The Five Domains Of Play: Mapping Psychology’s Five Factor Model to Game Design

[ Published in the May 2012 issue of Game Developer Magazine ]

Last year, while researching for a talk on player motivation, my PhD-in-psychology-wielding sister introduced me to the Five Factor Model (also known as the “Big 5”). The Big 5 is a system of human motivation that has come to dominate modern motivation psychology over the past twenty years—and learning it has had a profound effect on how I view players and design games to target certain kinds of players.

Over the last 12 months, I’ve been trying to draw correlations between the Big 5 motivational factors, and to game design elements that map to those factors by interviewing any gamer willing to take the test about their play behavior. In essence, I wanted to translate the work of motivation psychologists into game design—and I managed to draw a few correlations sooner than I had expected.

So I figured you guys might want to know about it, too.

Open Source Psychology

Before we get our hands dirty, I want to mention a few things about why the Big 5 is different from other systems.

For starters, the Big 5 doesn’t come from a single person—it was an international collaboration between dozens (hundreds?) of researchers. Instead of keeping the data driving the discoveries copyrighted and secret, or preventing correlative studies with other systems, the data behind the Big 5 was subjected to every imaginable cross-analysis – which is ongoing, and will be into the foreseeable future. And while the contributors could have owned the discoveries and charged for their use, they decided to release the entire thing into the public domain. In other words, the Big 5 is the Linux of motivation psychology.

Bells and Curves

When you take the Big 5 test (the best free one I’ve found: http://www.personal.psu.edu/~j5j/IPIP/), you get a report that shows you where you fall in 5 personality “domains”, as well as in the individual character “facets” that make up those domains. Each of these domains is defined as having a standard distribution (read: bell-shaped curve) when applied across humanity. A low score or a high score means that for that particular facet your motivation is a rarity, and a score in the middle  means that for that facet your motivation is similar to a majority of the population.

This means we have a statistical baseline for any analysis we’d like to do, which is a Very Good Thingas we apply this system to game development, we can swap out vague assertions like “most gamers want X,” for factual statements like “half the human population has a preference for X, and the other half by Y.” Even better, we can begin to accurately measure different player populations. Have you ever wondered whether the “core gaming” population is statistically different in their preferences than the rest of the world? Well, now we’re one detailed study of gamers away from having an answer.

Two Sides To Every Story

Each domain, or “factor”, is a two-sided spectrum with a positive motivation on each end.

This may seem obvious at first: Some people are open to new experiences, and others less so, for example. Well, compare that two-sided structure to the commonly-held-in-the-games-biz archetype of the Achievement Player. What is the positive opposite of an Achievement Player?

The developers I have worked with tend to talk about Achievers like this: If your game satisfies Achievers, you’ll probably get those players to play your game. If your game doesn’t satisfy Achievers, you’ll fail to attract those players, and by implication you make less money.

I have started calling this way of looking at players as the “thermometer model” of motivation – you stick a thermometer in the game and you measure its “achievement-ness”. High is good, low is bad. But from what I have learned so far, that view is completely wrong.

The opposite of an “achievement player” is a “contentment player”: someone who is perfectly happy to ignore your target goals, difficulty challenges, and medals, and just hang out. Someone who is motivated to be content with their current state… and who will buy games that let them act on that motivation.

Remember the bell-shaped curve discussion above? We’re talking about 50% of humanity being on the “contentment” side of the curve. That’s an awful lot of players who are not being discussed in most design meetings, thanks to a simple misunderstanding of how player archetypes work.

Swimming In The O.C.E.A.N.

Let’s get down to business. The Big 5 are: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. “O.C.E.A.N.”

Openness to Experience distinguishes creative, intellectual folk from down-to-earth, pragmatic ones. A high scorer would be Alice (in Wonderland), who is happy to drink whatever she comes across and follow rabbits into the unknown. Alice finds Wonderland a delight.  A low scorer would be Samwise Gamgee, who just wants to go home, have a predictable life, and not bother about with wizards quite so much.

Conscientiousness deals with our ability to control our impulses and order our world the way we want it. A high scorer would be Hermione Granger, who is the best in her class at almost everything, and who is the one you go to when you need to get something difficult accomplished. A low scorer would be Jeff “The Dude” Lebowski, whose primary ambition in life is to bowl, and who can turn nearly any simple outing into a near-complete disaster.

Extraversion deals with the desire for external stimulation, both social and otherwise. A high scorer would be Austin Powers, who is always ready to party, much prefers the company of others to solitude, and is the leader of the pack. A low scorer would be Edward Scissorhands, who is happy to do whatever you want, please, just leave him alone, in the dark.

Agreeableness deals with cooperation and social harmony. A high scorer would be Charles Xavier, who puts the needs of others ahead of his own, believes in the good in people, and who understands how you’re feeling better than you do. A low scorer would be Snake Plissken, who, if you want him to care about another human being (like the President), you have to inject explosives into his neck that will detonate if that person dies.

Neuroticism reflects how strongly one experiences negative (and only negative) emotions. A high scorer in Neuroticism would be Woody Allen (the character, not the man), for whom the world is a panoply of fears, anxieties, angers, and frustrations. A low scorer would be Obi-Wan Kenobi, for whom fear, anger, and jealousy lead to the dark side of the Force, and who met his death with a polite salute.

Facets Within Factors

Your score in each domain in the Big 5 is actually something like a weighted average of your score in six “facets” that that describe specific preferences within that domain. For example, Openness to Experience is comprised of Imagination, Artistic Interest, Emotionality, Adventurousness, Intellect, and Liberalism, Conscientiousness includes Self-Efficacy, Organization, Dutifulness, Achievement-Seeking, Self-Control, and Cautiousness, and so on. I won’t list them all exhaustively—take the test yourself, read the Wikipedia page (http://en.wikipedia.org/wiki/Big_Five_personality_traits), or just Google “Big 5 facets” to see them all.

As you will see, many of these facets describe polarities that game designers already use. These facets are where the proverbial rubber meets the road for game design.

The 5 Domains Of Play

With the psychology described, we can go looking for game elements that will satisfy these motivations. In order to do this, I have been interviewing gamers after they take the Big 5 test, with the idea that people with a similar score in specific facets should theoretically prefer similar games or game elements (such as PvP, achievements, grouping, and so on).

My database isn’t complete, but I have uncovered strong evidence for many direct associations between the Big 5 facets and game elements. The notable exception is Neuroticism – which refuses to produce predictable correlations so far. That said, based on the data so far, translating the Big 5 into game elements gives us these five domains of play: Novelty, Challenge, Stimulation, Harmony, and Threat.

Novelty (which maps to Openness to Experience) is the presence or lack of new, interesting, dramatic, or beautiful things in the game. A high Novelty game would be Minecraft, and a low Novelty game would be Flight Simulator.

Challenge (which maps to Conscientiousness) is the part of the game that requires the player to use self-discipline: overcoming obstacles, work, avoiding danger, and (literally) collecting achievements. A high Challenge game would be Splinter Cell, and a low Challenge game would be LEGO Star Wars.

Stimulation (which maps to Extraversion) is the part of the game that excites, be that through direct thrill-rides or through social interactions. A high Stimulation game would be Just Dance, and a low Stimulation game would be Flower.

Harmony (which maps to Agreeableness) is the part of the game where the player behaves in a particular way towards other people or characters. Do you shoot them? Or help them? A high Harmony game would be Little Big Planet, and a low Harmony game would be Street Fighter.

Threat (which maps to Neuroticism) is the negative tone of the game that can evoke negative emotions in the player, such as addiction, anxiety, anger, or sadness. As I mentioned, Threat is the domain that has so far resisted my efforts to find games that I can predict players will like, so I will save further discussion for when there’s solid data for this domain.

Mapping Facets To Game Design

Now we’re ready to correlate the individual facets in the Big 5 factors to a game’s ability to satisfy different types of player preferences. Remember: above we were measuring the player, but here we are measuring the game. The idea is that by correlating play preferences to game elements, we can predict what kind of player will like/play/buy games with those elements.

Let’s start with a simple example: just one facet of the Openness to Experience factor, which we’ve described in game terms as Novelty. In the Big 5, the facet of “Imagination”  reflects a person’s preference for their inner, imaginative world over the ‘real world’. I find (so far) that a player’s Imagination score often directly maps to their interest in fantastic/imaginative settings (such as Skyrim or Mass Effect) over realistic ones (such as Call of Duty or Madden). So, I call this facet of a game ‘World’, and describe it as the game’s “offer of fantastic or realistic settings”.

World is the first of six facets in Novelty. The other five are Predictability (offer of exploration and discovery mechanics over repetitive or “base-building” game mechanics), Melodrama (offer of emotionally evocative narratives), Artistry (offer  of compelling visuals/audio), Puzzle (offer of puzzle-solving play), and Message (offer of socially-progressive themes).

The Conscientiousness factor, is described in game terms as Challenge. Our game facets are Difficulty (offer of difficult-to-accomplish goals), Achievement (offer of accomplishment recognition, such as achievements), Order (offer of set completion mechanics, as well as grid-based play over free-board-play), Obligation (offer of guilds and other social obligation structures), Work (offer of labor-intensive tasks or ‘grinding’), and Cautiousness (offer of precise, calculated play over run-and-gun – said another way, the silenced pistol over the rocket launcher).

The third Big 5 domain is Extraversion, which for games we map to Stimulation. The game facets are Expression (offer of positive socialization opportunities – chat, emotes, etc), Crowd (offer of play with large groups of people), Role (offer of leadership roles versus follower roles), Pace (offer of a high volume of activities), Thrill (offer of high-intensity/exciting action), and Joy (offer of strong positive emotions in the player – happiness, joy, delight, etc).

Agreeableness is the fourth domain of the FFM, mapped to Harmony for a game. The game facets are Trust (offer of play that includes or does not include the capacity to be betrayed, especially in a way that feels “outside the rules”), Integrity (offer of, or the lack of, the ability to do the above to other players), Help (offer of support roles), Cooperativeness (offer of direct confrontation with other players – note that while pure PvP is a low Cooperativeness score, team-based PvP is a high one), Glory (offer of publicly-viewable medals, scores, character customizations, etc), and Compassion (offer of contexts that trigger and/or require an emotional comprehension of characters).

Neuroticism is the last doman, and while I call this domain of play Threat, the correlations are less clear. Currently, the facets I have are these: Tension (relates to the offering of player fear, such as in horror games), Provocation (relates to the opportunity to trigger player anger), Despair (relates to an offering of ‘hopeless’ contexts), Humiliation (relates to an offering that exposes player self-consciousness), Compulsion (relates to an offering of addictiveness in play), and Danger (relates to an offering where the game can actually hurt the player’s feelings). The issue here is that so far none of these facets have proven to be predictive – I have examples of players with very high and very low Anxiety scores (which maps to Tension, above) who both list Resident Evil 4 among their favorite games of all time. For now, Threat is still in ‘pre-production’.

To summarize:

  • Novelty: World, Artistry, Melodrama, Predictability, Abstraction, Message
  • Challenge: Difficulty, Order, Obligation, Achievement, Work, Caution
  • Stimulation: Expression, Crowd, Role, Pace, Thrill, Joy
  • Harmony: Trust, Integrity, Help, Competitiveness, Glory, Compassion
  • Threat: Tension, Provocation, Despair, Humiliation, Addictiveness, Danger

Now for some homework. If you have read this far, you’re clearly interested. Point your browser to http://www.personal.psu.edu/~j5j/IPIP/ and take the 300-question version of the Big 5 test. Then, with the mappings above, use those results to deconstruct what your motivations of play might be.

A note of caution: in giving the interviews, I have learned to strongly emphasize that we are trying to map preferences of play to game elements that are satisfying – not which elements players like. The idea of ‘liking’ a game element includes the player’s opinion on a lot of non-game things (how much time in the day they have for play, indie vs. AAA, etc) – and those are things that models like the Big 5 cannot predict.

How You Can Use This

The domains of play are a map to conclusions about how satisfying our games are, what motivations our games are not satisfying, what kind of players are enjoying our games, and what kind of players could be enjoying our games if we were to make specific changes.

Imagine that, for each motivation facet, every game has a “band” of that facet that it “offers”. For example, Skyrim will offer a high-to-medium Fantasy facet, which means that players with a high, average-to-high, or average score in Imagination will find Skyrim satisfying. Players with a low Imagination score will find the setting too exotic for their tastes (specifically, the existence and use of magic, in my experience). With this approach, we can map out how much “coverage” our games have for all of the motivations in the Big 5.

During game development, I often am confronted with this statement: “Players want (x),” where (x) is the speaker’s opinion on what everyone wants. These are usually inaccurate statements, and we often intuitively know this. But how do you respond?

Now, I can answer in this way: “True! Half of them do. The other half want (y),” where (y) is the motivation on the other side of whatever facet the speaker has referred to. That has changed my approach to many of my team’s design efforts. If the Big 5 system gave me only that, it would be worth it to me. But from what I can tell, that’s just the beginning.

Jason VandenBerghe is a Creative Director out of Ubisoft Montreal, which is a pretty good gig, actually. Comments, concerns, criticisms, and offers to help out can be aimed at jason.vandenberghe@ubisoft.com

6 thoughts on “The Five Domains Of Play: Mapping Psychology’s Five Factor Model to Game Design”

  1. Hi Jason,
    I’m a Game Design Student at Full Sail, with that being said, I had taken the OCEAN test and had noticed some things, this test may back up the claims for the average person’s behavior, but that is not exactly the average ‘gamer’, which may be why when I checked my score for “Neuroticism”, what you have labeled as “Threat”, I scored a 3. Realistically though, a 3 doesn’t cover any of the games you associate with that field, that I do find of interest. I recognize that it likely has a lot to do with the fact that, I myself, am not a normal person. I did not have a normal upbringing, I did not hang out with normal people, I did not live by normal cultural and societal norms, etc. Things that I thought were normal, tend to not be normal. There are quite a few inaccuracies in the test, however, what stood out the most to me, is that no one would play a game for despair, they play it in hope, hope that they can escape despair, they play because they are looking for the opposite of what they psychologically are feeling, so the results wouldn’t match up. I have been a gamer for about 30 years now. I can say gaming probably saved my life, and many other people I know. I would say all the gamers I know have something extreme in there personality. Point is, I think that the problem with threat is because the games you are assuming are for that category are actually better diagnosed elsewhere throughout multiple other categories. I would redo the bottom row entirely, some areas that pull peoples interest would be curiosity, spontaneity, surprise, atmosphere, survival instinct and nostalgia. And title it “Experiences” or something. Realistically, people aren’t playing these games for the negative experience, they are playing for the connection or tether to pull them out of that experience, or better yet, forget there current pain. In horror genre, playing something dramatic and dark that you can relate to or feel atm, allows you to normalize or interconnect with the character having that experience, or even regular people. You could be the type that wants an experience they never, which could just be curiosity. Anyway I could go on for quite awhile, and I spent too much time writing what I did already, and I have to evaluate games that fall into your older model of the 5 Domains of Play, you seem to have updated here. I see a few other things that should be changed, especially detracting and adding some questions more catered to life experiences and upbringing, if you wanted to narrow down the demographic of horror gamers and why they play what they play.
    If you read this, then thanks, hope my input helps? Remember nostalgia, experience. =)
    Natalie

    1. Hi! 🙂 Thanks for your comment.

      So, I hear in your comment the idea that I am proposing that Neuroticism is a domain that can predict player taste? To clarify: I’ve actually stated quite the opposite in all of my talks – it is clear in the data that all Neuroticism can reveal is some insight into why people *stop* playing games (not why they start). That is why all of my later work focuses on only the first four domains of the big 5. I have effectively rejected Neuroticsm as useful to game design. I would be happy to be wrong about this, but at this point all the data points in that direction.

      I also here in your comment the idea that upbringing/life experience can result in the test effectively becoming inaccurate. That isn’t really how these kinds of tests function. There is not any sort of predefined set of cultural interpretations that we must follow to interpret the tests – it’s just data about how you respond to each of these questions. There is no claim being made as to *why* the test results come out they way they do — that part is entirely provided by the human beings interpreting the results. So, if your subjective experience of the test is what you described above, great! That’s you. But there is no real ‘normal’ result for you to be different from. *Everyone* is wildly variable, as it turns out, and your struggles in extracting meaning from your results are actually quite the norm. The test is only accurate in *aggregate* — it is not a hardline crystal ball that can reveal The Truth about you, and thus the criticism that the results aren’t accurate for you are actually a normal part of the experience.

      Does that make sense?

  2. I posted this on the Gamasutra article, but can do it here too:

    I find it weird that Neuroticism is something psychologists take into account. To me, there are two polars: positive emotions and negative emotions. They will affect how we see the world and draw conclusions from it (from the book Emotions and Convictions – How Feelings Affect Our Thoughts). Someone with more negativity critiques more and draws conclusions by going into the smallest amount of aspects, where the person with positive emotions instead brush things over and take shortcuts in the decision-making. The research Yaysayers and Naysayers talk about how yaysayers meet the world with an open mind and is rewarded with adventure, where naysayers instead keep control over their world.

    A normal person takes different decisions depending on the emotions that human has at the moment. Filled with negative emotions, and you’re more closed to take in new things (control), you are more bound to see challenges everywhere, you judge with yourself as the center, and you tend to be more analytical. So negative emotions are all part of the extremes of Openness, Conscientiousness, Extraversion, and Agreeableness. The other polar, the positive emotions, are the opposite in these scales. So while the four talk about WHO you are, Neuroticism talk about HOW you do. At least this is my opinion after listening to Paul Ekman and reading Nico Frijda – leading experts on emotions and, for the later, how they affect decision-making.

    What I think is important is to realize that each one of these four differs from time to time for the same person, because they are affected by the level of Neuroticism.

  3. Hi Jason,
    I’ve been doing some work lately on player profiles, and stumbled on your page. Gotta say I love the idea of mapping game features/style to OCEAN, and especially the side effect that some features will cut off some users.
    Now, about Neuroticism. What I would say, based on my S/O’s gaming experience, is that it relates to calm vs tension/stress. She’s definitely an N+ while I’m an N-. She just can’t play any racing game, or anything slightly stressful or won’t be able to sleep afterwards.
    A nice case would be Age Of Empire, where we’d play coop. She loves the phase where she’s building her kingdom, and even attacking the computer AIs when she’s overpower. But for her to play, I have to be there, just as a safety. Even if in some cases I won’t be able to help, at least she knows she’s not alone.
    So, so far I would map according to the following:
    N+: slow rhythm, benign failure, auto-save
    N-: tension, timers, pressure, limited saves, perma-death

    This doesn’t take into account horror games. I don’t know those gamers enough to know if it’s N+ people trying to prove themselves or N- people happy to finally feel something.

    1. So, while I also have some nice examples of now specific people can map into a theory like this with Neuroticism, what I (and every other researcher who has looked at this) have found is that if you try to make predictive assertions with a large group of people, and then test them, that Neuroticism doesn’t perform that way.

      We’re not lacking ideas about how to make the map work 🙂 The problem appears to be that everyone’s relationship to their own ‘negative’ emotions is highly volatile and often very unique–and that makes the scores useless for predictive assertions. 🙁

      That said, they absolutely do provide a lens into why people _quit_. This has been validated through studies with several online games–the reasons that people wil_ quit_ playing your game can be predicted reasonably well by Neuroticism. But, alas, we can’t say anything definitive about what kinds of games people might _like_.

      Does that make sense?

  4. If we can already use it to predict players who’ll quit and why, it’s already much better than I thought. I was under the impression you were mostly leaving it aside because of unreliability.

Leave a Reply