Petri Lankoski, Staffan Björk
ACM, (2011). This is authors version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version was published in MindTrek’11 (Tampere ) http://dx.doi.org/10.1145/2181037.2181041.
Gameplay design patterns are semiformal interconnected descriptions features of gameplay. While most previous patterns have been identified through analyzing existing games, this paper proposed how patterns can be identified using theories as starting points. More specifically, we propose three different approaches to harvesting gameplay design patterns: 1) using theories as analysis foci, 2) distilling patterns from theories, and 3) using theories to understand the consequences of having or not having patterns present in a game design. The three approaches are presented together with examples of their use, and based upon this the concept of Theory Lenses as an analytical tool is introduced as a way of allowing theories independent of their research field to be applied to research on gameplay design.
Categories and Subject Descriptors
K.8.0 [General]: Games
Gameplay design patterns, gameplay design, design patterns, theory lenses, game research
Björk and Holopainen  introduced gameplay design patterns as a tool for describing gameplay aspects of game designs, and to help design games. The patterns introduced were primarily based on structural analysis of existing games to isolate reoccurring design features which were then described in a with semiformal notation. The central premise of the method was to give names to specific design possibilities in order to make them into perceivable concepts, and to provide structures for how the concepts relate to each other. But, how one can ensure that the patterns are of use? Are there methods for identifying patterns that arguable make them more likely to be interesting before they have been put to actual use?
We argue below that one way to increase the relevance of patterns to game design are to develop them in relation to specific scientific theories whose validity have already been motivated elsewhere. This approach, which we call Theory Lenses, can be applied to theories from any scientific discipline as long as they are based upon providing ontological models. The structure of this essay is as follows. First, we describe gameplay design pattern approach and its roots. After that, we present the Theory Lenses in detail and discuss the advantages and disadvantages of different Theory Lenses use. Last, in the Concluding Discussion, we summarize the benefits of using Theory Lenses in harvesting gameplay design patterns.
The idea of design patterns as concept for naming and describing design possibilities as well as how they relate to each other was first introduced within architecture . The idea and the original pattern collection was created by a group of professionals, lead by Christopher Alexander, with many years of practical experience. Their intention was to make explicit the design knowledge that had been built up over millennia and had in various, more or less implicit or structured, ways been transmitted between practitioners. Their intention in making this tacit knowledge explicit was so that the people who would live or work in buildings, and were not likely to be architects in most cases, could get access to a language that could help them describe the needs and desires of their environment. This would hopefully lead to designs that would more fit the actual functional requirements by empowering the people that would inhabit the spaces by letting them more efficiently communicate with architects.
Patterns have not gained popularity among architect professionals outside Alexander’s own projects but continue to be of interest to academics and hobbyists based on the continued availability of the original patterns books. In contrast, the use of patterns has become a well-established part of practices within software engineering after the idea was transplanted to that setting . It is not self-evident that patterns were adopted in one field and not another due to any one single reason, for example, having a different perspective on design or the age of the discipline. However, it is apparent that programmers perceived that they had a need that design patterns fulfilled, which quite likely related to being able to express possible design solutions and communicate this within groups of developers.
Both game designers and game researchers have voiced the same need for having a language when considering gameplay, for instance, Church , Spector , Costikyan , Falstein , and Zagal et al. —and in many cases also provided potential solutions. Björk and Holopainen  introduce design patterns to game research for the same reason. However, a challenge with game research is that it can readily be approached many different research disciplines; Björk  has categorize three groups of approaches based upon whether the research focus is upon gamers, gaming, or games. While this may seem to indicate that design patterns would only be applicable to those basing their research on specific disciplines or research interests, for example, designing games, a counter argument to this is that all type of research on gameplay is benefited by having access to a vocabulary with a high level of granularity. This could for example help describe the nuances of gamer preferences, label activities during gameplay with higher precision, and more succinctly state hypotheses of cause and effect chains in game designs. Therefore, the gameplay design pattern approach can be seen as theory neutral in that it can be used together with any type of theory that can be applied to the game phenomena.
Before continuing to look at Theory Lenses, we describe the gameplay design pattern format introduced by Björk and Holopainen and describe their method of harvesting patterns briefly. Their pattern format is as follows:
- name: a pattern having a short descriptive name;
- core definition is a brief sentence providing an overview of the pattern;
- description is a short general description of the properties of the pattern (including possible examples);
- using the pattern introduces description of the common choices that the designer faces when using the pattern;
- consequences discusses about the gameplay implications of using the pattern;
- relations accounts for the typical relations between patterns (a pattern can be instantiated, modulated, or conflicting another pattern);
- the references section points to earlier work where the pattern has been derived. 
The original pattern collection by Björk and Holopainen  was harvested mainly by analyzing games. Their main method was analyzing and comparing a wide range of games starting from concrete game elements and rules and from there inferring gameplay, but they also derived patterns by interviewing game designers. They also describe how the play testing of existing games was used to find patterns.  Although these methods were useful in casting a wide net to explore gameplay generally, it was also undirected in a sense that the process was not directed toward supporting a specific analysis or design process. Below, we introduce an alternative method of harvesting gameplay design patterns with a more precise focus: Theory Lenses.
3. THEORY LENSES
During research on gameplay design patterns, we have developed a method, Theory Lenses. The method aims to address the above-mentioned problem of judging the quality and usefulness of a pattern. In Theory Lenses, theories from various other scientific disciplines are used as a starting point for analysis when harvesting patterns from existing games. The choices of disciplines have been based primarily on the specific focus within gameplay that is explored rather than through selecting gameplay focus so it fits the specific same discipline each time. In previous studies, researchers have combined different theories when harvesting patterns in various ways . This theory use can be categorized as follows (we return to the categorization below in more detail):
- the scientific models and theories provide focus for the structural analysis of the game when harvesting patterns;
- a pattern can be based directly on theory without analyzing any game;
- the consequences of using a pattern can be derived from theories.
Not apparent from the format of patterns, they provide descriptions in two different ways. On the one hand, patterns describe the formal structure of games. The specific formal structure describe in one pattern often relate in some way to an everyday phenomena, but this may not be a structure that is meant to realistically model the phenomena. For example, Tetris  uses the idea of gravity as part of its gameplay, yet it hardly a model of the real world phenomena of gravity. On the other hand, the consequences section of the pattern refers to consequences on the level of gameplay experience.
These two ways echo two types of theories: those that are explanatory and those that are predictive . While explanatory theories can be understood as those that clearly describe a phenomena, Niiniluoto describes the main feature of predictive theory as being “able to produce reliable predictions about future events” and have a “predictive power as their central epistemic utility.” . Niiniluoto explains that practical astronomy, meteorology, and social statistics are examples of sciences where theories are typically predicative. He also stresses that explanatory theories do not have this kind of predictive power while they provide detailed models of a phenomenon, for example, that the theory of evolution can explain why species are evolving but not predict how they will evolve. 
Predicting the consequences of using a pattern, the category three above, quite naturally requires a predictive theory, but the other two categories can make use of both types.
Next, we look at the three categories that Theory Lenses can be used in more detail based upon specific uses. We aim to generalize principles of the above theory use, as well as discuss the strengths and weaknesses of these.
3.1 Using a Theory as an Analytical Focus
Existing theoretical models can function as a lens that narrows one’s attention towards specific aspects of games, thereby letting one only see gameplay patterns within that aspect. These can be used separately to support the identification of patterns in compliance with specific theories or in conjunction to allow the Mixed Method approach. Denscombe writes
“Researchers can improve their confidence in the accuracy of findings through the use of different methods to investigate the same. […] [T]he Mixed Methods approach provides the researcher with the opportunity to check the findings from one method against the findings from a different method.” 
This kind of Mixed Theory approach can give new insights for gameplay design. For instance, when studying how to design believable NPCs Lankoski and Björk  used theories from psychology, the philosophy of mind, and film studies to derive patterns by analyzing breakdowns in the NPC design of a game. This provided the basis for finding both presentational and behavioral patterns, for example, VISUAL BODY DAMAGE¹ (that damage to a character is represented through changes in the appearance of a character) and AWARENESS OF SURROUNDINGS (that NPCs seems to be able to detect phenomena that are relevant to it).
In another study, Lankoski and Björk  used Social Network Analysis  and Actor Network Theory (ANT)  to identify the presence or absence of concepts, and proposed patterns FACTION (a specific social network where membership is defined by what actions are allowed, disallowed, and required), and INTERNAL CONFLICT (reaching a goal makes reaching other goals more difficult or impossible). 
It should be noted that not only fully established scientific theories can function as these kinds of foci; for example, the classification of conversational systems were the basis for how NPC conversations can be part of gameplay  while the Mechanic-Dynamic-Aesthetics model of LeBlanc  drove Bergström et al  to explore aesthetical design patterns.
Common for all these cases is that the theories (or models) acted as a template which the games could be compared against. When a game did not comply with the template, the lack of functionality or structure was the basis for one or more potential design patterns. In this way, the Theory Lenses approach made it possible to in a structured way focus on the absence of features in games rather than that which was present. It should be noted that the games might also provide new insights into that which it is matched against, for example, looking at how games implemented conversations reveals a new (game state-based) category of dialogue systems .
This approach finds patterns based upon the lack of features or issues in terms of theory in existing gameplay design. Proponents wishing to have games have certain characteristics or behaviors can use this to find patterns that can make these appear.² This is not simply making the components of a theory into patterns but rather taking one step closer to concrete design elements; identifying the game parts that through their interrelations and affect on gameplay can give rise to the theory components. In addition, it does point to possible ways game designers could make games more detailed, or even base the core gameplay of a game around aspects not used previously. A specific case of this is to point to how one could apply existing technologies, currently not used in games, to provide new gameplay features.
3.2 Deriving Patterns from a Theory
We introduced above the idea of using a theory to focus on a specific aspect of gameplay. This requires that one is looking at a specific game. However, a theory can be used without looking at any specific game to suggest possible gameplay. While the approach in Lankoski and Björk  was mainly tied to the structural analysis of games, we also used the ANT as a source for gameplay design patterns. That study focused on character–character relations and, interestingly, the ANT proposes that social groups are not stable: “if you stop making and remaking groups, you stop having groups” . Based on this, Lankoski and Björk propose that “goals in a game can be created from the view that groups are not stable entities in the ANT, but rather something that constantly needs maintenance”, and formulated the pattern SOCIAL MAINTENANCE³.
The source of the pattern in the above case was in the theory rather than in any specific game analysis. Although this pattern could have been deduced through the approach describe in the previous section, it shows that a theory in itself can be enough.
Basing patterns purely on theories, the pattern models the theory domain in some degree. How well the gameplay based on the pattern mirrors the modeled domain is an important issue in some areas (for example, games for education), but another equally critical issue is if the use of pattern leads to interesting gameplay. The used theory can provide foundations for designing simulation which can function as the basis for a game. When theory is used in this way, the theory is considered as a model of a phenomenon in the target domain. However, there is no connection between scientific quality (how well theory seems to model the target domain, such as human social interaction) and quality of and gameplay (and design patterns) derived from the theory. For example, a model can lead to a pattern REGULAR WASTE DISPOSAL NEED (for example, that a character needs to use toilets regularly and failing to do so leads to embarrassing situations) that can (in many cases) lead to micromanagement that is repetitive and not supporting the focus of the game. This said, the Sims  uses the pattern successfully. Hence, one must be careful to choose a theory and patterns that can be used to support the main creative idea of a game.
This approach of harvesting patterns has two caveats to avoid: First, theories can lead to different, and sometimes incompatible, design conclusions regardless if they come from the same discipline or not. Second, even if the theories correctly predict consequences in the real world this may not apply to the models contained within a game. Thus, the suggested consequences need to be verified from actual game designs besides whatever evidence exists from the discipline of a theory. Since many theories are incompatible with each other, that also points to the impossibility of creating one single collection incorporating the knowledge from all research disciplines; it may be more feasible to have many different patterns collections based upon different theories, and choose between these based upon the specific requirements of whatever activity one is undertaking.
3.3 Deriving Implications of a Pattern from a Theory
As mentioned above, the gameplay design pattern format includes description of the consequences of using the pattern . These describe cognitive or emotional consequences, basing the implications for a mixture of common-sensical reasoning, personal experience from having vulgar competence , and analogies from scientific domains4. Although this allowed a large collection of patterns to be identified with limited resources, common-sensical conclusions are prone to different biases: for example, generalizing from limited evidence (such as using only supporting cases) or from failed introspection (see, for example, Glover , Kunda , and Noë ).
The theory lenses use in harvesting pattern can alleviate this issue. For example, when Lankoski and Björk explored the believability of NPCs  the theory of intentional stance (from the philosophy of mind, see ) and narrative comprehension (from cognitive psychology, see ) explained why design solutions for NPCs can initially evoke certain responses from players but later fail to do so; an example of how the consequences of a pattern may differ with player experience of a specific game. The intentional stance theory was not used as analysis foci, but as a theory to revealing some implications of the lack of the believability patterns; instead it postulated that when the believability patterns were missing, the NPCs would not be understood as an intentional agent, but rather as a simple machine or “law of nature”. On the other hand, the theories used in the study (international stance, narrative comprehension, and character recognition based on the person schema) could also motivate to reveal when characters would be more likely to be perceived as believable intentional agents.
Theories can also be used to revisit and revise patterns. Writing about the difficulty of knowing in advance if a gameplay design goal is reached or not, Björk and Holopainen write that
“player balance and emotional responses, are impossible to predict without play testing” .
Nonetheless, this statement should not be taken to mean that it is, in general, impossible to make the assumptions of likely outcomes without play testing. Lankoski argues that cognitive emotions theories can trace the affective implications of game systems , and this can be extended to make the point that theories can strengthen the predictability of a pattern’s consequences. In both cases, the consequences can be seen as hypothesis forming, which can be evaluated empirically, but with a theoretical underpinning the pattern can make use of previous scientific knowledge to improve its reliability5. Based upon this, Lankoski links theories about planning and cognition by Oatley  and Power and Dalgleish  to gameplay goals and emotions. For example, in Tetris , a player needs to create a new goal (where to place the new tile). The planning and cognition theories imply that emotions relate to goal-status evaluation: players are likely to feel happiness if they successfully guide the tile in the desired place, sadness if they fail, and fear when the speed threatens the player ability to manipulate the block in the desired way. This sequence is repeated for every block, until the final failure. The game session, then, is evaluated as success or failure depending on the overall goal (for example, new high score) of the player.
It is possible to describe implications of an event structure of Tetris in the above-described way. However, is it possible to generalize results from this kind of analysis? Björk and Holopainen  describe pattern TENSION (and it is typically used to describe a consequence of use of another pattern).6 They write:
“Tension occurs in games that uncertain outcomes, when players have emotional investment in which of these outcomes occurs, but cannot fully control them.” 
Combining Björk and Holopainen’s description and the above-mentioned theory linking goals and emotions, one can rewrite the pattern TENSION as a fear-based emotional state that is triggered by a perceived threat or by an expected threat: the tension is likely to occur when reaching a valued goal is not certain or when a valued goal is perceived to be threatened, but when the threat actualized or what kind of threat is not known.7
Patterns are not isolated, they instantiate each other and many patterns can modulate how other patterns work. How can this kind of complexity be addressed by the theory lenses approach? Using theories in making the hypotheses of psychological or other types of implications of a pattern can improve the quality of the pattern description for several reasons. First, the implications can be grounded to the used theory. Naturally, the theories used to predict the implications should be empirically tested so that there is enough empirical evidence supporting the used theory; if the predictions provided by the theory are inaccurate, the implications derived from that theory of the pattern are also inaccurate. It should be noted that this testing is not the same as the testing required finding supporting evidence for the theory as a scientific theory; it is the testing that the implications are valid for the gameplay that arises from a game. Second, the use of a theory lets the pattern make use of the vocabulary of that theory and through this have a more precise description (and in some cases also metaphors).
4. CONCLUDING DISCUSSION
In this paper we have introduced the idea of theory lenses approach to game research, and to harvest gameplay design patterns in particular. We argue that it useful in game research because the game phenomena can be approached through many different theories depending on one’s research question, and in many cases possibility of being able to use of multiple theories can be advantageous. Further, theory lenses is flexible in that it does not require that findings should need to model the source domain exactly, and there is a room for the creative interpretation for the specific area of gameplay. In fact, in some instances the approach has successfully used theories that are incompatible in relation to their original domain, and argue that rather than causing inconsistencies for gameplay research this allows several perspectives to be maintained simultaneously during the analysis of a game.
Our uses of Theory Lenses have been implicit in the above-mentioned studies and have been domain dependent. In this paper, we have focused on how to generalize the approach (that is, conjuncting one or more theory from a scientific discipline with a structural analysis of games) and argue that this forms a basis for a gameplay design patterns harvesting method usable in many different contexts.
The approaches we have described allow patterns to be indentified for specific usages. This provides a rationale for the initial usefulness of a pattern within a specific context. However, if this pattern is of more general use can thatcan only be known by observing if it becomes reused. We have noticed this with some patterns, for example, how the patterns AWARENESS OF SURROUNDINGS, SENSE OF SELF, EMOTIONAL ATTACHMENT, INITIATIVE, and CONTEXTUAL CONVERSATIONAL RESPONSES have been useful even if the patterns have been derived starting from several different perspectives.
One could argue that Theory Lenses can be used to affect gameplay design directly without formulating findings as gameplay design patterns. However, the strength of semiformal pattern description is in highlighting the fact that structural features in games are not isolated, but interconnected in various ways. Different Theory Lenses results are, as discussed above, not without weaknesses. Like other ways of identifying patterns, those harvested using Theory Lenses need to be validated in different ways, and should be developed with a broad perspective.
The introduction asked how to ensure that gameplay design patterns identified are of use. Theory lenses provided means to find new patterns, and gain knowledge on their potential consequences, based on scientific theories but without directly answering this question. Although the usability of any patterns depends on what they are to be used for, the three ways of using Theory Lenses provide some general ways of indentify patterns for specific uses.
- The first approach lets pattern identifiers to focus their identification on areas of their own choosing.
- The second approach can be used to derive gameplay features that do not exist yet, which can support innovation.
- Finally, theories, especially predictive ones, can allow the tracing of implications (psychological in our examples) which can improve the quality of the implications of harvested patterns, something which is valuable when one has a clear design vision (as long as the used theories relate to this vision).
Thus, the Theory Lenses shows three different ways of focusing ones work with identifying and describing patterns. Quite probably there are more ways of using theories, and other methods not based on academic theories, but the ones described in this paper show the possibility of developing more specialized ways of harvesting patterns with specific objectives.
An early draft of this paper was presented at Games Research Method seminar (Tampere, 2010). We are grateful of all comments given at that event.
 Alexander, C. et al. 1977. A Pattern Language: Towns, Buildings, Construction. Oxford University Press, Oxford.
 Bergström, K., Lundgren, S. and Björk, S. 2010. Exploring aesthetical gameplay patterns – Camaraderie in four games. In Proceedings of the 14th International Academic MindTrek Conference: Envisioning Future Media Environments (Tampere, Finland, October 6-8, 2010), ACM Press, New York, NY, 17-24.
 Brusk, J. & Björk, S. 2009. Gameplay design patterns for game dialogues. In Proceedings of Breaking New Ground: Innovation in Games, Play, Practice and Theory: Proceedings of the 2009 Digital Games Research Association Conference. (London, UK, September, 2009).
 Björk, S. and Holopainen, J. 2005. Patterns in Game Design. Charles River Media, Hingham.
 Björk, S. 2008. Games, gamers, & gaming – understanding game research. In Proceedings of the 12th International Conference on Entertainment and Media in the Ubiquitous Era (Tampere, Finland, October 8-10, 2008). ACM Press, New York, NY, 64-68.
 Church, D. 1999. Formal Abstract Design Tools. Gamasutra: The Art and Business of Making Games. URL= http://www.gamasutra.com/view/feature/3357/formal_abstract_design_tools.php.
 Costikyan, G. 2002. I have no words & I must design. In Mäyrä, F (ed.) Conference Proceedings of Computer Games and Digital Cultures. Tampere University Press, Tampere, 9-33.
 Currie G. 1995. Image and Mind: Film, Philosophy, and Cognitive Science. Cambridge University Press, Cambridge,.
 Damasio, A. 2005, Descartes’ Error: Emotion, Reason, and the Human Brain. Penguin Books, New York, NY.
 Dennett, C. 1989. The Intentional Stance. MIT Press, Cambridge, MA.
 Denscombe, M., 2007. The Good Research Guide: For Small-scale Social Research Projects, McGraw-Hill International.
 Falstein, N. 2002, Better by design: The 400 project. Game Developer magazine. 9, 3, (March 2002).
 Gamma, E., Helm, R., Johnson, R. and Vlissides, J. 2001, Design Patterns – Elements of Reusable Object-Oriented Software. Addison-Wesley, Reading.
 Garfunkel, H. 2002. Ethnomethodology’s Program: Working Out Durkheim’s Aphorism. Rowman & Littlefield Publishers.
 Glover, J. 1991. The Philosophy and Psychology of Personal Identity. Wipf & Stock Publishers, Eugene.
 Kunda, Z. 1999. Social Cognition: Making Sense of People. The MIT Press, Cambridge MA.
 Lankoski, P. 2010. Character-driven Game Design: A Design Approach and Its Doundations in Character Engagement. Aalto University, Helsinki.
 Lankoski, P. 2007. Goals, affects, and empathy. In The Philosophy of Computer Games Conference (Reggio Emilia, Italy, January, 2007).
 Lankoski, P. & Björk, S. 2007. Gameplay design patterns for believable non-player characters. In Situated Play: Proceedings of the 2007 Digital Games Research Association Conference. University of Tokyo, (Tokyo, Japan, September 24-28,2007), 416-423. URL= http://www.digra.org/dl/db/07315.46085.pdf.
 Lankoski, P. & Björk, S. 2007. Gameplay design patterns for social networks and conflicts. In GDTW 2007 Proceedings (Liverpool, UK, November 14-15, 2007), 76-85, Liverpool John Moores University, Liverpool.
 Latour, B. 2005. Reassembling the Social: An Introduction to Actor-network-theory. Oxford University Press, Oxford, CA.
 LeBlanc, M. 2006. Tools for creating dramatic game dynamics. In Salen, K. and Zimmerman, E. (eds.) The game Design Eeader: A Rules of Play Anthology. The MIT Press, Cambridge MA, 438-459.
 Maxis Software 2003. Sims. Electronic Arts.
 Niiniluoto, I. The aim and structure of applied research. Erkenntnis 38, 1 (January 1993). 1-21.
 Noë, A. 2004. Action in Perception. The MIT Press, Cambridge, MA.
 Oatley, K. 1992, Best Laid Schemes. Cambridge University, Cambridge MA.
 Pajitnov, A. 1984. Tetris. Pajitnov.
 Power, M. and Dalgleish, T. 1997. Cognition and Emotion: From Order to Disorder. Psychology Press Ltd, Hove.
 Spector, W. 1999. Remodeling RPGs for the new millennium. Gamasutra: The Art and Business of Making Games. URL= http://www.gamasutra.com/view/feature/3309/remodeling_rpgs_for_the_new_.php.
 Wasserman S. & Faust K. Social Network Analysis. Cambridge, Cambridge University Press, 1994.
 Zagal, J., Mateas, M., Fernandez-Vara, C., Hochhalter, B. & Lichti, N. 2005. Towards an ontological language for game analysis. In Changing Views: Worlds in Play: Proceedings of the 2005 Digital Games Research Association Conference. Vancouver, Canada.
- Small caps are used to indicate a pattern name. No full descriptions are provided, because each pattern takes several paragraphs or even pages in when completely presented. In addition, the main purpose of this paper is not to extend the pattern collection, but rather to look at how the theory lenses approach affects harvesting patterns.
- It should be noted that this is in no way guarantees improved gameplay experience to players.
- For the detailed description of the pattern SOCIAL MAINTENANCE, see Lankoski .
- The pattern RED QUEEN DILEMMAS is an example of a pattern that builds upon the evolutionary hypothesis with the same name (see Björk and Holopainen ).
- Note that such testing would still have an issue with distinguishing negative results arising from faulty “implementations” of the pattern in the game and false hypothesis.
- This is a borderline case of deriving a pattern from a theory and deriving implications of a pattern from a theory, because the psychological implication is formulated as a pattern.
- Hypothetical reasoning or possible or likely future events can trigger emotions (see Damasio  and Power & Dalgleish ).