners to undertake actions or to adapt their strategies.
The location builder The relevant state values of an e-mail message would be
This building tool is used for the definition of locations ‘accessed by player X’ and ‘not accessed by player X’.
that are relevant to the game scenario. Locations are
simply represented on the screen by a background The selection builder
graphic or photograph and a label. By defining a loca- This tool is used to offer learners the opportunity to
tion rule (using the scenario builder), the background explicitly select relevant options. This includes single
graphic may be switched. This way, the location may answer questions (a deliberate decision) and multiple
have different appearances at different conditions (day- answer questions (filtering of a relevant subset of
light, night-time, weather appearance, etc.). An impor- options). Also, the ranking of items can be included.
tant state parameter of a location is ‘visited by player’ Naturally, the choices are logged and applied in the
(yes/no). Such parameters are logged and are available game logic to generate new events or feedback.
for defining Booleans with the scenario builder. The list
of locations is fed into the navigation controls. The interview builder
The interview builder enables the game developer to
The task builder create a predefined interview. It applies lists of ques-
The task object builder is used to specify learning tasks. tions that are linked to video streams of the interviewee
It simply comprises a text field for the description of the (see Fig 4). This way, non-playing characters within the
task assignment and some attributes to indicate whether game play can be consulted (i.e. a subject matter expert)
the task is optional or not, whether it requires previous or can be assessed (i.e. diagnosing a patient).

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

Serious games design framework 427

The interview object’s state parameters include Table 1. Summary of the general state attributes of the various
‘question asked’ (yes/no) and ‘video played’ (yes/no). game components with regard to each player.

Evaluation of the interview may be carried out by check- Game component State parameter
ing what questions have been asked and what questions
have been skipped. Alternatively, the learner may be Location Visited/not visited
Number of visits
asked to make a summary of the interview by using the
Time of visit
report tool, which is also part of the Emergo toolkit. Duration of visit
Activities Number of start-up trials
The object evolution builder Accessed/not accessed
Tool contents Viewed/not viewed
This tool allows the game developers to define autono-
Question asked/not asked (interviews)
mous changes of an object’s abstract states. This is Used/not used
achieved by identifying relevant properties and associ- Learning tasks Started/not started
ated state values (for instance room temperature: 38 °C, Completed/not completed
or colour: blue). The object’s state changes may be trig- Learners products Started/not started
Completed/not completed
gered by time (evolution, life cycle), by player interac-
Non-playing Met/not met
tions or by other changing conditions of the game play characters
world. Time Nett time since game run start
Nett time since user start
Resources Viewed/not viewed
The role builder
Interrupted/not interrupted
The role builder is a management tool which allows the
specification of multi-user scenarios by assigning dif-
agendas, portfolios and reports and tools for addressing
ferent roles and powers to each user.
learning resources. Moreover, the modular architec-
ture of the toolkit allows extension with additional
The scenario builder game-building tools. In accordance with the design
The scenario builder is the intelligent core of the game framework, an ADDIE-based methodology for scena-
development system. It is connected to all objects and rio-based serious games has been described by (Nadol-
their state values and it is largely used to define the rules ski et al. 2007) (ADDIE comprises respective stages of
of the game dynamics. The general form of a rule reads: Analysis, Design, Development, Implementation and
Evaluation). The methodology offers important consid-
If state parameter (X) = p, (1) erations for designing scenario-based games. But still,
then state parameter (Y) = q
the natural tendency of toolkits and methodologies is to
The state parameters may concern any objects, subjects enable a wide range of game applications differing in
or locations within the game environment. The anteced- size, purpose and complexity, and therefore, game
ent of the rule may be extended to any Boolean combi- designers are likely to stumble into the pitfall of unman-
nation of state parameters. Likewise, the consequence ageable complexity, turning the toolkit’s strengths into
of the rule (the event) may be extended to describe mul- a fundamental weakness. Since the states of many game
tiple state changes. The state changes cover various objects will be greatly interlinked with the learners’
operations, which include showing and hiding of states, a combinatory explosion of required specifica-
objects, generating feedback, generating an e-mail, tions may occur. In order to counteract this risk, meth-
setting an internal game parameter and locking or odologies for scenario-based game design need to be
unlocking of locations. The state parameters reflect the extended with explicit guidelines how to deal with
key elements of the game dynamics; Table 1 summa- complexity.
rizes the general state attributes of the various game
components with regard to each player.
Framework description part 3: practical level
It should be noted that the Emergo toolkit contains
various additional building tools that are not explained On top of the conceptual framework, the development
in this paper, for instance, tools for creating tests, system architecture and the design methodology, a set

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

428 W. Westera et al.

of design principles are needed for controlling and Often many branches of the decision tree can be
reducing game design complexity. Any scenario-based omitted. Shallow distracters can be applied to create
game-design methodology should incorporate such the impression of option spaces. These can be dummy
principles to support the affordable development of options that suggest a sensible alternative, but when
complex serious games. The principles cover the topics activated, a message is displayed that the option is not
of game structure, feedback and game representation, available. Second, a dead end distracter is used
respectively. We will explain these below. involving some tasks or activities, but at certain con-
ditions (e.g. time past or completion), these are termi-
Reducing structural design complexity nated and the learners return to the original decision
From Fig 2, it can be observed that the complexity of the node to reconsider their behaviours. Third, a dis-
decision tree increases rapidly with increased depth. tracter can also act as a single sidestep of the main
Although such complexity matches the starting points road; completion of the distracter would redirect the
of educational games, it would take a lot of effort learners back to the preferential scenario, with or
to elaborate and specify all decision nodes. Four without any corrective comments. Eventually, learn-
approaches are proposed that reduce structural design ers will find out, either by inhibition rules or by cor-
efforts. rective feedback that they have spend valuable time at
irrelevant activities.
• Option width rather than option depth • Closures
In conflict with the sequential nature of scenarios or Game closures are certain game states that corre-
narratives, the complexity of the decision tree is best spond with the completion of one or more achieve-
reduced by choosing small values for the depth. For ments and that trigger access to a new episode,
instance, swapping the values of depth (n = 4) and activity or level of the game. Closures reduce design
width (m = 2) of Fig 2 reduces the number of nodes complexity since they mark the start of a new tree.
from 63 to 29 (cf. Fig 5). Paradoxically, learners may They often come with methodologies, which usually
experience larger complexity in this alternative struc- comprise a number of subsequent phases, as in scien-
ture because experienced complexity is determined tific research. Closures are removed when learners
by instantaneous option space (width) rather than have met certain criteria in their game play. Basically,
sequential option space (depth). For large tree depth three different types of criteria can be distinguished:
m, halving of m reduces the tree complexity of order activity-based, route-based and performance-based.
nm approximately a factor √n. • Parallel task execution
• Shallow distracters Complex tasks often require the parallel or iterative
execution of activities. Many strict methodologies are
sequential in kind, but cannot be applied unthinkingly
to practice. Parallel task execution is difficult to
describe in a single decision tree. As the learner is
supposed to occupy only one single position in the
tree, various cross-referential bypasses would be nec-
essary to enable traffic from one branch of the tree to
the other. Ultimately, all nodes might be connected to
each other, which would produce very complex struc-
tures and routes. Instead, it is important to identify the
separate underlying activity classes and model these
as separate decision trees with their own structures
and their own state logging. Thus, multitasking condi-
tions are created which allow learners to temporarily
switch from one activity tree to a second activity tree,
Fig 5 Reducing design complexity by offering option width and to return after a while to the recorded position in
rather than option depth (cf. Figure 2). the initial tree.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

Serious games design framework 429

games are promising environments for experiential

Reducing feedback design complexity
learning, but they tend to focus on action rather than
For educators, it is often hard to resist the tendency of reflection, which greatly conflicts with experiential
tracking and evaluating any learner action so that they learning theory of Dewey (1938) and Kolb (1984).
are able to provide detailed feedback. Commenting on Since peer feedback establishes reflection as a distinct
any single step in the game, however, would require an activity, it may improve the intrinsic quality of
almost indefinite number of extended rule sets to gener- learning.
ate the appropriate feedback. Two separate approaches
are proposed to reduce the efforts of feedback design:
Reducing representation design complexity

• Strategic performance feedback Many entertainment games owe their successes to out-
Rather than commenting on a microlevel, feedback standing graphic sceneries, character animations and
should be generated at high aggregation levels. Learn- sound. These high representation standards put high
ers would not be too interested in receiving direct pressures on educational game developers to attain the
feedback on every single action or decision, but same level of sophistication and complexity. To this end,
instead, they would want to be informed about their it is important to distinguish between authenticity of
overall progress, about their approach, how well they content and authenticity of representation. Naturally,
perform and how they can improve it. This nicely both are relevant, but there is some evidence that authen-
matches the strategic nature of scenario-based ticity of representation is less critical than authenticity of
problem cases. The associated assessment model content. In serious games, the content itself contributes
should be restricted to only a few aggregate dimen- substantially to tension and involvement. In addition, the
sions, for instance, the quality of performance, the media equation theory of Reeves and Nass (1996) claims
speed of performance and performance style. Quality that only very little representational or technological
of performance presupposes a rating method that is efforts are necessary to provoke true interpersonal
based on start or completion of necessary activities. responses. An example would be original games like
Speed of performance can be easily related to (nett) Pac-Man wherein simple graphical interfaces arouse
session time. Performance style measurement would great tension and excitement of players, even when the
incorporate some of the learners’ routing behaviours, monsters hardly look like monsters. Apparently, what
the degree of parallel processing and probably the counts is not realism or authenticity but credibility.
learners’ responses to natural feedback. Without trivializing the scenery of the game play world,
• Including peer feedback it is important to note that a functional analysis of the
As Amory (2007) puts it, educational games should game contents and the game scenario is necessary to
support authentic learning activities that are designed decide on representation topics. The proposed frame-
as narrative social spaces. The multi-user dimension work provides a useful vocabulary for recognizing the
opens up great opportunities for the arrangement or functional importance of game components. Naturally,
emergence of peer feedback. Several authors report the representational efforts in game components should
that giving and receiving peer feedback has positive properly align with their functional importance.
effects on motivation, reflection, self-esteem and
commitment (Fantuzzo et al. 1989; Anderson et al.
Discussion
2000) and produces higher learning outcomes (Fan-
tuzzo et al. 1989; Wong et al. 2003; van Rosmalen In this paper, a conceptual framework for scenario-
et al. 2005). First, peer feedback may be part of the based game design has been proposed. The framework
game scenario, as in many management games where has been applied to develop a system architecture and
company teams compete against each other. Second, the associated tools for efficient scenario-based game
peer feedback can also be arranged outside the game development. A corresponding design methodology for
scenario, transcending the game play world by stimu- scenario-based games has been explained elsewhere
lating the community of players to comment on their (Nadolski et al. 2007). An important goal of the
game performances. As Barab et al. (2005) point out, approach is to reduce the efforts needed for design,

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

430 W. Westera et al.

development and exploitation. It is a response to the our distance learning institution, which inevitably intro-
inherent diversity and complexity of game design which duces some bias of teachers in favour of e-learning, rich
calls for applicable guidelines and methods for the media and gaming. The teachers appreciated its flexibil-
improvement of game design. The foundation of the ity and the large range of learning scenarios it caters for.
framework is theoretical in kind, but its development It must be said that in the case of sophisticated sce-
was strongly linked with the practice of educational narios, additional support remains essential. Especially,
game development, in particular, with the development the design of conversations and game logic are likely to
of the Emergo toolkit and methodology. So far, the become too complex for teachers, as was the case in the
Emergo methodology and toolkit has been used to domain of environmental policy. To this end, it can be
develop seven separate scenario-based games in various remarked that once a scenario is finalized, its structure
domains. These include environmental sciences, psy- becomes available for the community of teachers to
chology and education as part of the educational serve as a template for reuse with different content. With
programmes of the Radboud University, Utrecht Uni- respect to produced games themselves, teachers valued
versity, Maastricht University and the Open University their usefulness for efficiently preparing students for
of the Netherlands. Additional games are considered for handling real-life problem situations. Preliminary tests
management and law. The games involve complex with students indicate high appreciations, especially
problem scenarios which address higher order (aca- with regard to the game structure, the alignment with
demic and professional) skills. The size of the games is prior skills and, interestingly, the credibility of the game
up to 60 h of study load, while production efforts representation. The latter seems to conform to the prin-
including production of video and graphics design, have ciple that authentic representation is less critical than
been reported up to 1500 h. This yields a production content credibility because simple graphics dominated
ratio of 1 : 25, which is much lower than common stan- the games’ user interfaces. Additional research will be
dards for computer-based learning programmes (cf. necessary to establish the framework’s effectiveness
ranging from 1 : 100 towards 1 : 600; Alessi & Trollip and to evaluate the learning outcomes in connection
2001). The approach supports the instructional design with particular scenario types. This would also provide
process well. A brief evaluation among eight teachers the opportunity to link the work with different strate-
showed that they were well able to use the authoring gies, descriptions and tools in order to get closer to the
tools without any support. It must be added that almost ideal of a shared serious gaming paradigm.
all game development took place within the context of

Appendix

Actor Game object which may represent a non-playing character or a human game player
Event A state change of any game object
Game logic The collection of triggers and the underlying rules acting on the game world and its users
Game management Subsystem of the educational game environment where the different game runs are arranged and
world co-ordinated
Game object Any component of the world of game play which represents an artefact, actor or concept
Game play Context of operation defined by the rules of play and the physical and temporal boundaries of the
game
Learner world Subsystem of the educational game environment which contains the learner and its specific context
Location Game object which conforms to a spatial metaphor by representing a particular environment for
specific actions
Location builder Technical tool for defining game locations
Object builder A set of different technical tools for specifying game objects and the associated objects’ attributes
Option space The number of alternative actions that learners may choose at a particular location or point in time

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

Serious games design framework 431

Appendix Continued

Role builder Technical tool for specifying different player roles, associated with different tasks and different
powers
Scenario-based Game with some predefined narrative structure for the triggering of appropriate learning
game experiences and activities
Scenario builder Technical tool for constructing the logical conditions that make up the game play scenario
Serious game Game in which education (in its various forms) is the primary goal, rather than entertainment
(Michael & Chen 2006)
State logging Technical tool for the monitoring and registering of the state data of locations, objects and players
Subject Game object which may represent a non-playing character or a human game player
Teacher world Subsystem of the educational game environment which contains the teacher and its specific
context
World of game play Subsystem of the educational game environment where learners interact with game objects,
locations and other actors

Dickey M.D. (2005) Engaging by design: how engagement

References strategies in popular computer and video games can inform
Alessi S.M. & Trollip S.R. (2001) Multimedia for Learning: instructional design. Educational Technology Research and
Methods and Development. Allyn and Bacon, Needham, Development 53, 67–83.
MA. Fantuzzo J.W., Riggio R.E., Connelly S. & Dimeff L.A.
Anderson A., Cheyne W., Foot H., Howe C., Low J. & Tolmie (1989) Effects of reciprocal peer tutoring on academic
A. (2000) Computer support for peer-based methodology achievement and psychological adjustment: a component
tutorials. Journal of Computer Assisted Learning 16, analysis. Journal of Educational Psychology 81, 173–
41–53. 177.
Amory A. (2007) Game object model version II: a theoretical Galarneau L. (2005) Authentic Learning Experiences
framework for educational game development. Educa- through Play: Games, Simulations and the Construction
tional Technology Research and Development 55, 51–77. of Knowledge. Digital Games Research Association
Barab S.A., Thomas M.K., Dodge T., Carteaux B. & Tuzun H. (DiGRA), Vancouver, BC, Canada. Available at: http://lisa.
(2005) Making learning fun: Quest Atlantis, a game without socialstudygames.com/digra_galarneau_final.pdf (last
guns. Educational Technology Research and Development accessed 15 December 2007).
53, 86–107. Garris R., Ahlers R. & Driskell J.E. (2002) Games, motivation
Björk S. & Holopainen J. (2005) Patterns in Game Design. and learning. Simulation and gaming. An Interdisciplinary
Charles River Media, Hingham, MA. Journal of Theory, Practice and Research 33, 43–56.
Burgos D., Tattersall C. & Koper R. (2007) Re-purposing Gredler M.E. (1992) Designing and Evaluating Games and
existing generic games for E-learning. Computers in Simulations. Kogan Page, London.
Human Behavior 23, 2656–2667. Gredler M.E. (2004) Games and simulations and their relation-
Crawford C. (2005) Chris Crawford on Interactive ships to learning. In Handbook of Research on Educational
Storytelling. New Riders Publishing, Indianopolis, IN. Communications and Technology (ed. D.H. Jonassen),
De Freitas S. (2006) Learning in Immersive Worlds. Joint pp. 571–581. Lawrence Erlbaum Associates, Mahwah, NJ.
Information Systems Committee. Available at: http://www. Huizinga J. (1938/1955) Homo Ludens: A Study of the Play
jisc.ac.uk/media/documents/programmes/ Element in Culture. Beacon Press, Boston, MA.
elearninginnovation/gamingreport_v3.pdf (last accessed Hunicke R., LeBlanc M. & Zubek R. (2004) MDA: A Formal
15 December 2007). Approach to Game Design and Game Research. Game
De Freitas S. & Oliver M. (2006) How can exploratory learn- Design and Tuning Workshop at the Game Developers
ing with games and simulations within the curriculum be Conference, San Jose, CA. Available at: http://www.cs.
most effectively evaluated? Computers & Education 46, northwestern.edu/~hunicke/MDA.pdf (last accessed 15
249–264. December 2007).
Dewey J. (1938) Experience and Education. Macmillan Pub- IMS (2003) IMS Learning Design Information Model, Version
lishing, New York. 1.0 Final Specification. Available at: http://www.imsglobal.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd

432 W. Westera et al.

org/learningdesign/ldv1p0/imsld_infov1p0.html (last Quinn C.N. (2005) Engaging Learning. Designing

accessed 26 February 2008). E-Learning Simulation Games. Pfeiffer, John Wiley and
Kiili K. (2005) Digital game-based learning: towards an expe- Sons, San Francisco, CA.
riential gaming model. The Internet and Higher Education Reeves B. & Nass C. (1996) The Media Equation: How
8, 13–24. People Treat Computers, Television, and New Media Like
Kiili K. (2007) Foundation for problem-based gaming. British Real People and Places. Cambridge University Press,
Journal of Educational Technology 38, 394–404. Cambridge, UK.
Kolb D. (1984) Experiential Learning: Experience as the Rieber L.P. (2005) Multimedia learning in games, simula-
Source of Learning and Development. Prentice Hall, Engle- tions, and microworlds. In The Cambridge Handbook of
wood Cliffs, NJ. Multimedia Learning (ed. R.E. Mayer), pp. 549–567. Cam-
Koumi J. (2006) Designing Video and Multimedia for Open bridge University Press, Cambridge, UK.
and Flexible Learning. Routledge, New York. van Rosmalen P., Brouns F., Tattersall C., Vogten H., Van
Lindley C.A. (2005) Story and narrative structures in com- Bruggen J., Sloep P. & Koper R. (2005) Towards an open
puter games. In Developing Interactive Narrative Content, framework for adaptive, agent-supported E-learning. Inter-
Sagas/Sagasnet Reader (ed. B. Bushoff), pp. 218–245. national Journal of Continuing Engineering Education and
High Text, Munich, Germany. Lifelong Learning 15, 261–275.
Littlejohn S.W. (1983) Theories of Human Communication. Salen K. & Zimmerman E. (2004) Rules of Play, Game Design
Wadsworth Publishing, Belmont, CA. Fundamentals. MIT Press, Cambridge, MA.
Ma Y., Williams D., Prejean L. & Richard C. (2007) A Squire, K. (2002) Cultural framing of computer/video games.
research agenda for developing and implementing educa- Game Studies 2. Available at: http://www.gamestudies.org/
tional computer games. British Journal of Educational 0102/squire/ (last accessed 15 December 2007).
Technology 38, 513–518. Steinkuehler C.A. (2004) Learning in massively multiplayer
Michael D. & Chen S. (2006) Serious Games: Games That online games. Proceedings of the 6th International Confer-
Educate, Train and Inform. Thomson Course Technology, ence on Learning Sciences, pp. 521–528. Lawrence
Boston, MA. Erlbaum, Mahweh, NJ.
Montola M. (2005) Exploring the edge of the magic circle. Westera W. (1999) A didactic framework for audiovisual
Defining pervasive games. Proceedings of DAC 2005 design. Journal of Educational Media 24, 87–102.
Conference. Available at: http://users.tkk.fi/~mmontola/ Westera W., Sloep P.B. & Gerrissen J. (2000) The design of
exploringtheedge.pdf (last accessed 15 December 2007). the virtual company; synergism of learning and working in
Murray J. (1997) Hamlet on the Holodeck: The Future of Nar- a networked environment. Innovations in Education and
rative in Cyberspace. MIT Press, Cambridge, MA. Training International 37, 23–33.
Nadolski R., Hummel H., van den Brink H., Hoefakker R., Westera W., Hommes M.A., Houtmans M. & Kurvers H.J.
Slootmaker A., Kurvers H. & Storm J. (2007) Emergo: (2003) Computer-supported training of psychodiagnostic
methodology and toolkit for efficient development of skills. Interactive Learning Environments 11, 215–231.
serious games in higher education. Available at: http:// Wong W.K., Chan T.W., Chou C.Y., Heh J.S. & Tung S.H.
dspace.ou.nl/handle/1820/1046 (last accessed 5 March (2003) Reciprocal tutoring using cognitive tools. Journal of
2008). Computer Assisted Learning 19, 416–428.
Prensky M. (2006) Digital natives. Learning in the Digital
Age 63, 8–13.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd