Category Archives: Games Studies

Empirical research on narratives

A few weeks ago I published a post that looked at mathematical models of narrative comprehension that could be used for empirical research into how viewers understand narrative films. I also complained that the reason there was a lack of empirical research on narrative comprehension was simply because no-one in film studies has undertaken such research, whereas in many other disciplines an empirically based approach is fundamental. You can read the earlier post here. This week I include some abstracts and some links to papers that do look at the relationship between agents and narratives empirically. An interesting aspect that should also be noted in many of these papers is the way in which concepts of film style have moved into other media, while thinking about narrative in virtual environments and interactive fiction provides new ways of thinking about narrative in the cinema.

As ever, the versions of these papers linked may not be the final published version.

Bizocchi J 2005 Run, Lola, Run: film as a narrative database, Media in Transition 4: The Work of Stories, May 6-8, 2005, MIT, Cambridge, MA. [I’ll have more to say about this paper and the topic of database narratives at a later date in a piece on paraconsistency in narrative cinema].

Clarke A and Mitchell G 2001 Film and the development of interactive narrative, International Conference on Virtual Storytelling: Using Virtual Reality Technologies for Storytelling, 27-28 September 2001, Avignon, France.

This paper explores narration in film and in videogames/virtual environments/interactive narratives. Particular attention is given to their use of the continuity of time, space and action and this is used as a means of classifying different types of work. The authors argue that the creators of these videogames etc. need to have more authorial presence and that this can only be done through abandoning their traditional reliance on the continuity of time, space and action.

Johnson K and Bizzocchi J forthcoming Lost Cause: an interactive film project, Journal of the International Digital Media and Arts.

The paper describes the design, the aesthetics, and the experience of the interactive film Lost Cause. The film is examined from several theoretical perspectives: cinematic roots, narrative construction, interface design, and new media artifact. Lost Cause extends the complex plot structure used by filmmakers such as Altman or Tarentino into an explicitly interactive format. The plot has three interrelated and synchronous threads which are represented in a multiscreen user interface. It culminates in an ending determined by the history of user navigation choices. The paper analyzes the work to reveal critical insights into database narrative, expressive interface design, user agency, and the construction of micronarrative.

Marsh T, Nitsche M, Liu W, Chung P, Bolter JD, and Cheok AD 2008 Film informing design for contemplative gameplay, Sandbox Symposium, 9-10 August 2008 Los Angeles, California.

Borrowing from film and filmmaking styles, techniques and devices that manipulate spectators’ attention and experience, this paper proposes an approach to inform design of games and gameplay to manipulate player’s focus of attention and encourage contemplation — in design features, characters, story elements, etc. or even break the player’s engaged attention in the game/virtual world altogether — to provide meaning, experience and opportunities for learning. Focusing on film styles alternative to the continuity style of Hollywood filmmaking, we discuss examples of design for contemplative gameplay in game-based learning environments/serious games, machinima and augmented and mixed reality games in previous, current and future projects. We propose that one goal of game design is to establish a rhythm between contemplation and engagement, and the appropriate rhythm is determined largely by a game’s genre, platform and/or narrative.

May J and Barnard PJ 1995 Cinematography and interface design, in K Nordby, PH Helmersen, DJ Gilmore, and SA Arnesen (eds.) Human-Computer Interaction: Interact’95. London: Chapman and Hall: 26-31. [NB: there isn’t a direct URL for this paper, but if you google the title you should find the pdf version easy enough].

Interface designers are increasingly relying on craft based approaches to compensate for a perceived lack of relevant theory. One such source is cinematography, where film-makers succeed in helping viewers follow the narrative across cuts which change the information on the screen. Cinematography has evolved over the last century, and its rules of thumb cannot be applied directly to interface design. We analyse film-makers’ techniques with a cognitive theory (ICS) and show that they work by preserving thematic continuity across cuts. Expressing this theoretically allows us to extrapolate away from film, applying it to screen changes in interface design.

Nath S 2004 Narrativity in user action: emotion and temporal configurations of narrative, 4th International Conference on Computational Semiotics for Games and New Media, 14-16 September 2004, Split, Croatia.

One of the core problems in Narrative Intelligence is maintaining the narrative nature of event sequences that emerge owing to user participation. This paper challenges the common premises and assumptions about the nature of human action and experience that underlie common approaches to finding a solution to the problem of narrative structuration. An in-depth analysis of the temporality of human action and experience provides important indicators on how the problem can be approached. It is argued that user emotion is not just a by-product of narrative structure, but a critical factor in maintaining narrativity. Finally, it is indicated as to how patterning of emotions can regulate user action and the creation of a subjective experience.

Rowe JP and Lester JC 2010 Modelling user knowledge with dynamic Bayesian networks in interactive narrative environments

Recent years have seen a growing interest in interactive narrative systems that dynamically adapt story experiences in response to users’ actions, preferences, and goals. However, relatively little empirical work has investigated runtime models of user knowledge for informing interactive narrative adaptations. User knowledge about plot scenarios, story environments, and interaction strategies is critical in a range of interactive narrative contexts, such as mystery and detective genre stories, as well as narrative scenarios for education and training. This paper proposes a dynamic Bayesian network approach for modelling user knowledge in interactive narrative environments. A preliminary version of the model has been implemented for the CRYSTAL ISLAND interactive narrative-centred learning environment. Results from an initial empirical evaluation suggest several future directions for the design and evaluation of user knowledge models for guiding interactive narrative generation and adaptation.

This paper by Rowe and Lester is from the Intellimedia group at North Carolina State University, which publishes a wide range of papers on human-computer interaction, virtual learning environments, and narrative interaction. Their website can be accessed here.


The ontology of games

As games studies has made its way into the academy, it has (like film studies before it) been forced to establish itself as a field of inquiry and to determine the nature of its object of study. As film studies was forced to address the question ‘what is cinema?,’ games studies has been forced to consider the nature of games. It is the ontology of games that has been the subject of debate. This debate has primarily focussed around the question of whether games are narratives or something else, and has been played out as a conflict between ludologists and narratologists. I do not intend to go into any detail on this debate, as it has been covered extensively elsewhere (see Simons 2007). I also don’t see the point in dealing with much that has been said on this issue because this is an intellectual argument that was flawed from the start, as fundamental questions regarding the ontology of games were not been adequately explored. This post addresses questions that are relative to both sides of the ludology/narratology split by focussing on two areas: the ontological and psychological uses of the term ‘narrative,’ and the problematic nature of distinguishing between ergodic and non-ergodic texts.

The ontology and experience of narratives

The term narrative may be used in two different senses:

  • It may be used in an ontological sense to classify the nature of a text. In this sense, we are using ‘narrative’ to refer to what a text is and what it is not: Pride and Prejudice is a narrative, North by North West is a narrative, but Shakespeare’s sonnet 116 is not a narrative and the Argos catalogue is not a narrative.
  • ‘Narrative’ may also be used to refer to the cognitive operations in processing information, and in this sense refers to the psychological structuring of events by an agent. These ordered events may then be retold, and most definitions of narrative refer to some idea of the telling of an event, but this is not strictly necessary. It is required only that an agent will organise their experience as a narrative for the term to apply in this sense, and whether or not they tell anyone is a moot point. See, for example, János Lázló’s The Science of Stories (2008).

The etymology of ‘narrative’ is from the Latin ‘gnarus’ – to know. The use of ‘narrative’ as a noun in its ontological sense appears to emerge in the 16th century.

The flaw in some theories of games studies is that they fail to distinguish between the different uses of ‘narrative’ (as do many narratologists), and statements about the ontological nature of a text are assumed to apply to an agent’s psychological experience: a game is not a narrative therefore narrative theory cannot be of use in explaining the experience of an agent (i.e. a player) in games studies. This reflects a strong argument about the difference between games and narratives – i.e. that a game is categorically (ontologically) not a narrative (see, for example, Eskelinen 2004). A weaker argument is that games have some elements in common with narrative forms, but there are other elements that cannot be explained by (some version of) narratology.

The strong argument assumes that the ontology of a text determines the nature of an agent’s experience. However, there is no reason why an agent cannot organise their experience of a non-narrative text as a narrative at the psychological level. A game of Cluedo, for example, is not a narrative in an ontological sense, but the experience of a player may be expressed as a narrative.

Eskelinen (2004) gives the following argument: ‘Luckily, outside theory, people are usually excellent at distinguishing between narrative situations and gaming situations: if I throw a ball at you, I don’t expect you to drop it and wait until it starts telling stories.’

Simons (2007) responds to this by pointing out that a distinction between event and narrative of the sort outlined above can exist: ‘If there is still anybody waiting, it must be for Eskelinen to explain the point of this stab at narratology. Narratologists will be happy to explain to him the difference between the act of throwing a ball and the act of recounting that (f)act.’ Narrative thinking aspires to make sense or to establish coherence’, and events become socially visible through their retelling as narrative (Lazlo 2008: 3). But this does not depend on the fact that those events were narrative events (in the ontological sense).

The psychological experience of a situation (a narrative) is not determined by the ontology of the situation (a gaming situation). That an agent can make an ontological distinction between narrative and gaming situations is not in dispute, but it is for Eskelinen to demonstrate that there is a necessary relationship between ontology and experience, so that for an agent a gaming situation cannot be experienced as a narrative. Equally, there is no reason why we should experience Pride and Prejudice as a narrative just because it has a particular ontology as a text. Statements of ontology (‘this is not a narrative’) are not statements about the epistemological state of an agent.

A secondary issue here is the role of non-narrative passages in narrative texts: not everything in a story will necessarily be part of a narrative and so we have to make allowances for the fact that the narrative/non-narrative dichotomy will not be as simple as we would like. For example, in Moby Dick, Herman Melville puts in a great amount of information about whales and whaling, but not all of this is relevant to the furtherance of the narrative: chapter 32 – ‘Cetology’ – sets out in great detail zoological information on many different types of whales, which Ishmael attempts to classify by book size (e.g. folio, quarto, etc). This is not narrative information, even though we would class Moby Dick as a narrative – it is more akin to an encyclopaedia or scientific text. Abbott (2000) argues that narrative ‘tolerates’ non-narrative because the latter ‘can sit on top of it.’ In this way, Moby Dick can tolerate non-narrative passages because it has a narrative platform. This approach may be useful in thinking about some games (e.g. games that have been spun-off from films and which have a pre-existing narrative form), though not necessarily all games. It may also be the case that for some games a non-narrative platform tolerates narrative information.

What kind of text is this for you?

There is a further problem regarding the ontology of games relating to the when a text can be classed as ergodic, and the nature of experience is crucial understanding this problem. The concept of an ergodic text was introduced by Aarseth (1997) to distinguish texts that are characterised by a topological structure enabling variable expression and which require non-trivial extra-noematic effort of an agent (e.g. cybertexts, games, etc) from texts which lacked these feature (i.e. non-ergodic texts, such as novels, films, etc).

In an earlier post (here), I argued that the distinction between ergodic and non-ergodic texts as presented by Aarseth is flawed as the extra-noematic element may be absent from some texts capable of variable expression, or that in some cases it is too ill-defined to be a useful concept. The distinction between these two types of texts should be made on the basis of the types of reasoning they demand of an agent. Non-ergodic texts require an agent to reason about the nature of the text-world and events and characters in that world, and to form beliefs about that world; while ergodic texts require this type of reasoning and to go further to make decisions and to act on the basis of the beliefs held about the world. The types of reasoning involved can be expressed in terms of the concepts of credal and pignisitic reasoning used in the transferable belief model (Smets and Kennes 1994), where credal reasoning is reasoning about the state of a world and pignistic reasoning is involved in decision making.

Consider the following example: two people are in a room – one is playing James Cameron’s Avatar: The Game on a Wii, whilst the other watches the player in the room and the game on the screen. Is James Cameron’s Avatar: The Game an ergodic or non-ergodic text? The answer would seem obvious: this is an ergodic text – the word ‘game’ is even in the title. Things, however, are not so straight forward.

For the player the argument that this game is an ergodic text is straight forward. Unlike the film, this version of Avatar has the topological structure that enables variable expression – the game can be played several times with different decisions made by the player so that no two performances of the game will be the same. The extra-noematic element is present in this case, as the player manipulates the game’s control systems to move her in-game character. This fulfils Aarseth’s definition of an ergodic text. The player is required to reason about the world of the game at the credal level, but also to make decisions about what action to take in playing the game and so is required to engage in pignistic reasoning. This satisfies my definition of an ergodic text. It would seem that in this instance we have a clear-cut case of an ergodic text.

The same argument could not be made for the non-playing observer. Aarseth (1997: 4) uses the example of a spectator and a passenger to characterise the position of an agent (reader, viewer, etc) who is unable to influence the text-world they experience.

A reader, however strongly engaged in the unfolding of a narrative, is powerless. Like a spectator at a soccer game, he may speculate, conjecture, extrapolate, even shout abuse, but he is not a player. Like a passenger on a train, he can study and interpret the shifting landscape, he may rest his eyes wherever he pleases, even release the emergency brake and step off, but he is not free to move the tracks in a different direction. He cannot have the player’s pleasure of influence: ‘Let’s see what happens when I do this.’ The reader’s pleasure is the pleasure of the voyeur. Safe, but impotent.

The non-playing observer has no means of making decisions in the context of the game, and so while he may reason at the credal level about the nature of events in the world of the text he does not make the shift to decision making at the pignistic level. He is in the ‘safe, but impotent’ position described by Aarseth. The observer of a game is in the same situation as a viewer in a cinema, and so from his point of view James Cameron’s Avatar: The Game is non-ergodic.

Therefore, the answer to the above question depends on whose perspective you take. To the player, James Cameron’s Avatar: The Game is ergodic. However, from the perspective of the non-player, the game is a non-ergodic text. We could certainly make an argument that for the player, the game is ontologically different from a film; but from the perspective of the non-player, there is no ontological difference between the game and the film of Avatar.


The ontology of games is a complex matter that requires exploration beyond saying ‘this is an ergodic text’ or ‘this is not a narrative.’ In both of the cases discussed above, the common thread is that the experience of an agent matters, and the proper basis for a theory of texts and the experience of texts (in both games studies and film studies) should perhaps be the epistemology of an agent rather than the ontology of the text.


Aarseth EJ 1997 Cybertext: Perspectives on Ergodic Literature. Baltimore: The Johns Hopkins University Press.

Abbot HP 2000 What do we mean when we say ‘narrative literature?’ Looking for answers across disciplinary borders, Style 34 (2): 260-273.

Eskelinen M 2004 Towards computer game studies, in Noah Wardrip-Fruin and Pat Harrigan (eds.) First Person: New Media as Story, Performance, and Game. Cambridge, MA: The MIT Press: 36-44.

Lazlo J 2008 The Science of Stories: An Introduction to Narrative Psychology. Hove: Routledge.

Simons J 2007 Narrative, games, and theory, Game Studies 7 (1):, accessed 4 December 2009.

Smets P and Kennes R 1994 The transferable belief model, Artificial Intelligence 66 (2): 191–234.

The transferable belief model in film and games studies

This post is another draft paper, this time focusing on the difference between ergodic and non-ergodic texts and so is of relevance to games studies as well as film studies. I use an approach that I don’t think has been applied to either of fields yet: the transferable belief model, which is a mathematical theory of evidence. Hopefully soon I will be able to outline how this model will be of use in film studies in more depth, along with considering its relationship to Bayesian approaches to modelling viewer behaviour. The idea is to apply these models to the empirical analysis of the beliefs of real spectators so that it may be possible to make some statements about how we understand films that are more than theoretical but which have a solid evidential basis. The article can be downloaded as a pdf file here:Nick Redfern – Credal and pignistic reasoning in ergodic and non-ergodic texts.


This paper discusses the difference between ergodic and non-ergodic texts by considering the different levels of reasoning required of an agent in each case. The difference indentified between such texts is based on the distinction between credal and pignistic reasoning in the transferable belief model. It is argued that non-ergodic texts require an active agent to reason about the state of the world, and thus operate at the credal level; while ergodic texts require that the belief function of an agent be transformed into a probability function for the purposes of decision making, and therefore entail both credal and pignistic reasoning. The difference between ergodic and non-ergodic texts considered in these terms is illustrated through comparing narratives from the CSI: Crime Scene Investigation franchise.