Why You Play Games

Over on the ihobo site today, a post explaining why people enjoy videogames. Here's an extract:

Whatever reason you enjoy playing videogames, there is one specific part of the brain that lies behind it.

For some time, game designers have known how to apply the work of the behaviourist B.F. Skinner to videogame design, creating reward schedules (such as levelling mechanics in computer RPGs) that will hook players in and keep them addicted while they play, or setting up short and medium term challenges to overcome in order to produce the same kind of pattern. These reward schedules work because when we win or attain something, the pleasure centre in the brain (the nuclues accumbens) releases a neurotransmitter called dopamine, which is chemically similar to cocaine.

But recent research into cognitive functions using functional magnetic resonance imagine has shown that these are not the only ways to trip the pleasure centre – and in fact, it's starting to look suspiciously as if whichever aspect of videogames you enjoy, the pleasure centre lies behind your enjoyment.

This represents the final results of the DGD2 study, which is now essentially concluded. My thanks to everyone who was involved in the surveys and case studies!


Is It I? (Terranova)

In my feeds this week was this piece by Greg L at the virtual worlds mega-blog Terranova. Here's an extract:

At a conference recently, I heard someone say that she had several avatars in several virtual worlds, including World of Warcraft and Second Life.  I was surprised, though, that she uniformly referred to her avatars in the third person.  She said things like: "It isn't a good fighter - it was exploring and it got killed by wolves."  This surprised me.  Though it is perfect logical to see your avatar as a separate thing, I still would have expected her to say "I was exploring and I got killed by wolves."  Does the usage of the third person here sound strange to you too? 

My comment (among the hundreds...):

For me personally, I use the third person to refer to videogame characters in games that give me multiple characters, such as turn-based strategy games which give me a team to deploy. In this situation, it is difficult to use "I" as you would have to pick one character to identify with. The immersion centres upon your management of the team, not on your identification with a single avatar. Perhaps players with multiple characters in multiple games would have the same perspective?

When I think back to my play of online virtual worlds, I struggle to remember how I referred to my avatar. But I imagine I said "I became a religious official and presided over several marriages after I retired from the Romulan Embassy" rather than "She retired from the Romulan Embassy". I really can't be sure though.

You can comment on Greg's post over at Terranova by clicking here. If you have any commentary solely concerned with my response, you can use the comments here as a sidebar since Terranova is so jam packed. But if you want to share your perspective on this issue, you should probably do it at Greg's original post so he can see what you think.

This is part of my new attempt to foster more dialogue with other blogs in my cluster, although since this is Terranova, I don't see it happening in this particular case.


Auctions and the Fear of Failure

Last week, the BBC reported on some research about behaviour in auctions that caught my attention. The researchers in question concluded that it was predominantly the fear of losing that drove people to overbid in auctions, and not the joy of winning as had previously been assumed. Here's an extract:

Brain scans of people taking part in an auction showed those "overbidding" had a greater response to losing than to winning, the Science journal reported... A type of scan called functional magnetic resonance imaging (fMRI) showed that in the auction game there was an exaggerated response to loss in the striatum - part of the brain associated with reward - but hardly any response to winning. The greater the tendency to overbid, the stronger the response to loss suggesting that the prospect of losing the competition caused participants to bid too high, the researchers said.

Now this interests me as a game designer and researcher because the obvious interpretation of the auction game is that the participant wants the emotional reward of fiero (triumph over adversity) from being victorious - something ebay have capitalised upon in their "shop victoriously" campaign. But instead, what we see here is that the desire to avoid being beaten by the other players (the other bidders in auction terminology) is showing up as a prevailing force.

I don't want to jump to conclusions here, because if there's one thing my company's research has shown it's that fiero-seeking isn't a majority pursuit (my best estimate at the moment is about one fifth of players have this as their primary drive, but this figure is vague at best) and this is something the researchers won't have considered since the predominant paradigm for this kind of scientific research presumes all participants can be treated as instances of the same archetype (i.e. that all humans are essentially the same) which holds far better for some traits (such as enjoying food) than for others (such as enjoying pain).

But it does beg the question: do some or all challenge-oriented players strive to beat the games they are playing because they refuse to admit defeat? Are they seeking fiero, or just striving not to lose? Are they driven by a subconscious fear of failure rather than a desire for the emotional reward of winning? That's not to say that winning isn't fun... just that it might not be the anticipation of winning that drives certain videogame players to push for the win.

The situation could be far more complicated than we previously assumed: we've talked a lot about the carrot (fiero) but not really considered the stick (fear of failure). Alas, finding a way to investigate this further is going to be especially challenging. If anyone has a spare fMRI scanner they want to loan me I'd appreciate it!


Playing Together

What’s the difference between a gamer who prefers to play multiplayer games, and a gamer who prefers to play alone? In this final look at the DGD2 survey (for now, at least – I have some major posts to come, but they will have to wait) I will very briefly examine differences between players who prefer to play alone, and those that prefer to play with other people. (Values in brackets are the statistical significance of a 2-tailed t-test – the lower the value, the more significant the finding). 

The 1,040 respondents in this survey divided more or less evenly into the two camps. 40.6% preferred to play single player games on their own, with an additional 7.1% preferring single player play, but enjoying playing such games with other people via pad passing and similar play-sharing techniques. The remaining 52% of the survey preferred a form of multiplayer gaming: multiplayer in the same room was the most popular at 17.1%, followed by virtual worlds and MMORPGs at 16.3%, and multiplayer gaming over the internet 13.6%. Finally, team or clan play over the internet represented just 5.3% of the sample.

One thing immediately stands out of the results: those who prefer multiplayer are much more focussed on challenge (and thus fiero – the emotion of triumph over adversity) than those who prefer single player. Multiplayer respondents gave much higher ratings for challenge-oriented play (.000) and both for the emotion of fiero (.009) and the fiero-enhancing emotion of anger (.009). That’s not all: multiplayer-preference players had a statistically significant higher preference for social emotions (.000) and random elements in games (.000), and a lower preference for sandbox play (.017). 

This paints a particular picture of these two kinds of players.

Multiplayer gamers (statistically speaking) tend to be challenge-oriented, and willing to be aroused to anger as this enhances their eventual reward in fiero when they attain victory. They are not only enjoying fiero, though, they are also enjoying the social element of multiplayer games such as the sense of belonging to a team, feelings of envy and gratitude, and the feeling of naches – the satisfaction of seeing someone you taught to play perform well. 

Conversely, single player gamers (statistically speaking) are showing greater interest in having control over the space of their play. This is one way to interpret the lower interest in random elements – these add variety to play, but they also mean the player has less direct control over outcomes. The higher interest in sandbox play can also be interpreted as an increased interest in having complete control over the play space, although undoubtedly other interpretations are possible.

Regarding the skills of play, multiplayer gamers rate themselves fractionally higher on basic game literacy (.001) – perhaps a sign of higher self-confidence rather than anything connected with game literacy – and (in the context of Temperament Theory) had a slightly higher mean rating for Logistical skills i.e. tolerance of repetition (.028) although this was a marginal result at best. However, the multiplayer gamers rated themselves much higher in terms of Tactical skills i.e. real time decision-making and action (.000) which is not surprising given that the most popular games to play in multiplayer all depend upon Tactical skills (first person shooters and racing games, for instance). 

We do not usually think about the split between those that prefer single player and those that prefer multiplayer games as enormously significant, but there are clearly patterns of difference to be detected. For one thing, it seems that the emotional reward of fiero may be more attractive when it is earned against (or with the assistance of) human players – beating a single player game might be less satisfying because it was not a person that was overcome. For the 36% of gamers for whom multiplayer competitive play is appealing, playing together is doubly rewarding: not only do they get the emotional benefits of social play, but the taste of victory appears to be all the more sweet when it is won from a human opponent.

This post concludes the statistical analysis of the DGD2 data for now. I may have one more analysis in the future concerning game genres, but this has not yet been conducted. Concerning the conclusions of the DGD2 study – this will have to wait!


Redefining Hardcore & Casual

The persistence of the terms “Hardcore” and “Casual” can perhaps be credited to the simplicity of the audience model implied: people find it easier to grasp an idea that divides people into two boxes than to comprehend what is implied by a more complete audience model. 

Furthermore, many people working in videogame design are resistant to audience modelling as a driver for the game design process. Usually this is because the game designer in question wants to trust their instincts as to what is fun rather than a model; there’s merit to this approach, but surely one’s instincts can be better refined by learning the available models? Earlier this year, football game expert Dino Dini contended to me that game design could be driven by intuition rather than theory; I agreed with him – it is certainly possible to design this way, and I often do rely upon my intuition – but also pointed out that whether you used theory or intuition to guide the design process, you didn’t really know what you have until you try it with players who have never seen it before. Whatever assumptions you’re making, the players are the ultimate test.

If transitioning to the widespread use of more detailed audience models (Lazzaro’s Four Keys, Bartle Types, DGD1 etc.) is hindered by a barrier caused by the relative complexity of such models, perhaps we can at least improve upon what we mean by the paper-thin model we do have in common use – the Hardcore/Casual split.
 

Game Literacy 

In August last year, I submitted a post to the Round Table which attempted to look at Hardcore and Casual from the point of view of how much experience of games the player had – in terms of game literacy. There is a lot of merit in what was proposed here, and I believe the key points still stand: the market for videogames does indeed consist of a “head” of game literate players – which I refer to as gamer hobbyists – who buy the most number of videogames, and rack up the greatest number of hours playing. And it also consists of a “tail” of less game literate players – the mass market – who are gradually in the process of replacing the gamer hobbyists as the primary source of cash flow in the games space. Nintendo’s ongoing success with their mass market friendly Wii and DS platforms emphasises this shift in the marketplace.

Despite the saliency of this viewpoint, there are flaws to the substitution of Hardcore for gamer hobbyist, and Casual for mass market. We are coming to the end of our analysis of the DGD2 survey data of 1,040 gamers (both Hardcore and Casual): as with DGD1 one way we divided players was according to whether they self-assessed as Hardcore gamers, Casual gamers, or didn’t know. This allowed us to examine the differences between the two (strictly, three) groupings. 

Most of the findings in this regard are trivial. Hardcore gamers rated themselves higher for the importance of all the emotions we inquired about (and all these findings were highly statistically significant) – which is to say, Hardcore gamers were more emotionally invested in their play, or at least more likely to rate the importance of any emotional factor in their play higher. Hardcore gamers also rated themselves higher on every aspect of game literacy or player skills in the survey (and these results were even more statistically significant). Finally, Hardcore gamers were more interested in games of challenge, structured play (Caillois’ ludus) and games of escapism (acting out in a virtual world) – all of which broadly validated the findings from the earlier DGD1 survey.

But these results obscure something interesting about the players who self-identified as Casual. Firstly, Casual players still play games very often. 81% of those who self-identified as Hardcore said they played videogames everyday, but 49% of Casual players also said they played everyday. Hardcore players gave themselves high marks in game literacy (more than 95% of Hardcore respondents claiming the top two marks, and about three quarters the very top mark), but Casual players didn’t exactly rate themselves low on this (around 85% of Casual respondents claimed the top two marks, and roughly half the very top mark). So while some of these Casual players might be mass market players, many of them are highly game literate players who play videogames every day. (Incidentally, those who were unable to choose between Hardcore and Casual looked remarkably similar to those who self-identified as Casual).

What other factors might be in play?
 

Punishing versus Forgiving

In January of this year, Corvus (our inimitable host for the Round Table discussions) shared his thoughts about what characterised Casual games (as opposed to Casual players). He characterised these games as forgiving, as shown in this extract:

Casual games are typically very forgiving games. They don’t harshly penalize failure, they have gradual increases in difficulty, they don’t demand you spend large blocks of time in one sitting. They don’t have complicated control schemes or complex mechanics. Typically, you don’t even have to read to be able to play (excepting Bookworm and its ilk, obviously). 

Corvus’ formulation of Casual games as forgiving is, I believe, a major step forward in understanding Casual players. Because without a doubt, on the basis of case studies at the very least, Casual players are looking for games that are more forgiving – and along the same lines, more welcoming. They don’t necessarily want a big time commitment (but may still spend a lot of time playing a particular game), and they certainly don’t want to be punished for their failures – they want failure to be forgiven.

This was part of the genius behind the design of PopCap’s evergreen favourite Bejewelled. It not only allowed you to excuse yourself from additional stress (by opting out of a timer – a major source of excitement in play, but also a source of unpleasant panic for certain players), but it doesn’t penalise you for making a mistake. Swap two jewels that don’t make a line and you’ll just be warned that you made a mistake – no score penalty, no penalty of any kind. This was a break from a tradition of punishment that runs throughout the history of videogames, and it found an eager audience waiting for it.

Conversely, the gamer hobbyists contain a great many players for whom the “old school” sensibilities of the arcade game and the early home videogame are more desired – games in which you are up against impossible odds, where you will fail often, and be punished for the slightest misstep. Why are these games enjoyed? Presumably because punishing for failure makes success all the more vital to strive towards and so the threat of punishment adds not only excitement to the play of the game, but it intensifies the reward in fiero (the emotion of triumph over adversity) that is received when success if finally attained. 

This, then, is the other side of the Hardcore/Casual split – not the division of the market based upon game literacy, but the division of the players according to whether they are looking for a forgiving game (one that will welcome them, and behave in a civil and friendly manner) or a punishing game (one that will raise the degree of challenge and dare the player to rise to the level of difficulty that it demands, in order to earn the maximal payout of fiero when victory is eventually attained).

There will probably be a gender influence at work here, in that the majority of female players would probably prefer a forgiving game, but it is a gross simplification to assume that this is an adequate and complete explanation. There are female players looking for punishing games, and there are plenty of male players who want a forgiving game – Animal Crossing is a quintessentially forgiving game, and its audience appears to show no obvious gender bias. The DS version has sold more than 9.5 million units, almost twice the audience that a punishing first person shooter can even hope to attract.
 

Conclusion 

The Hardcore/Casual split doesn’t work any more. It’s an incomplete description because as games have pushed deeper and wider into the demographic landscape the old assumptions don’t work any more. Hardcore might mean game literate, and it might mean seeking punishing games, but there are players who self-identify as Hardcore and yet detest any game that will make them feel angry (a feeling that enhances fiero, and can be associated with punishing games). We have no way of distinguishing between those two state of affairs in our current language.

Similarly, Casual might mean less game literate, but there are a great many players who self-identify as Casual but who are clearly well versed in the language of gameplay. And Casual might mean desiring more forgiving games, but about one in five players who self-identify as Casual still say they looking for (or willing to tolerate) anger in their play – roughly the same proportion as in Hardcore players. Once again, the term describes multiple different kinds of players, between which we cannot distinguish in our current language. 

If we want to better understand the marketplace for games, perhaps we should start thinking in terms of two very different splits. The split between game literate gamer hobbyists, and less experienced mass market players on the one hand, and players seeking punishing play (challenge-oriented, fiero-seeking players – perhaps we might call them punishers, or punishing players, that is, players seeking punishing games) and players seeking forgiving play (forgivers, or forgiving players, that is, players seeking forgiving games) on the other. Hardcore and Casual is a compromised terminology – it means too many different things, and it no longer reflects the state of the marketplace. The time has come to move forward into a new language for describing the basic splits in the audience for videogames.

Can we redefine the way we talk about the basic splits in the audience for games? Or will we be stuck with Hardcore and Casual as our only widespread terms for decades to come? Share your thoughts in the comments!


Player Skills

One of the many things we were testing in the DGD2 survey data was the validity of the skill set model from Temperament Theory. We arranged the data in order that, whether or not this was validated, we would still have some useful data about player skills from which we could derive some correlations. However, to my surprise, the Temperament Theory skill sets (Logistical, Tactical, Strategic, Diplomatic) were validated – although owing to a bug in the data gathering program, we lost most of the data on Diplomatic skills and were unable to reach any conclusions about this. 

We also gathered data about basic game literacy (things like understand how a game works without looking at a manual and moving around a 3D world using mouse and keyboard or two joysticks). 90% of the 1,040 respondents gave themselves one of the top two grades in these skills, despite the fact only 50% self-identified as a Hardcore gamer (40% self-identifying as Casual, and the remainder being unsure). This brings into doubt our system of using self-identification to separate Hardcore and Casual players, or suggests that Casual gamers are more game literate than perhaps is usually assumed.

We used exploratory factor analysis with principal component analysis as the extraction method, and varimax with Kaiser normalization as the rotation method, and a rotation converged in 3 iterations. Logistical, Tactical and Strategic groupings all held together under this analysis, as did basic game literacy, although there were signs that this could be further reduced – that Tactical and basic game literacy formed a set (possibly a bias because the 3D world controls are essentially Tactical in terms of Temperament Theory), and Logistical and Strategic also formed a set (which supports some of the claims of DGD1, in which the Conqueror archetype was shown to possess this combination of skills in varying degrees). There is probably more work that can be done here. 

Cross-tabulation confirmed that there were overlaps in the composition of these sets – which is to say, some people might always respond high, and some always respond low, but this was not the case in the data gathered.

To explore the influence of player skills, we divided the sample into pairs of groups based upon how high respondents rated themselves in the given skill sets, using an average value of 4 or higher as the split. These splits did not wholly reflect the expected proportions predicted by Temperament Theory, but the skewing effect is likely to be a result of the biases in videogame players we have already observed. Nonetheless, high Logistical skills were still more common than high Tactical, which was more common than high Strategic, all of which fits the general predictions of Temperament Theory. 

A t-test of the groups also showed that the different groupings based on skill set were statistically significant when compared to the other skill sets (and also with basic game literacy).

 

Player Skills and Game Types

A first set of data explorations compared the skill sets against measures composed to match game types, on a system inspired by Caillois’ patterns of play, but constructed to work independently of this model. Several of the measures do not correspond directly to Caillois’ patterns, and thus it should be seen as an inspiration for the construction of this part of the analysis, and not a rigidly applied formula. 

For Logistical skills, 40.1% of respondents fell into the high group. The general pattern for this high group was that they showed greater interest in all but one game type, and greater emotional response across all the emotional measures. The greatest significance values were attained for interest in challenge (Caillois’ agon), structured play (Caillois’ ludus) and escapism (interest in world-based play, playing such games just to mess around, and enjoying acting out in ways that they would never even consider in real life). This somewhat validates some of the results of DGD1, which associates Logistical skills with challenge-oriented play, but it is hard to push this conclusion too far.

There was one measure in which Logistical skills did not seem to be a factor – namely sandbox play. It seems that having high or low Logistical skills had no influence in a player’s interest in games such as The Sims or the sandbox modes of the various sim games on offer. Of course, this might be a skew in the sample of players, but the lack of statistically significant findings on this measure was interesting since every other measure was affected by Logistical skill ratings. 

For Tactical skills, 38.7% of respondents fell into the high group. The same general pattern was repeated with these groups – the only measure which did not produce a statistically significant finding was sandbox play. However, the strongest correlations were in different places – both challenge and escapism were once again two of the strongest significance values, but excitement (Caillois ilinx) and role-playing were also in this highest correlated group. This links in with predictions I have made concerning the relationship between Tactical skills, real time play, and excitement.

Finally, for Strategic skills, 30.2% of respondents fell into the high group. There was considerably more definition here – and in precisely the predicted areas. The two correlated measures were challenge, and structured play (Caillois’ ludus). This seems to validate the results of the DGD1 which links Strategic skills with precisely these two elements of play.

 

Player Skills & Emotions 

Another set of correlations that were explored were the player skills versus emotional patterns. In this part of the data, certain assumptions were used in constructing the measures which perhaps lacked strong prior experimental evidence, but in all cases the measures were not used unless they were shown to hang together convincingly.

For Logistical skills, all but one of the emotional measures were shown to be statistically significant, with the higher skill group rating higher the value of emotions to their play in every category but negative emotions, for which there was no statistical pattern. Unusually, it was fiero which came out “worst”, pulling in a statistical significance of 0.06 while all others were statistically significant at a degree which showed up as a flat 0 in the results (i.e. highly significant). 

For Tactical skills, the pattern was very similar, except in this case even negative emotions were shown to correlate (higher skills showed a higher rating of negative emotions at a 0.05 significance level). It’s very hard to know what to make of such a finding. All emotional measures had a statistical significance which showed up as 0 for the Tactical skill groups comparison, including fiero.

Finally, for Strategic skills there was one major difference: anger was not a statistically significant correlation in the comparison between the two groups (which it was for both Logistical and Tactical) – which is to say, players who rate high on Strategic skills had no greater interest in anger than those who rate lowly. Since anger serves to heighten fiero, this suggests that players with strong Strategic skills are less interested in being driven to frustration prior to victory. Although differences in the fiero measure were statistically significant between the high and low Strategic skill groups, the significance was 0.025 – like the Logistical groupings, this was the “worst” performing measure, and like the Logistical group negative emotions did not correlate at a statistically significant degree.

 

Conclusion 

The major result in this part of the DGD2 data analysis was that the skill sets described by Temperament Theory formed valid patterns in our data – a validation of Temperament Theory, which also serves as a validation of aspects of the DGD1 results.

Although players strong in Logistical skills were shown to be more obsessive than those who were weaker in this skill set (which was a predicted result), players high in Tactical and Strategic skills were also shown to produce the same pattern (which was not predicted). On the whole, it seems the higher someone rates their game abilities, the more likely they are to be an obsessive player. The same kind of pattern is found with positive emotions, social emotions, curiosity and amusement: it seems whatever a player is good at, the higher they rate themselves on those particular skills, the more they will value these kinds of emotions and behaviours. This might be an artefact of this kind of self-assessment study, rather than any kind of deeper conclusion. 

The study produces faint evidence of a connection between Tactical skills (which includes and implies competence in real time activities) and enjoyment of excitement – this was another predicted result, and perhaps warrants further exploration. The odd thing about the Tactical results is this correlation with negative emotions – suggesting that players strong in Tactical skills are consistently better at shrugging off negative feelings than those players strong in other skills. (The mean value was closer in this case to “I don’t care” than “I don’t like feeling this way”). This fits with the idea of Tactical competence being “of the moment” – focused upon the present.

Both Tactical and Logistical skills (which share in common the Sensing trait in Myers-Briggs typology) correlated with an interest in escapism, that is, acting out in a world-based environment – something that did not correlate with Strategic skills. This is another predicted result, although we might have expected this to correlate more strongly with Tactical than Logistical, which was not found.

The results in respect of players strong in Strategic skills were more or less as predicted: these are players looking for challenge in a strongly ludic game (a game with complex and detailed rules, such as a strategy game). It is interesting to note the decreased interest in anger, however – the Strategic player archetype is perhaps more dispassionate about their play; they want to win, but are perhaps less interested in having a fight to achieve this goal. A sign of a desire to dominate rather than overcome? We can only speculate. 

On the whole, the results of the DGD2 survey analysis in terms of player skills were interesting, but lacked definition. The most general conclusion we can draw concerning these results is that players who rate themselves highly on any aspect of player skills are more likely to rate highly their interest in particularly game types and emotions of play. But what is clear is that there are distinctions between Logistical, Tactical and Strategic skills – it would be interesting to discover if the same validation can be achieved for Diplomatic skills, about which we are still quite uncertain how they connect to play.

Coming soon - the remaining DGD2 analysis: Hardcore vs Casual, Single player vs Multiplayer and a mystery finale!


Female Players

What does the DGD2 survey data tell us about female players?

Well, first and foremost it tells us that female players consistently rate themselves lower than male players in terms of their gaming skills. This doesn’t mean that aren’t as skilled as their male counterparts – we have no way of assessing this from survey data – but it means that (depending upon your perspective) either female player underrate themselves or male players overrate themselves, when compared to the other gender. (A t-test showed that this finding was significant at the 0.01 alpha level, which means this finding is extremely significant in statistical terms).

We also found some interesting patterns in the self-assessment of emotions of play. Once again, the trend was for men to enter higher numbers into the survey, but there was a statistically significant pattern to the deviations. (The significance of the t-statistic is given in brackets in the rest of this piece).

In the case of emotions such as excitement and surprise, which we relate to the neurotransmitter epinephrine, men self-assessed higher than women (0.10), and in the case of the emotions such as anger and schadenfreude, which we relate to the neurotransmitter norepinephrine, men also self-assessed higher than women (0.16). Similarly, men self-assessed higher the importance of fiero (triumph over adversity) in their play (0.13). All of this collectively can be interpreted as meaning that female players are slightly less interested in gameplay resulting from the fight-or-flight response than male players, and that this distinction is statistically significant at the 0.05 alpha level.

One more finding in connection with emotions: female players rate lower the importance of curiosity to their play (0.0). This one is especially mysterious as it is the opposite of what might have been predicted. However, this may be a consequence of the wording of the curiosity question, which may have narrowed the meaning of the term artificially. Further investigation is recommended.

On the subjects of negative emotions (such as sadness, disgust, contempt, guilt, embarrassment), social emotions (gratitude, naches, envy, belonging) and amusement there was no discernible difference between genders, nor on the issue of obsessive play (which was the only emotions measure in on which women scored more highly than men, but the difference was extremely small and not statistically significant).

When examining the patterns of play respondents enjoyed, a similar trend was revealed: women, for the most part, gave lower numbers than men. This was true for games of fiero i.e. certain forms of agon (competition) and alea (chance) (0.13) and games of excitement i.e. ilinx (vertigo) (0.09), as well as games of escapism i.e. certain forms of mimicry (simulation), paidia (unstructured play) and social ilinx (e.g. “sandcastle stomping”) (0.00). Furthermore, male players rated ludus (structured play) higher than female players (0.02).

So what patterns of play did female players rate higher than male players? Well although they scored more highly on what we term role-play (various forms of mimicry) this finding was not statistically significant in this sample. However, women did rate higher than men the importance of sandbox-type play to their enjoyment (0.10), which was an expected result.

Overall, the main finding of this part of the statistical analysis is that which was introduced at the start of this piece: women consistently provide lower numbers to describe their game playing competences, the importance of emotions to their play, and their enjoyment of various patterns of play. However, it is also telling in which specific areas this finding proved statistically significant.

It is probably premature to make a statement of the kind that women are less interested in fiero or excitement than men, but the findings in respect of gender do suggest that men are more interested in gameplay generated by the fight-or-flight response than women, and that women would in general prefer to play in an apparently unstructured way, or in a form with few penalties for experimentation (e.g. sandbox games).

Remember, however, that these are trends that have been detected at a statistical level: you can’t reason from the general to the specific in this case, so what seems to be shown by a sample of 141 women tells you nothing about an individual you meet who also happens to be female e.g. the fact that statistically most women likes flowers doesn’t allow you to assume that every women you meet likes flowers – they might have a pollen allergy, or an anti-barbie complex, or they might simply not like flowers. You just can’t know anything about individuals without talking to them as individuals.

Also, it is worth nothing that there were more than 6 times as many male players in the DGD2 sample (891 vs 141) which in itself shows up a problem in gathering data about female players in an industry which has geared itself quite heavily towards making games for teenage boys.

More DGD2 number crunching soon.


Malone on Curiosity

Pf_973641curiosityposters One of the more interesting emotional behaviours associated with videogames is curiosity – that powerful drive to seek out new and interesting information. Yet there is very little written on the subject, unless you count Nicole Lazzaro’s “Easy Fun” key. Or at least – so I thought. Super-heroic Only a Game fixture zenBen (whose blog can be found here), however, has given me nothing short of three papers on the subject of curiosity in the last fortnight. I can scarcely keep up with his deep academic pockets! One of these papers we will come to shortly in another context; for today, I want to talk about two papers by Thomas W. Malone, one from 1980 and the other from 1981.

Malone was conducting research into games as tools for learning – now a very popular topic, but at the time, videogames were far from spectacularly impressive. To put this in context, the most advanced coin-op videogames at this time were Asteroids (Atari) and Pac-man (Namco/Midway). Nonetheless, Malone’s papers make for fascinating reading, and contain numerous ideas still pertinent to the games industry. In fact, what is most disturbing to me is that Malone’s papers aren’t cited more often, or indeed, required reading for game designers.

The papers are packed full of little observations which remain as poignant today as ever. For instance, in the 1980 paper Malone notes in the context of the way the game communicates success and failure to the player:

...performance feedback should be presented in a way that minimized the possibility of self-esteem damage.

This is a lesson that a staggering number of videogames have never learned! Most players are easily discouraged, and yet a macho, conqueror-style ethos is still quite prevalent, with failure being met with abuse and ridicule (even in an otherwise charming game such as Katamari Damacy – although at least in this case a touch of humour offsets the problem).

The most salient line in the 1980 paper states succinctly what should have been the mantra for the videogames industry for the past twenty five years:

If computer game designers can create many different kinds of fantasies for different kinds of people, their games are likely to have much broader appeal.

The same idea is re-iterated in the later paper:

...fantasies can be very important in creating intrinsically motivating environments but that, unless the fantasies are carefully chosen to appeal to the target audience, they may actually make the environment less interesting rather than more.

This is a claim I have been making with ever-increasing force in recent years, and it stuns me to read that someone else could make this observation back when the industry was in its infancy. How does Malone reach his conclusion? By analysing the components of a videogame and the response that players have to the game with different elements removed. He finds that the inherent fantasy of the game (the setting, or the focus of the mimicry) is the single largest factor in player’s enjoying a game – a fact that remains as valid today as it was in 1980.

The discussion of curiosity is mainly in the 1980 paper (although it is summarised in the later piece), and is presented in a pre-existing psychological framework:

Curiosity is the motivation to learn, independent of any goal-seeking or fantasy-fulfilment. Computer games can evoke a learner’s curiosity by providing environments that have an optimal level of informational complexity (Berlyne, 1965; Piaget, 1952). In other words, the environments should be neither too complicated nor too simple with respect to the learner’s existing knowledge. They should be novel and surprising, but not completely incomprehensible.

This observation ties up with recent research into a neurobiological mechanism for interest (or curiosity) by Biederman and Vessel, but we will review this work at another time when we begin to dig into the biology of play more explicitly.

Malone divides curiosity into two variants: sensory curiosity, which is about maintaining interest in the senses (and matches up with Biederman and Vessel), and cognitive curiosity, which is more about the semantic content of information. For example, one picks up a National Geographic because the photo on the cover is intriguing – this is sensory curiosity. One picks up a newspaper because of a surprising headline – this is cognitive curiosity.

The idea of sensory curiosity is not enormously explored beyond the basic statement, although there is some discussion about the work of Jerry Mander’s 1978 work on television and TV commercials in particular. The discussion here focuses on “technical events” – that is, camera cuts, zooms and other changes which apparently serve to keep the viewer’s interest solely on the level of sensory interest. I believe there is considerable more work to be conducted in exploring sensory curiosity in videogames.

On the subject of cognitive curiosity, Malone makes an interesting (although intuited and therefore essentially unsupported) claim:

...people are motivated to bring to all their cognitive structures three of the characteristics of well-formed scientific theories: completeness, consistency and parsimony. According to this theory, the way to engage learners’ curiosity is to present just enough information to make their existing knowledge seem incomplete, inconsistent, or unparsimonious.

This idea strikes me as worthy of further investigation, and even suggests something concerning the nature of science itself. Since until recent centuries “science” meant “domain of knowledge”, perhaps the element that has allowed what we now term “science” (i.e. empirical research)  to gain so much ground is that its mechanisms produce statements that are cognitively pleasing. The alternative interpretation – that we aim for cognitively balanced statements because of the influence of science – seems somehow less plausible, but there is room for inquiry in either case.

There is much to explore in the context of videogames in terms of these three conditions: each suggests a way to sustain the interest of players. By comparison, Lazzaro’s work highlights three aspects of curiosity that can be leveraged: ambiguity, incompleteness and detail. ‘Ambiguity’ seems to match Malone’s ‘inconsistency’ to some extent, ‘incompleteness’ matches ‘incompleteness’ perfectly, while Lazzaro’s ‘detail’ seems to match Malone’s sensory curiosity. Only Malone’s ‘unparsimonious’ (that is to say, ideas that violate the principle of Occam’s razor that knowledge should be succinct) seems unmatched in this comparison. I’m uncertain to what extent players are interested in parsimonious game rules, or to be more precise, while I’m certain some players are interested in developing parsimonious knowledge, it’s unclear how one leverages the absence of parsimony to provoke curiosity.

One aspect of how Malone suggests making use of player curiosity is particularly intriguing. In the 1981 paper, he includes the following bullet point under the subheading of curiosity:

Does the interface use randomness in a way that adds variety without making tools unreliable?

This matches up to our exploration of the use of luck in videogames – the landscape function that I have suggested is one of the modern expressions of Caillois’ alea (games of chance and fate) in videogames. Malone is suggesting that randomness is useful in games because it can provoke curiosity – and on examination, it seems he is on to something here. It is undeniable that the benefit of randomly generating content in a videogame is that the chance-fuelled combinations will produce something intriguing, memorable or simply bizarre. Malone even lists randomness as one of four factors most strongly correlated with a game’s popularity (the other three being explicit goals, score-keeping, and audio effects – but since he was working in 1980 it is important to remember just how crude the games used in his studies would have been).

The idea that an uncertain outcome can fuel a player’s interest is one of the most fascinating elements of the Malone papers, and suggests a link between chance and curiosity in videogames. Malone notes:

Randomness and humor, if used carefully, can also help make an environment optimally complex.

And also:

...if randomness is used in a way that makes tools unreliable it will almost certainly be frustrating rather than enjoyable.

Malone’s observations that uncertain outcomes are inherently part of the draw of videogames warrant further investigation. It seems to me that there are games where the outcome is not really uncertain – in most RPGs, you know you’re going to level up, you just don’t know how, for instance – but even in these cases there is always a level of uncertainty at work. Consider how a player who has mastered a particular game produces a new uncertainty by adding a higher level goal (in speed runs, for instance) – thus restoring uncertainty to the situation.

Perhaps in uncertainty we have a definitive link between chance and curiosity, something which will expand the emotions associated with chance in games (namely the excitement of an unknown outcome, the fiero of winning against the odds, and the sadness of failure) and potentially suggest to us a whole new avenue of exploration in videogame design.

Although the games which were the subjects of Malone’s papers have aged terribly in the intervening decades, Malone’s work has not. I heartily recommend both these papers as a fascinating and oddly fresh perspective on the play of videogames.

The opening image is a Jon Bertelli print entitled Curiosity, which is available for purchase here. As ever, no copyright infringement is intended and I will take the image down if asked.

The papers referred to in this post are as follows:

Malone, Thomas W., “What makes things fun to learn? Heuristics for designing instructional computer games”, ACM, 1980.

Malone, Thomas W., “Heuristics for designing enjoyable user interfaces: Lessons from computer games”, ACM, 1981.


Top Ten Videogame Emotions

Emotions_small What are the most popular emotions of play in videogames?

Based on the 1,040 responses to the DGD2 survey, I have ranked the top 10 emotions with their average score out of 5 to get a rough-and-ready estimate of the popularity of various emotions. This isn’t a strict scientific measure, as such, but the highest scoring emotions are those for which the majority of people not only recognised having that emotion while playing games, but recognised it enhanced their enjoyment.

(For reference, the top answer that could be given in each case was “Yes, [I recognise this emotion in my play] and I seek out games that give me this feeling” and the next highest was “Yes [I recognise this emotion in my play] and it enhances my enjoyment of a game”. The bottom answer in each case was “No, I never feel this way when playing games.”)

I have included my hypothetical deductions concerning the underlying neuro-biological mechanisms where I have some idea of what is involved.

 

10. Bliss (3.26)

At the bottom of our Top 10, the feeling of utter joyfulness, which is probably the experience of highly elevated levels of the neurotransmitter seratonin. While 27.7% of respondents said no videogame had given them this feeling, 59.9% of people gave this emotion one of the top two responses (with 22.1% actively seeking out games which give them this feeling). I’m actually quite doubtful that so many people have experienced bliss in the sense intended by emotions-expert Paul Ekman (although a study could easily determine this), and I find it more likely that people are taking the description “utter joy and bliss” to mean fiero (the emotion of triumph over adversity), which we will come to below.

 

9. Relief (3.28)

Relief, which may be the experiential analogue of the hormone cortisol, has already been acknowledged as an important emotion of play (as we discussed before in the piece on rushgames). Despite this, 21.5% of respondents said no videogame had ever given them this feeling. However, 43% said it enhanced their enjoyment of games, and 14.4% said they sought out games that gave them this feeling.

 

8. Naches (3.57)

Here’s a curious one – the emotion of pride in the accomplishments of one’s students or children, referred to by emotion researcher Ekman by the Yiddish term naches. Players seem to really enjoy training their friends and family to play games, with a whopping 53.4% saying it enhances their enjoyment, and another 12.9% saying they seek out games that give them this feeling. (I don’t have the data yet, but I wonder if such people play mostly MMORPGs?) Only 10.9% had never had the experience in the context of videogames. Perhaps, as Katherine Isbister has suggested, more videogames should include a co-operative Tutor mode?

 

7. Surprise (3.59)

Another emotion we’ve seen in the context of rushgames, surprise is closely related to fear and thus probably relates to the hormone and neurotransmitter epinephrine (adrenalin). Few people (8.1%) had never been surprised by videogames, while more than half the respondants (51.9%) said it added to their enjoyment, and another 14.4% saying they sought out games that gave them this experience.

 

6. Fiero (3.89)

Yes, arguably the most prominent of the videogame emotions, fiero (the feeling of triumph over adversity – probably a cocktail of norepinephrine, epinephrine, and dopamine) didn’t even make it half way up the top ten! It wasn’t because it wasn’t highly rated – in fact about three quarters of respondants (77.1%) gave it the top two marks, with about a third (32.7%) saying they seek out games that give them this feeling. Still, there were five other emotions that scored more highly, and three other emotions which scored higher in terms of players actively seeking out the feeling...

 

5. Curiosity (3.92)

I wasn’t surprised to see curiosity in the Top Five, but to see it edge out fiero was unexpected! Curiosity, which is an expression of what some psychologists refer to as interest (and could be seen as a behaviour rather than an emotion) seems to relate to the beta-endorphin neurotransmitter, which is involved in a mechanism encouraging animals to explore and seek new stimulus. Nicole Lazzaro was the first person to relate it to videogame play, and with good cause! It pulled in big numbers, with once again about three quarters rating it highly (78.8%) and of these about a quarter (24.3%) seeking out games that give them this feeling. Just 5.4% had never had the experience in videogames.

 

4. Excitement (4.02)

Well no surprise to see this one near the top! Excitement, as discussed previously, is an expression of epinephrine (adrenalin), and an extremely common experience – just 2.7% of respondents claimed they had never experienced it in the context of videogames. 8 out of 10 people (82.1%) gave it one of the top two responses, with about a quarter (26.3%) actively seeking it out. This emotion also produced the highest incidence of the second-to-highest response (55.8%) in the survey, that is, a strict majority of players recognise excitement as a major contribution to their enjoyment of play.

 

3. Wonderment (4.07)

Another expression of the interest mechanism mentioned under curiosity, wonderment is probably also related to beta-endorphin. Here, the feeling is more intense – and it seems players respond to the greater intensity. Whilst a larger number of people could not relate the experience to their play (8.1% had no experience of it in videogames), 41.5% said it enhanced their enjoyment and an additional 41.2% (for a total of 82.7%) said they sought out games that gave them this feeling. In fact, of all the emotions studied in this survey, this was the highest scorer in terms of respondents actively seeking it out, as even the top 2 emotions did not clear 40% in seeking out the emotion. It seems amazing players is one of the most effective techniques videogames can muster.

 

2. Contentment (4.09)

I said before the survey began that I suspected that the research community had underestimated the importance of contentment to videogames, and although this crude ranking is far from definitive, it does seem I was correct! 82.7% gave this emotion one of the top two marks, with 38.2% seeking out games that would give them a sense of contentment. Like bliss, this probably connects to serotonin, but whereas more than a quarter of players had no experience of bliss to draw upon, just 5.8% could find no memory of contentment in their play.

 

1. Amusement (4.28)

But head and shoulders above every other emotion in the survey was amusement (for which I have no biological mechanism, although psychologists link it to the resolution of inconsistencies, and it will involve an endorphin of some kind as well as the pre-frontal cortex). The fewest number of people responded that they had no experience of amusement in videogames (just 1.7%) while a whopping 92.6% gave this emotion one of the top two responses, and 39.7% stating they actively sought out this feeling (second only to Wonderment for the rate of response in the top answer).

It seems that if we want to make better games for everyone, we should be looking at how to make our games funnier, not more challenging!

 

Bottom of the List

Finally, you might be interested to know what the bottom three emotions were. At number 20, it was Sadness (2.08), at number 21, Guilt (1.91) and bottom of the barrel at number 22 was Embarrassment (1.70). In all three cases, more than half the respondents said no game had made them feel this way. Oddly, 1.1% of respondents said they actively sought out games that made them feel embarrassed – even allowing for some fatuous respondents, this is still odd. I guess it truly is different strokes for different folks!

More preliminary results from the DGD2 survey data soon. 


Gamers ♥ Stories (and other titbits)

We are about to begin the number crunching for our new player model, DGD2, and I thought I'd share a few titbits of raw data with you. Sadly, a bug has corrupted two of the data fields from the survey, but the information lost is not vital to the study, and we should be able to proceed with the wealth of data we have.

We've received 1,040 responses to the survey, of which 55% (576) are from North America, 30% (317) are from Western Europe or the UK, 5% (52) are from Australasia, and a few responses from everywhere else in the world besides.

The majority of respondents play games every day (66%), with many of the others playing every week (26%). Interestingly, of those that self-identified as "Hardcore", 81% play every day, and of those that self-identified as "Casual", 49% play every day. It seems that even people who see themselves as a Casual player are still playing amazingly often.

The most popular approach is to play alone (40%), with just a few playing single player games with pad passing or some similar group play (7%). The remaining players all prefer some kind of multiplayer format, whether in the same room (17%) or over the internet (19%, of which 5% is team or clan play), with the remaining 16% preferring virtual worlds and MMORPGs.

On the subject of game stories, there is overwhelming consensus, with 93% saying either that stories are very important to their enjoyment of videogames (36%) or that stories help them enjoy videogames (57%). A mere 5% say stories are not important, and just 1.25% say they prefer videogames without stories. Clearly, story occupies a vital space in the modern world of videogames gamers love stories!

Next week: Top 10 Videogame Emotions!