"Play, Memory": Shadow of the Colossus and Cognitive Workouts

by David Ciccoricco

This paper applies the distinction of episodic and procedural memory from cognitive science to the experience of contemporary video games. It aims to illustrate how participation in the simulative digital environments of "coherent world games" not only draws on but also relies on both forms of memory. Toward this end, the paper employs Fumito Ueda's _Shadow of the Colossus_ (2005), a game that combines a complexity of interaction (play and puzzle-solving) with a narrative complexity that allows for - and encourages - an interpretative understanding of its characters and storyworld.




1. A Literary Turn?
As video games continue to occupy their place as a culturally dominant form of popular entertainment, it is clear that some forms of game design and production are growing in artistic – and indeed, literary – sophistication and complexity. Game designers see themselves as artists in their own right involved in a defining moment of cultural production (Freeman 2004:16). It is also clear that the commercial video game industry is pursuing game creation with more emotionally compelling and complex story mechanics (the high profile contracting of Stephen Spielberg by video game giant Electronic Arts in 2005 offers a crude indication of this trend). In his endorsement of David Freeman's Creating Emotion in Games (2004), Scott Miller, CEO of 3D Realms and co-creator of Max Payne, writes, ‘It's time game designers grasp the power of creating games with emotional depth, meaningful characters and dialogue, and themes that reach players’ hearts and really – I mean REALLY – make them care about their game's story’ (Freeman 2004: jacket blurb).

Despite these popular trends, the emphasis on storytelling in games and the concomitant project of intensifying their emotional impact – what Freeman (2004:7) has dubbed ‘emotioneering’ – does not mean that narrative complexity is coming to the rescue of an otherwise deficient form and saving it, in the nick of time, from the brink of inadequacy (as the language of the conventional narrative arc might frame it). Nick Montfort (2003) has cautioned against such a view, casting a critical eye on the ‘hope […] that action-oriented games, which people clearly do not play for narrative reasons, can be enhanced with good stories – as if “story” might be the deus ex machina that could arrive to save an otherwise incomplete gaming experience’ (499). It is difficult to assert, unconditionally, that narrative mechanics do not motivate gameplay or contend that games cannot be enhanced in this manner.
[1] The underlying premise of Montfort's statement, however, is instructive: what makes games theoretically engaging artifacts is not their function as storytelling machines but rather the experience they yield as rule-based simulations.

It should come as no surprise that out of a range of critical approaches invoked to study this emergent field, narrative theory and the custom-built discipline of game studies – or ludology – have been cast as opposing methodologies. The opposition is rooted in a tension between the unambiguous formal rules and algorithms that describe how the game works, and the ambiguous (and often incomplete) representation of the gameworld that is, arguably, non-essential to the game in a strict sense. That is, remove the formal rules and you no longer have a game; remove the fiction and you are still left with an abstract or abstracted game. But a radical ludology, however necessary it may have been in order to clear ground for a new discipline, ultimately fails to justify its insistence on a mutually exclusive approach whereby either a video game is 1) read and interpreted as a text (which is the wrong way according to the ludologist), or 2) treated as a rule-based simulation (which is the right way according to the ludologist). The partition is a particularly unproductive one given that some games place a heavy investment in constructing thematic frameworks and portraying a fictional world – what Jesper Juul (2005) describes as ‘Coherent World Games’, while others do not.

In coherent world games, it is possible to suggest that, aesthetically, the rules of the game are inseparable from the fiction, and further explore the relationship between story mechanics and game mechanics. Juul (2005:15) considers the question of an ‘essential’ relationship between ‘theme’ and ‘structure’, a question that not only takes this inter-relationship as a given but also offers a productive starting point for criticism of actual games that deliver coherent worlds. It is possible to further qualify the player's experience of coherent world games by looking through other lenses outside of – and perhaps in between – ludology and literary theory (and there is after all much more to literary theory than narratology). This paper considers the relationship between story mechanics and game mechanics by applying the basic distinction of episodic and procedural memory from cognitive science, illustrating how participation in digitally simulated environments not only draws on but also relies on both forms of memory. To demonstrate this distinction in practice, I discuss Fumito Ueda's Shadow of the Colossus (2005), a game that combines the complexity of gameplay (puzzle-solving, exploration, combat) with a narrative complexity that encourages an interpretative understanding of its represented world.

2. Accounting for Memory
The distinction between short term memory and long term memory is a familiar one. Short term memory (or working memory) is what allows you to keep the words you are reading at this very moment in your head long enough to make sense of them. Long term memory refers to everything remembered, from childhood to this morning. It can be reliably divided into two subgroups: declarative and procedural (Squire et al. 1993: n.p.). Declarative memory can be consciously recalled and described in words, and for this reason it is also called explicit memory. In contrast, procedural memory, which is also called implicit memory (Schacter 1987:501), refers to skill sets or organizational routines, such as riding a bicycle or swimming. It is a form of memory that operates without you being conscious of it. The distinction basically describes a form of memory that is expressed through recollection and one that is expressed through performance (Squire et al. 1993: n.p.).

As Figure 1 shows, declarative memory is further subdivided into semantic and episodic memory (Tulving 1983). Semantic memory refers to factual or general knowledge; it allows us to remember the name of a significant historical figure, for example. Semantic memory is not context dependent: the time and place you acquired this form of knowledge is not significant to your storage and retrieval of it. Episodic memory is what most people think of when they think of memories generically, and it differs slightly from semantic memory. It refers to the recall of certain episodes in one's life and amounts to our personal stable of autobiographical details. Episodic memory, then, is context dependent in that it is always tied to a specific time and place. Because it involves sequencing of events into coherent and meaningful stories episodic memory is also understood as a form of “narrative” memory. The separation of semantic and episodic memory is still a matter of debate, as it is unclear to what extent they imply discrete cognitive functions (Squire et al. 1993: n.p.). But the governing distinction of declarative and procedural memory is widely accepted in the field (Squire et al. 1993: n.p.). In turn, different modes of learning have been recognized to accommodate for these different memory systems in practice.

Figure 1: Common distinctions for Long Term Memory. The memory systems discussed in this paper, procedural and episodic, are highlighted.

There is a groundswell of research on games and cognition that looks at the (positive) pedagogical impact of playing video games. Scholars and critics such as James Paul Gee (2003a) and Steven Johnson (2005) have already established that video games, to put it plainly, can “exercise your brain”. This body of research focuses mainly on the procedural learning that results from being immersed in simulated worlds – the skills of pattern recognition, systematic thinking, even patience. Decision-making also plays a crucial role, specifically time-constrained decision-making that involves solving all manner of problems and puzzles in order to advance through the game. Typically, the puzzles become more subtle and complex as the player progresses through the levels, which in pedagogical terms is a process directly analogous to “regime of competence” learning. Other research on games and cognition has focused on visual and spatial intelligence: ‘Video game play frequently activates cognitive skills by calling upon the player to construct mental representations of space in order to win the game. Repeatedly playing video games may cultivate visual spatial skills’ (Calvert 2005:127). In all of these cases, the idea is that there is a ‘clear transfer of skills from gameplay to real-world application’ (Johnson 2005:209-10); ‘[p]layers are learning HOW to think as much as learning WHAT to think’ (Malloy 2000; emphasis in original). Players are thus learning processes in the game that can be transferred – or recalled – outside of the game as well. Procedural memory, which is effectively performed recollection, is what allows them to do so.

Coherent world games invested in the story and themes of their fictional worlds, however, mobilize both procedural and episodic memory in order to make the gameplay meaningful for the player. After all, such games are not only dynamic real-time interactions, but also elaborate fictions teeming with narrative detail; acts of interpretation can – and do – accompany the act of play (as the following analysis of Shadow of the Colossus will illustrate). It is possible then to align the distinction of procedural and episodic memory with the ludic and narrative elements of gameplay respectively: performative gameplay calls on our procedural memory, whereas the fictional and narrative elements of the game call on our episodic memory.

Because video games and adventure games in particular typically force players to engage with simulated environments in a rapid and reflexive way, it only makes sense that there is a focus on procedural response systems in terms of the player’s ‘cognitive workout’ (Johnson 2005). But that is still not the whole story, so to speak, and the comparably slower, more reflective response implicit in episodic memory comes into play. A discussion of both systems can move toward a further understanding of the kind of “meaningful play” Katie Salen and Eric Zimmerman (2005) describe as ‘the emotional and psychological experience of inhabiting a well-designed system of play’ (61).

Because episodic memory refers to one’s personal narrative experience, it is necessary to make a further assumption before pursuing this theoretical model: it is necessary to allow a conflation of our own embodied narrative memory with the fictionalized narrative memory of our playable character,
[2] our prosthetic agent in the gameworld. All gameplay requires some degree of make-believe, and in this regard such an assumption might seem unproblematic. At the same time, it is important to acknowledge that the relationship between the embodied gamer and the playable character is always layered, and the player’s consciousness can be framed in at least three ways: ‘as character in a simulated world, as a player in a game, and as a person in a larger social setting’ (Salen and Zimmerman 2004:454; emphasis in original). For the purpose of conveying the model, however, it is necessary to make this assumption given the simple fact that the playable character is not cognitively equipped to draw on a “personal” narrative memory; for this, he or she needs you as much as you need them.


3. The Play of Memory's Shadow
In Fumito Ueda's Shadow of the Colossus (2005), a game made for the Sony PlayStation2, “you” are the nameless wanderer who must travel across a vast expanse on horseback and defeat sixteen giant “colossi”. This young wanderer (also referred to as “Wander” or “Wanda”) strikes a deal with a disembodied entity known as Dormin, who agrees to bring his sacrificed lover back to life if these sixteen giant beings are killed. He has only a bow and his ancient sword to complete his task, and only his horse, Agro, to accompany him. No two beasts are the same. Some stand on two legs, others on four; some fly; some are intensely aggressive, whereas others do not even seem to notice your presence until your first arrow pierces its hide. Furthermore, they each must be brought down in a different way, by exploiting a different weakness peculiar to each one.

For example, one colossus carries a massive rectangular club, and has a distinctive arm band around its club-wielding arm. In order to destroy it, you must run your blade through its weak spot – denoted by a luminous “sigil” – that sits on the back of its head. You must first, however, find a way up there. A slightly raised, circular platform made of seemingly solid rock offers the best clue: you must stand on the platform and bait the giant, who will laboriously swing the club so that it hits the platform, just after you roll out of the way of course. The force of the club on the platform the first time breaks the arm band of the giant, which will allow you, upon its second swing, to climb directly up the club and continue up the exposed arm (see Figure 2). From there, you have to either find parts of the giant’s armor that will act as ledges to stand on, or – failing that – you can grab hold of its thick pelt. Grabbing on too long, however, will drain your energy, so you will have to time your ascent. When you summit the beast, you will repeatedly run your sword into the designated weak spot. The rest is academic.

Figure 2: Ascending the club of the colossus. Copyright © StrategyInformer.com

In these fight scenes, the player is drawing on procedural memory to puzzle-solve and conquer each new beast. With the repetition of failed attempts for each particular giant, as well as each successful kill that allows you to find the next one, you are filtering out non-essential information and streamlining the interaction to reach your goal more quickly and efficiently. As Robert Sylwester (1997) explains, this type of response is fast and reflexive, and involves an action that typically has a direct and immediate outcome in the environment: one ‘identifies the fearful and survival elements in a situation, and quickly activates automatic response patterns if survival seems problematic’ (n.p.). In this case, our environment is virtual, but one glance at our life meter confirms that our “survival” is problematic nonetheless.

This is not the only way to respond to challenges or changes in our environment, however, whether we are in an actual environment or a virtual one. A comparably slower response system considers ‘objective factual elements’ (Sylwester 1997: n.p.) of a given situation, and draws from our related store of personal experience (our episodic memory bank) to make sense of it. This response ‘often functions through storytelling forms and sequences, and so is tied heavily to our language and classification capabilities’ (Sylwester 1997: n.p.). A lot happens in between the epic fight sequences – or “boss” battles – that mark your progression through game levels, though some would suggest that what happens is a lot of nothing. And that is the point. As adventure games go, Shadow of the Colossus is unique in that it makes time and space for a reflective response (perhaps even more remarkable for a game meant to entertain and sell, not educate); that is, it makes you think about what you’ve done after you've done it while, experientially, it may seem as though you’re not doing anything in terms of a material outcome in the gameworld.

The game accommodates this response in two ways. First, what follow after each successful battle are long cut scenes, cinematic intermissions that can advance the plot and project the fictional – and indeed literary – qualities of the game (see Figure 3).
[3] In these scenes, you return to the temple, have a dialogue with Dormin, the disembodied entity, who briefs you on the next beast, and you then carry on, using your sword’s magic to guide you toward the next colossus.

Figure 3: The wanderer brings his deceased lover to the temple in the cinematic prologue. Copyright © StrategyInformer.com

In addition to these cut scenes, long periods of riding punctuate your experience of the gameworld. There are no smaller, intermittent enemies (common in adventure games); there is no-one at all besides you and your horse (see Figure 4). You ride across vast sepia-toned deserts and rambling plains, valleys full of mist and dense forests that splinter whatever sunlight they allow in. Occasionally, you lift your magic sword to the sun, and a conspicuous convergence of light will indicate that you are moving in the right direction. Then you keep riding.

Figure 4: The wanderer and Agro take a break from riding. Copyright © StrategyInformer.com

What does one do when riding alone all that time? You think. You think about the fact that you are about to bring down another one of these awe-inspiring creatures even though you know that they have not wronged you in any way, and they are definitely not expecting you and your sword. All you know – all you remember – is that you must kill them in order to complete your quest.

The sense of moral ambiguity evoked by the storyworld is one of the main reasons why Shadow of the Colossus is an aesthetically compelling video game. One reviewer suggests that the reason for these long stretches of riding is that they make ‘the player feel the isolation that the story tells the player they should feel’ (Sherman 2006: n.p.). The observation speaks to Juul's (2005:15) question of an essential relationship between ‘theme’ and ‘structure’ in video game design, and serves as a productive example of story mechanics and game mechanics working in concert. With regard to gameplay and cognition, furthermore, the lengthy cut scenes and long stretches of riding in isolation are both design qualities that allow the player to respond reflectively to the storyworld, a response that calls on the player’s own form of narrative memory. In effect, the player not only inherits the task of the wanderer and the tools with which to accomplish that task, but also (potentially and ideally) the psychological baggage that his ordeal entails. The process of drawing on procedural and episodic memory systems is thus critical to the player’s experience as the character of the wanderer in the simulated world of Shadow of the Colossus.


4. Reading into Games
Even though the player experiences the world of Shadow of the Colossus AS the wanderer, the player’s response to the game clearly transcends this “in-character” response. The subjective experiences of an empirical player color both what is brought to the game and what is taken away from it. In addition, the creator(s) of the game to some extent prompt and pre-structure the player’s response through various elements of the game’s design, and in such a way that is similar to how we arrive at meaning when we read a fictional text.

Literary theorist Wolfgang Iser (1978:27-38), for instance, differentiates between 1) the ‘actual reader’, who brings his/her individual experiences and preoccupations to the text, and 2) the ‘implied reader’, who is established by the ‘response-inviting structures’ of the text and in a sense produced by the text itself. Coherent world games invite a parallel notion of an actual and implied gamer, in which an implied gamer is similarly constructed by various components of the game’s fiction – such as characterization, themes, or even expressly narratological elements such as point of view.[4] In this regard, the player’s episodic memory is also informing what might be described as "extra-textual" acts of interpretation, which is a productive and inevitable consequence of reflective engagement with coherent world games. As Gee (2003b: n.p.) writes, players should ‘think reflectively and actively about the games and their connections to other texts, activities, and the world’ (Gee 2003b: n.p.). From the perspective of game design, some games do not simply allow for such interpretation: they cry out for it.

Like your typical adventure story fare, Shadow of the Colossus borrows from diverse mythologies. There is a magic sword, a Sleeping Beauty, a Faustian pact, and of course a David and Goliath. Each giant has a weak spot reminiscent of Achilles, and these are revealed to the wanderer in the form of the sigils, which are derived from Jewish mysticism and the Kabala. (It can be said that a coherent world game does not necessarily imply a coherent “theme” game). Some of the allusions are more subtle. The name of the disembodied spirit/demon in the temple, Dormin, is Nimrod spelled backward. Nimrod is the first super-human, God-on-Earth figure in Biblical lore (Genesis 10.9), and is said to have persuaded the masses to build the Tower of Babel. Upon Nimrod’s death he was cut to pieces and spread throughout the earth. In Shadow of the Colossus, we learn that Dormin’s soul was separated into sixteen parts and sealed in the Colossi, who are spread out across the sacred land. Killing these giants [spoiler warning] will in turn set him free.

Beyond their self-contained and unambiguous rules, the inter-textuality of such games is obviously part of what makes them meaningful – from politically activist games by independent designers to the pulp of the PlayStation mainstream. But it remains a matter of debate how far we can or should take this practice of interpretation.[5] In the gameworld authored by Ueda, for instance, we might consider the allegories implicit in the notion of the colossus itself. After all, it names not only a mythological giant but also a giant of a much different nature: the monstrous computer used by the British to decode German messages during World War II, which was arguably the earliest electronic digital computer produced. Twelve of these Colossi contributed to the Allied victory, but because they were top secret, Churchill ordered their destruction immediately after the war, much to the horror of the engineers who had devoted themselves to their creation (Copeland 2006:172). Like the unique giants killed by the wanderer, the computer colossi were to be erased from history, and erased from memory, before their popular discovery (which has only come with the recent declassification of documents in 2005). Moreover, German media theorist Friedrich Kittler's (1999:257-63) conception of the Colossus as an informatic monstrosity, one that has subsumed all power of symbolic manipulation, could also have a bearing on our reading of this game, and what we are (symbolically) destroying. And, if we read the game in light of its contemporary cultural and political context, it would be difficult to ignore the fact that piercing a giant’s head with a final death blow results in a geyser-like explosion of black fluid that looks unmistakably like an oil well.

Are all or any of these inter-texts and observations fair game? Or are we just reading into things? In the end, it might all be a matter of how we want to exercise our brains, and what kinds of deposits we have made in our episodic memory banks. Nonetheless, what remains clear is that distinctions from the cognitive sciences can inform both ludologically and narratologically inclined approaches to gameplay, and more generally lead to a richer understanding of games in aesthetic and pedagogical terms.



1. Montfort is writing in the context of Sherry Turkle’s ‘Video Games and Computer Holding Power’ (1984), which is based on extensive empirical data on the experience of gamers. The advances in graphical capability, cinematic special effects, and, as I am suggesting here, the increasing literary sophistication of games, however, has to some extent invalidated Montfort's statement. At least in the popular realm, empirical evidence on fan forums would suggest that what sets a game apart is quite possibly its story. On a fan forum for Sony’s God of War (2005), for example, one player writes, ‘[R]ecently, I borrowed God of War from a buddy of mine. The storyline intrigued me so much I decided to write a plot summary of the game' (http://faqs.ign.com/articles/675/675093p1.html, cited 20 February 2007).

2. The concept of “playable character”, which denotes agency, clearly transcends “character”, which denotes a fictional person that elicits a sense of identification from a reader or viewer as it may be. Nevertheless, I use the term “playable character” as opposed to “avatar” to reserve the latter, in line with its origins in role-playing games, for discussions of playable characters that are configurable to some degree by the user.

3. During cut scenes, the player is temporarily a viewer only, not engaged in gameplay. For this reason, cut scenes have been considered as problematic “non-game elements” in some circles and indispensable for conveying the game’s fiction in others (see Juul 2005 [16, 135] for more on this point).

4. Diane Carr (2006:41) has drawn a similar distinction in ‘Games and narrative’ – and even refers to the “implied player” as a ‘textual construct’ (my emphasis).

5. Discussing varied interpretations of Tetris, Juul (2005:133) has established that some interpretations are more convincing than others. Furthermore, Juul (2005:195) has discussed a divide between more experienced players who are more interested in game rules and less experienced players who are more interested in game fiction. I would propose adding a third "category" here: those who, whatever their level of gaming experience, are watching someone else play. These people are participating in the performative aspect of gameplay (as audience) as well as its social aspects, offering verbal input – and perhaps criticism – of the player’s moves. These participants (also referred to as “backseat button-mashers”) are arguably more attuned to analyses or reflection given that they are not otherwise occupied with direct cybernetic control of the game. Not only can these indirect participants help figure out puzzles for the player by observing the gameworld in a less constrained manner, they are more likely to view the game as an aesthetic object for the same reason.



CALVERT, Sandra L. (2005). Cognitive effects of video games, in J. Raessens and J. Goldstein (eds) Handbook of computer game studies, pp. 125-32. Cambridge, MA, MIT Press.

CARR, Diane, BUCKINGHAM, David, BURN, Andrew and SCHOTT, Gareth (2006). Computer games: text, narrative and play. Cambridge, Polity Press.

COPELAND, Jack (2006). Colossus: the secrets of Bletchley Park's code-breaking computers. Oxford, Oxford University Press.

FREEMAN, David (2004). Creating emotion in games: the craft and art of emotioneering. Berkeley, CA, New Riders Publishing.

GEE, James Paul (2003a). What video games have to teach us about learning and literacy. New York, Palgrave Macmillan.

______ (2003b). Live colloquy: ‘Video games in the classroom?’. [online]. Chronicle of higher education, 27 August. Last accessed 20 February 2007 at: http://chronicle.com/colloquylive/2003/08/video/.

ISER, Wolfgang (1978). The act of reading: a theory of aesthetic response. Baltimore, MD, The Johns Hopkins University Press.

JOHNSON, Steven (2005). Everything that's bad for you is good. London, Riverhead Trade.

JUUL, Jesper (2005). Half-real: video games between real rules and fictional worlds. Cambridge, MA, MIT Press.

KITTLER, Friedrich (1999). Gramophone, film, typewriter. Stanford, CA, Stanford University Press.

MALLOY, Thomas E. (2000). Teaching deductive, inductive, and inferential logic through interactive online computer simulation. [online]. Last accessed 20 February 2007 at: www.psych.utah.edu/malloy/J-ITM_Difference_to_Inference00-09-26/JITM-Diff_to_Inf.htm.

MONTFORT, Nick (2003). Introduction to ‘Video games and computer holding power’, in N. Wardrip-Fruin and N. Montfort (eds) The new media reader, pp. 499-513. Cambridge, MA, MIT Press.

SALEN, Katie and ZIMMERMAN, Eric (2004). Rules of play: game design fundamentals. Cambridge, MA, The MIT Press.

______ (2005). Game design and meaningful play, in J. Raessens and J. Goldstein (eds) Handbook of computer game studies, pp. 59-80. Cambridge, MA, MIT Press.

SCHACTER, Daniel L. (1987). Implicit memory: history and current status. Journal of experimental psychology: learning, memory, and cognition, 13, 501-18.

SHERMAN, Ben (2006). Story mechanics as game mechanics in Shadow of the Colossus. Gamasutra. [online]. Last accessed 20 February 2007 at: http://gamasutra.com/features/20060328/sherman_01.shtml.

Spielberg takes film magic to EA (2005). BBC news, 14 October. [online]. Last accessed 20 February 2007 at: http://news.bbc.co.uk/2/hi/technology/4341118.stm.

SQUIRE, L. R., KNOWLTON, B. and MUSEN, G. (1993). The structure and organization of memory. Annual review of psychology, 44, 453-95.

SQUIRE, Larry R. and SCHACTER, Daniel L. (eds) (2002). Neuropsychology of memory. 3rd ed. New York, The Guilford Press.

SYLWESTER, Robert (1997). Bioelectronic learning: the effects of electronic media on a developing brain. Technos quarterly, 6 (2). [online]. Last accessed 20 February 2007 at: www.ait.net/technos/tq_06/2sylwester.php.

TULVING, Endel (1983). Elements of episodic memory. Oxford, Oxford University Press.

UEDA, Fumito (2005). Shadow of the Colossus. Sony Computer Entertainment Inc.