ADAPTIVE GAME MECHANICS FOR LEARNING PURPOSES

Making Serious Games Playable and Fun

Jonathan Tremblay, Bruno Bouchard and Abdenour Bouzouane

LIARA laboratory

Université du Québec à Chicoutimi, (Québec)(Rolling and Adams, 2003), Canada

555 boul. Université, G7H 2B1

Keywords: Game design, Adaptive game mechanics, Serious game, Challenge, Game studies, Gameplay, Flow,

Adaptive difficulty.

Abstract: This paper investigates adaptive games mechanics and how to implement them. First, a comprehensive

review of existing adaptive models is presented. Next, we propose a new adaptive model, which combines

dynamic difficulty adaptation, the player’s performance, and adaptive flow. An implementation of these

new adaptive mechanics is presented in the form of a simple serious game called Number to Number

Combat. This game was released freely on the internet in order to be tested by the gaming community. It has

shown very promising results that will help us to improve our adaptive model.

1 INTRODUCTION

The video game industry as entertainment provides

the player with many different game genres. A

player can choose from puzzles to first person

shooters (FPS), casual to hardcore, Play Station to

PC, etc. These choices are meaningful for the player

(Salen and Zimmerman, 2003), they give the player

choices about his game experience based on his own

abilities and aspirations. Therefore, they enhance the

player’s immersion in the game and contribute to his

desire to play.

In the last few years, video games have also

begun to be more than just fun. An increasing

amount of scientists (Jiang et al., 2006) and

companies are exploring how games can be used as

teaching tools (Gee, 2003) or as a rehabilitation

platform (Jiang et al., 2006) for cognitively impaired

patients, such as Alzheimer’s patients. For instance,

one may note the recently published game “Brain

Age” on Nintendo DS for which the idea is to

provide a series of small simple puzzle games

making it possible to play a few minutes per day in

the aim of “improving brain performance”.

This new way of using video games brings a lot

of emergent challenges. One of the most important

challenges corresponds to designing a game in

which the player can make meaningful choices in a

serious context. The fact is that currently, in the

industry, players cannot make very many

meaningful choices in serious games, even though

the diversity of types of players in serious gaming is

greater than in the entertainment industry.

This diversity of players introduces different skill

levels. Serious games lack the luxury of a vast

budget. Despite this lack, games should serve every

kind of player and every kind of skill. The process of

design for serious games needs to adapt the

difficulty level of the game to different sets of skills.

In general, the entertainment industry provides

different difficulty levels for games: casual, normal,

hard (Gilleade and Dix, 2004). This design process

does not take into consideration the game semantic

mastered by the players or the player who has

difficulty where they are not supposed to. The

designer has to understand every kind of player

playing the game and make sure that the game is

enjoyable for all sorts of players.

A designer should concentrate his efforts on the

player’s experience, instead of trying to anticipate

the player’s skill level. A serious game should let

every kind of player play the game. Casual to

hardcore players should learn from the game and

enjoy the game. The player’s experience is described

as an optimal flow (Chen, 2006), which can be

defined as a state of total immersion where the

challenge should match the player’s abilities

perfectly. The player’s enjoyment is closely linked

to the appropriate level of difficulty. If the challenge

465

Tremblay J., Bouchard B. and Bouzouane A. (2010).

ADAPTIVE GAME MECHANICS FOR LEARNING PURPOSES - Making Serious Games Playable and Fun.

In Proceedings of the 2nd International Conference on Computer Supported Education , pages 465-470

DOI: 10.5220/0002855604650470

 SciTePress

is too hard, the player will suffer anxiety. On the

other hand, if the challenge is too easy, the player

will experience boredom. As Juul (Juul, 2009) points

out, the player needs to experience failure and

difficulty in order to enjoy the game. A game where

the player is winning all the time is no fun and the

opposite is also not enjoyable. When flow is

experienced, the player feels control over the game,

they are mastering it. Mastering the semantics of a

serious game will lead to mastering the subject of

the game (Gee, 2003). Would it be possible for a

serious game to allow players to experience flow? If

so, would it be possible for serious games to adapt

the difficulty of the challenges to the player’s skill

level?

This research investigates the possible ways for

games to adapt to the player’s skills and how to

implement this adaptation. We found four

mainstream adaptive game mechanics: Dynamic

Difficulty Adaptation (DDA) (Hunicke, 2005),

adaptive flow (Chen, 2006), Game Play Schemas

(Lindley and Sennersten, 2008) and using frustration

(Gilleade and Dix, 2004). After having reviewed

each of them, we developed a new adaptive model

which combines feedback (Salen and Zimmerman,

2003) based on DDA, the player’s performance, and

includes adaptive flow. We implemented these new

mechanics into a simple serious game called Number

to Number Combat, which was released freely on the

internet in order to be tested by the gaming

community. This game is made so that a frequent

player will be challenged more than a casual one.

The results obtained after a first testing phase are

encouraging and will help us to improve the

adaptive model.

The rest of this document is organized as follows:

Section 2 discusses previous works related to this

research. Section 3 describes our approach to

designing our game, our implementation and some

early results. Section 4 summarizes our conclusions

and presents possible future work.

2 RELATED WORK

The level of difficulty in a game is created linearly

by a designer. The design process depends upon play

testing, so that the designer can understand the

difficulty and tweak the game for a particular kind of

player (Chen, 2006). The designer needs to repeat

this step until the game is balanced. This is even

more time consuming when catering to every kind of

player (casual, normal, hardcore, etc.). In reality,

when developing a serious game with a low budget,

the designer does not have all the time he/she needs

to tweak the game perfectly. Introducing adaptive

game mechanics makes the game more accessible

and enjoyable for the player. It makes the game

more challenging for any kind of player, therefore

more enjoyable and playable for the player (Juul,

2009). Adaptive game mechanics also require

tweaking (Hunicke, 2005). In the last few years,

researchers (Hunicke and Chapman, 2004, Chen,

2006, Lindley and Sennersten, 2008, Gilleade and

Dix, 2004) have explored different avenues to

implement this kind of adaptive mechanics. These

sources explain the player’s experience using flow

theory. We can distinguish four proposed

approaches: Dynamic Difficulty Adaptation,

Adaptive Flow, Game Play Schemas and using

frustration.

2.1 Dynamic Difficulty Adaptation

Dynamic Difficulty Adaptation (DDA) offers

alternative-modulating in-game systems to respond

to a particular player’s abilities over the course of a

game session. DDA is based on the mathematical

analysis of structures and relationships within a

game system (Hunicke, 2005) and on the player’s

flow experience. DDA uses the flow principle in

order to keep the game intriguing and enjoyable.

With the right structure, everything from narrative

structure to the game menu can possible adjusted

(Mateas, 2002). It is very important to completely

understand the design and how the system could

interact with the game in order to challenge the

player.

DDA uses a system that changes the game

mechanics without the player knowing it. These

changes are made in order to keep the player

challenged and interested (Hunicke and Chapman).

First, the system computes the player’s data;

player’s position, player’s health, player’s ammo,

etc. Following the system assessment, the system

chooses the data that reflects the player’s state of

flow. The system analyses the player’s state of flow

and notifies the game of any changes. Lastly, the

game apply the changes (Chen, 2006).

For instance when the player is playing a first

person shooter (FPS), the system could notice if

he/she has low health. The game could be too

difficult for the player’s skills. The system then

could decide to make a health pack available to the

player. An important element would be to ensure

that the player does not know about systems such as

the DDA (Hunicke, 2005).

CSEDU 2010 - 2nd International Conference on Computer Supported Education

466

The system analyses the player’s data based on

the player’s flow experience. However, one of the

major problems with DDA is that the system bases

its decisions on the player’s flow state using only

raw data. The raw data used represents the

performance of the player, which is objective, while

flow is subjective (Chen, 2006) . On the other hand,

DDA is straight forward to implement and

understand (Hunicke and Chapman, 2004).

2.2 Adaptive Flow

Chen (Chen, 2006) introduces flow as a design

process. Based on the assumption that the player’s

flow experience is subjective, Chen (Chen, 2006)

proposes giving the player control. Control is a

requirement for the flow experience, the player must

feel in control over his/her actions in order to

experience flow (Salen and Zimmerman, 2003). The

sense of control comes from the sense of progression

and positive feedback (Chen, 2006). In the design of

the game, the player should control the level of

difficulty. Figure 1 shows how a player can make

choices that can result in changing his flow

experience.

Figure 1: Adaptive flow based on player's choices.

In order to design such a game, the designer needs to

include a wide spectrum of game mechanics for a

variety of levels of difficulty and tastes. The game

should provide a player-oriented active DDA to

allow different players to play at their own pace.

This system must be embedded in the game core

mechanics and let the player make their own choices

through play (Chen, 2006). For instance, the player

could experience an intense challenge or choose to

explore and power up his avatar which will attenuate

the challenge’s intensity.

Designing a game with embedded meaningful

choices regarding the difficulty of play is not a

simple task. The designer has to think of the game as

a DDA platform from the beginning. Should the

game already exist, it is almost impossible to include

adaptive flow after the fact.

2.3 Game Play Schemas

Lindley and Sennersten (Lindley and Sennersten,

2008) introduce game mechanics based on schema

theory. A schema is a semantic representation of

knowledge integrated into the decision process.

Schema applied to game mechanics becomes an

algorithm representation of the semantic knowledge

needed to perform an action within the game.

Therefore, the player’s actions can be reproduced as

an algorithm.

Example of a schema game mechanics algorithm for a

typical adventure game:

if(player.sick)

player.getHeal();

else

player.Attack();

This algorithm is typical in adventure, before

attacking we check if the avatar is sick. Based on the

data, the player’s action consists of either attacking

or healing. Using this algorithm a system could find

where a player makes mistakes during play. In this

case, if the player never heals his avatar; this could

indicate that the player does not understand the

meaning of being sick. With this mistake found, the

system could intercept in order to help the player.

This help could be meaningful for a casual player,

such as Alzheimer’s patient.

In other situations, a schema game play could be

used to adapt the game to the player’s style (Lindley

and Sennersten, 2008). The system, by determining

what type of player is playing, can introduce

elements that the player enjoys based on the player’s

repetitive actions such as how the player is defeating

None Player Character (NPC), how the player is

driving his car, how the player interacts with the

menu, etc.

A major issue with the game play schemas’

model is that it has never been implemented or

tested (Lindley and Sennersten, 2008). Moreover, it

seems that it would be very time consuming to

implement. In order to understand the player’s

interaction with the game, the designer would need a

lot of case studies and would have to calculate the

possible mistakes a player may make.

2.4 Using Frustration

Frustration in a game is something that every player

has experienced at least once. Frustration arises

ADAPTIVE GAME MECHANICS FOR LEARNING PURPOSES - Making Serious Games Playable and Fun

467

when the in-game progress towards achievement is

impeded (Gilleade and Dix, 2004). When the player

is unable to complete a command, he/she becomes

frustrated as he/she is not able to progress.

Furthermore, the player becomes frustrated when

he/she cannot complete a certain challenge in the

game due to a misunderstanding of the game

challenge.

Using DDA, when the player is getting

frustrated, the system could change the game

difficulty or provide help to the player. For example,

the game system could change the width of a hole

the player had fallen into so that their next attempt at

jumping over the hole would be successful and the

player would be less frustrated.

The major issue with this model is the detection

of frustration. Frustration can be measured using

blood pressure, heart rate and conductivity (Gilleade

and Dix, 2004). These measurements are related to

the level of arousal. From a commercial point of

view, using these measurements is almost

impossible; the only existing connection between the

player and the game is the gamepad. Some research

(Sykes and Brown, 2003) indicates that it is possible

to measure the player’s level of arousal by

monitoring button pressure on the gamepad.

However, this idea has never really been officially

used as an adaptive mechanic.

3 IMPLEMENTATION AND

RESULTS

We have developed a digital learning game

prototype called Number to Number Combat to use

as an experimental test for our research in adaptive

game mechanics. A screenshot of the game is shown

in Figure 2. Number to Number Combat is a game

that is designed to teach and master basic arithmetic;

addition and subtraction.

Figure 2: Screenshot of Number to Number Combat game.

The current version of Number to Number

Combat is simple; there is no end to the game. The

player is playing the left avatar, by answering

correctly the equation; the avatar will hit his

opponent and inflict damage. On the other side, the

NPC will randomly hit the player. The goal is to

defeat the NPC before it defeats you. When the

player wins, he/she has to build up his/her avatar’s

force, defence and luck. Force gives the player more

powerful hits. Defence protects the player from his

opponent. Luck increases the player’s chances of

hitting a weak spot.

3.1 Adaptive Game Mechanics

In Number to Number Combat, we propose a new

adaptive model which combines feedback (Salen

and Zimmerman, 2003) based on DDA as well as

the player’s performance, and includes adaptive

flow. The DDA system is based on the player’s

health. As represented in Figure 3, when the player

finishes a combat with a low health bar, the next

opponent will be easier to vanquish. On the other

hand, when the player is mastering a combat and

his/her health bar is full, the next opponent will be a

little harder. The system works without the player

noticing. The way the system changes the NPC

difficulty is by feedback. The system uses negative

feedback, that consists of reducing the gap between

two related elements (Salen and Zimmerman, 2003).

In other words, if the challenge is too hard for the

player, the negative feedback will reduce the

difficulty of the gap.

Figure 3: DDA using negative feedback in Number to

Number Combat.

After a fight, the player must choose what he will

put the 2 points he has earned towards. This is

adaptive flow at work. This part maps the kind of

fight the player will encounter. If a player adds a lot

of points to their Strength, the fights will need a fast

paced answer entry. On the other hand, if the player

adds a lot of points towards his Defence, the fights

CSEDU 2010 - 2nd International Conference on Computer Supported Education

468

will be slower, and both the player and the NPC will

be well protected against attacks. Finally, if the

player adds a lot of Luck, the fights will be

completely random.

3.2 Validation and Results

We proceeded to a first validation phase for our new

adaptive approach by releasing Number to Number

Combat as Freeware on the website

www.newgrounds.com. Newgrounds provides a free

online video game platform for independent

developers. There are about 2 million registered

members and 500 000 flash submissions (Fulp,

2010). Players can comment on the game and rank

it. Since the game came out, it has been played by

over a thousand players. The results obtained so far

are very promising, knowing that, even if this

serious game is very basic and simple, the players’

average review is 8.2/10 and the game received very

positive comments. One of the aspects of the game

most appreciated by the players is that the game

improves theirs math skills with basic arithmetic in

an enjoyable combat role playing game world. We

believe that the new adaptive mechanics included in

the game contributed to the great appreciation from

the players.

Conversely, some players were disappointed that

the game does not get harder after a certain time.

Also, the aesthetic aspect of the game does not seem

to be appealing enough for the players and the fights

are too repetitive and do not offer enough sense of

control over the avatar. Finally, the role playing

elements do not seem to make a difference in the

game experience. Implementing a more complex

game system with more attractive graphics and more

controls will help us to alleviate these concerns and

to conduct enriching further tests on the adaptive

model.

4 CONCLUSIONS AND FUTURE

WORK

In this paper, we investigated adaptive mechanics in

games and how to implement them. This has been

achieved by reviewing existing adaptive models, by

proposing a new adaptive approach combining

dynamic difficulty adaptation, the player’s

performance, and adaptive flow, and by presenting

an implementation of this new adaptive mechanic in

the form of a simple serious game called Number to

Number Combat. We also presented the promising

results of a priliminary validation phase conducted

by releasing the game freely on the Internet to be

tested by the gaming community.

Every game teaches something about a system

(Koster, 2004, Gee, 2003). The preliminary

comments gathered by releasing Number to Number

Combat show promise, but the game is not perfect

and has a couple of design flaws. One important

element that we bring to light in this paper is that

serious games need to be treated like any

entertainment game; the game has to be appealing

for the player. Therefore, we plan on working on a

narrative framework for the game, redoing the

graphics and getting the game balanced.

Number to Number Combat was a first step in

the LIARA laboratory new project, which aims to

give way to video games as a new software platform

allowing the support of medical and learning tools,

less expensive and more accessible, that will be

used, for instance, for palliative care for those

suffering from Alzheimer’s disease. Using this kind

of serious game will be enjoyable, fun, and will

make a real difference in the life of Alzheimer’s

patients by slowing the degenerative process of their

disease, thus contributing to giving them a better

quality of life (Tárraga et al., 2006). We plan on

developing prototype games for Alzheimer’s

patients, for people suffering from head traumas and

for people with other cognitive impairments. The

Alzheimer’s patients are casual gamers, we need to

adapt our design process to make sure the game they

play is accessible, fun, and that it answers their

needs.

The next logical step in this project consists of

implementing our adaptive mechanics in a more

complex game system and testing it. We also plan

on exploring different adaptive game mechanics

such as game play schemas (Lindley and Sennersten,

2008). Implementing these models could help the

design process for serious games. Finally, we

recently signed a collaborative agreement with the

rehabilitation center of La Baie, which is an

institution that treats Alzheimer’s patients. This will

allow us to test our prototype on the targeted

audience.

ACKNOWLEDGEMENTS

We would like to thank our main sponsors for their

financial support: the Natural Sciences and

Engineering Research Council of Canada (NSERC),

the Quebec Research Fund on Nature and

Technologies, and the Université du Québec à

Chicoutimi (UQAC). We also want to thank the

Cleophas-Claveau center for their future help in

testing our ideas. Finally we would like to thank all

the players who played Number to Number Combat

and commented on it.

ADAPTIVE GAME MECHANICS FOR LEARNING PURPOSES - Making Serious Games Playable and Fun

469

REFERENCES

Chen, J. 2006. Flow in games. MFA in Interactive Media,

University of Southern California.

Fulp, T. 2010. Main Page [Online]. Available:

http://newgrounds.wikia.com/wiki/Main_Page.

Gee, J. P. 2003. What video games have to teach us about

learning and literacy. Comput. Entertain., 1, 20-20.

Gilleade, K. M. & Dix, A. 2004. Using frustration in the

design of adaptive videogames. Proceedings of the

2004 ACM SIGCHI International Conference on

Advances in computer entertainment technology.

Singapore: ACM.

Hunicke, R. 2005. The case for dynamic difficulty

adjustment in games. Proceedings of the 2005 ACM

SIGCHI International Conference on Advances in

computer entertainment technology. Valencia, Spain:

ACM.

Hunicke, R. & Chapman, V. 2004. AI for Dynamic

Difficulty Adjustment in Games.

Jiang, C.-F., Chen, D.-K., LI, Y.-S. & KUO, J.-L. 2006.

Development of a Computer-Aided Tool for

Evaluation and Training in 3D Spatial Cognitive

Function. Proceedings of the 19th IEEE Symposium

on Computer-Based Medical Systems. IEEE Computer

Society.

Juul, J. 2009. Fear of Failing? The Many Meanings of

Difficulty in Video Games. In: PERRON, M. J. P. W.

B. (ed.) The Video Game Theory Reader 2. New York:

Routledge

Koster, R. 2004. A Theory of Fun for Game Design,

Paraglyph Press.

Lindley, C. A. & Sennersten, C. C. 2008. Game play

schemas: from player analysis to adaptive game

mechanics. Int. J. Comput. Games Technol., 1-7.

Mateas, M. 2002. Interactive drama, art and artificial

intelligence. Carnegie Mellon University.

Rolling, A. & Adams, E. 2003. Adrew Rollings and Ernest

Adams on Game Design, New Riders.

Salen, K. & Zimmerman, E. 2003. Rules of Play: Game

Design Fundamentals, The MIT Press.

Sykes, J. & Brown, S. 2003. Affective gaming: measuring

emotion through the gamepad. CHI '03 extended

abstracts on Human factors in computing systems. Ft.

Lauderdale, Florida, USA: ACM.

Tárraga, L., Boada, M., Modinos, G., Espinosa, A., Diego,

S., Morera, A., Guitart, M., Balcells, J., López, O. L.

& Becker, J. T. 2006. A randomised pilot study to

assess the efficacy of an interactive, multimedia tool

of cognitive stimulation in Alzheimer’s disease. J.

Neurol.

CSEDU 2010 - 2nd International Conference on Computer Supported Education

470