17
\$\begingroup\$

Specifically, how difficult would it be for an indie developer to implement a system similar to the MBTI so that the player character's dialogue, actions, and choices will change, according to the assigned MBTI type?

This may be limited to the snarkiness of the player character's responses, the character being more reluctant to go to a social event, how well the character will respond to criticism, etc.

It would help to reference games that have done this or something similar.

EDIT: For those that don't know, the Meyers Briggs Type Indicator is a way of measuring someone's personality based on 5 variables, (the last of which is usually ignored) in order to get 16 (or 32, including the last variable) different archetypes.

Extraversion vs. Introversion determines if you 'regain energy' by interacting with others, being the centre of attention, etc. or by being alone, doing such activities as walking, reading, and introspection.

Sensing vs. iNtuition determines if you prefer to absorb new information by focusing on facts, statistics, and details, preferring ideas with practical applications, or imagining the possibilities, exploring concepts for their own sake, and looking at deeper meaning.

Thinking vs. Feeling determines if you make decisions using logic, value justice, enjoy finding flaws and playing devil's advocate, or base decisions on personal values, and are more about how things should be.

Judging vs. Perceiving determines if you use T/F attributes to interact with the world around you, or S/I attributes.

Perception tends to make people want to keep their options open, and make external dealing decisions on the spot, rather than in advance, while Judging tends to be linked to making a plan and sticking with it.

Then, there's Assertive vs. Turbulent, which determines confidence in decisions. Assertive individuals supposedly are less affected by stress, which has the side effect of it not affecting their performance that much, whereas Turbulence is related to perfectionism, completionism, and panicking.

People cannot be fitted neatly into 32 different categories, as many people are near the centre for some of the variables, and it is therefore better to consider these archetypes than actual categories that everyone falls into. If as well as left and right there was also a 'middle ground' classification, instead of 32 personalities, there would be 243.

The personality system in question would not have to be the Meyers Briggs, and indeed it probably should deviate from it in a few areas for copyright reasons. The purpose of giving the PC an in-game personality is so that the player can relate to them, not so they can emulate the player, so the Forer effect actually works in the Game Designer's favour.

How feasible is it, therefore, to integrate a system such as this into a game such as an RPG?

\$\endgroup\$
  • 4
    \$\begingroup\$ I noticed that you tagged this question with ai. Neither DMGregory's nor my answer contain anything ai-related, though. That's because your question only seems to ask about the player-character, while in the context of game development, "ai" is usually what controls non-player characters. Did you have any specific intention when you picked that tag which is not reflected well by the answers you got? \$\endgroup\$ – Philipp Aug 28 '17 at 15:20
  • \$\begingroup\$ @Philipp I chose AI because the player character is a distinct entity from the player themselves, particularly in the game I'm envisioning, the majority of whom's actions will be decided according to the character rather than the player, or else it would get tedious. The player's input near the beginning decide the traits that the player character displays, and the player's own decisions only occur at the more 'important' moments. The personality system used might be loosely based around the MBTI, not for emulation purposes, but relatability. \$\endgroup\$ – Piomicron Aug 28 '17 at 19:59
  • \$\begingroup\$ @Piomicron An idea I had when playing a game with multiple dialogue options; have the personality type affect what options show up first, with a last "none of the above" option leading to a choice between the rest, with no way of going back to the original ones. It would keep all possibilities open but represent what options come to mind intuitively. \$\endgroup\$ – JollyJoker Aug 29 '17 at 10:23
  • 1
    \$\begingroup\$ The Ultima game series, notably Ultima 7, made your character start out with a personality quiz which affected how you would start out in the game. See tk421.net/ultima for the eight virtues. \$\endgroup\$ – Adder Aug 29 '17 at 10:38
  • 2
    \$\begingroup\$ @Piomicron as the asker of the question it's your responsibility to make it clear to potential answerers, and also to others who might benefit from it. Telling someone to look up the meaning of part of your question is arrogant and rude. \$\endgroup\$ – barbecue Aug 29 '17 at 14:01
35
\$\begingroup\$

Firstly, you should know that the Myers-Briggs Type Indicator is a proprietary system - if you wanted to use exactly this system in your game you'd need to license it from the foundation that commercializes it. (It's also, in terms of scientific validity, kind of bunk - it's popular mainly due to effective promotion and application in flashy motivation programs, rather than for any accuracy or even internal consistency).

Other, open/non-commercial, and better-vetted models of personality exist, with some already being applied in game development.

The broader idea of establishing a personality type for the player / their character through early choices in the game and using that to steer future outcomes is certainly viable.

In Interactive Fiction games and those that focus on branching narrative, this is called the "sorting hat" choice structure, with the initial choices often referred to as a "personality quiz," establishing internal variables called "personality stats" to control branching down the line. Choice of Games' style and their ChoiceScript format emphasize this use of stats.

Tooling around the IF links above, you'll find tons of advice for choosing effective personality stats, setting them, and using them for interesting branching structures and variability in your game.

Silent Hill: Shattered Memories is one example I know of a larger-scope game using this approach. There, the player goes through a literal psychological exam, answering a therapist's questions and filling in a questionnaire, which ultimately determines what they'll encounter later in the game. Morality & alignment systems common in RPGs like Neverwinter Nights could also be considered a simple form of personality metric, with actions your character takes shifting their position on a scale that can be used to selectively offer / hide particular dialogue & quest options or change NPC reactions toward you.

The MBTI example illustrates a helpful bias we have as humans, called the Forer effect: when we are told that a particular outcome was tailored specifically for us based on our individual responses, we tend to read into it the things we identify with, and subjectively rate it as much more personally accurate than if we'd seen exactly the same outcome but without this framing. So, you don't need to get carried away with extremely granular choices and thousands of varied outcomes - as long as you do a good job conveying the idea of an individually personalized character, your underlying model can be relatively simplistic and still deliver a compelling positive player experience.

\$\endgroup\$
  • 4
    \$\begingroup\$ It's true that the Big 5 is very popular in research, but it's far less useful than Myers-Briggs, or most other personality systems. The traits are not even remotely orthogonal, which is one of the most important features of a comprehensive personality theory. Without orthogonal traits, it's merely a system of measuring a couple of traits. In other words, it's a vague psychological measurement rather than a personality theory or system. \$\endgroup\$ – Attackfarm Aug 28 '17 at 15:22
  • 3
    \$\begingroup\$ The Forer effect is a very specific cognitive bias that is not applicable to the MBTI. It explains systems such as astrology, where the descriptions are vague, almost entirely positive, and often different according to the author. The MBTI is specific, features value-neutral traits, and are consistent regardless of the author. The Myers-Briggs system is not due and cannot be due to the Forer effect. \$\endgroup\$ – Attackfarm Aug 28 '17 at 15:23
  • 2
    \$\begingroup\$ @Attackfarm Psychology is not an exact science. Any model is hard to prove or disprove, so one can argue endlessly about how accurate they are. I am not a psychologist, but I would argue that the human psyche is far too complex to press it into any classification system which can be summarized with a single infographic. \$\endgroup\$ – Philipp Aug 28 '17 at 15:29
  • 1
    \$\begingroup\$ @DMGregory With that we agree. Though, I took the OP's question less literally than you, assuming that "similar to" meant merely a personality system loosely based on the MB system but tailored enough to work within the specific game at hand. It's totally possible that you're right and the OP meant to use the MBTI peg and cram it into a square hole. \$\endgroup\$ – Attackfarm Aug 28 '17 at 16:51
  • 2
    \$\begingroup\$ Just remember not to buffer underflow your aggression factor or Ghandi will nuke you. \$\endgroup\$ – corsiKa Aug 28 '17 at 23:20
3
\$\begingroup\$

As DMGregory pointed out, the Myers-Briggs Type Indicator system has some problems both legally and scientifically. But my answer should be applicable to any personality system which uses two or more axis' to categorize people. Whether or not such systems have merit would be a topic for cogsci.SE. But we game developers always oversimplify complex real-world systems into simplified models so we can build comprehensible game mechanics around them. So let's roll with it.


To make the players MBTI category affect their responses, you simply have a set of responses for each situation and a bunch of if-statements to decide which one to use. It might be difficult to come up with 16 different phrases for every possible situation, so in some situation you might only decide based on one or two of the four different cognitive learning styles mentioned by Myers and Briggs.

However, keep in mind that most games are about making decisions during the game, not making one decision at the beginning and then seeing how it plays out. That's why many RPGs which have some character personality mechanic use a different approach:

  • Don't start with a defined character personality. Form it over the course of the game, based on the player's choices during decisions. For example, when the player chooses the answer "Would you like to join me?" they get +1 Extroverted. When the say "I prefer to go alone" they get +1 Introverted. Those scores then later affect their character's abilities and/or the story of the game.
  • Have the character's personality affect what options they have over the course of the game. For example, when the player picked the character type "INTJ" and they have to make a decision, you give the player one option which represents introversion, one which represents intuition, one which represents thinking and one which represents judging. That means you need at most 8 options per decision.

Pillars of Eternity combines those two mechanics. Choosing personality-specific dialog options gives you points in that personality-direction, and having a lot of those points later in the game unlocks special dialog choices which might have more beneficial (or at least more interesting) results than the neutral options.

Now the hard part is dealing with the explosive combination of so many options. If you have 4 possible options per decision, and you let the player make 10 such choices, you end up with 4 to the power of 10 = over a million possible combinations. The number of possible outcomes increases exponentially with the number of decisions to make. Writing an unique story for each leaf of this huge decision tree is impossible. To keep the development scope bearable, you have to prune some of these branches. These are some techniques you can use:

  • Localize the effect of decisions. Have a choice affect one future event, and then never mention it again.
  • Fake it. Have multiple options result in the same effect. A decision might have 3 options which turn out as "yes" and 5 which turn out as "no" and then only take into account if the player picked a "yes" or a "no" option. You can even have decisions which do not actually affect the story at all and serve no purpose but to give the player the illusion of a meaningful decision. Bethesda does this a lot in the 3d Fallout Games. Whenever the player-character is about to say anything, the player has to pick between exactly 4 options. But more often than not they do not even affect the next sentence by the NPC (and sometimes not even the sentence spoken by the PC).
  • Aggregate decisions. Use a points system where points are added or subtracted based on choices during multiple decision and then later branch the story based on the total score. This can be how much the player is following a specific personality archetype, but can also be entirely separated from it, like how much an NPC likes the PC.
\$\endgroup\$
  • \$\begingroup\$ It's very rare to make the story branch out in sixteen different directions at every choice. I would expect most branching out to be short term, with maybe two different paths the narrative could go down, keeping track of certain variables. Most of the time, not all the aspects of your personality are going to come into play, and I would expect the choices to have similar effects no matter what your personality is (for the most part), the requirements to choose those being different. For example: in order to object to an action, whether it must 'feel' wrong, or not add up. \$\endgroup\$ – Piomicron Aug 29 '17 at 8:51
0
\$\begingroup\$

Instead of issuing a test, why not present a chart of stereotypical characters to choose from, and let the user play as that character with that personality set? It can be presented like these charts:

https://www.google.com/search?tbm=isch&q=mbti%20characters&tbs=imgo:1#imgrc=_

With your characters, or it can have several well-known archetypes for each category. However I would recommend the 4 major functions to keep it simple: SJ, SP, NT, NF.

I've created psychometric personality tests before (specifically for Big5). If you really want a test, for MBTI, you will need 4 scales (E/I, N/S, T/F, P/J) and a bunch of questions which measure a score on that scale. For example, 10 questions for E/I like "Do you like parties?"; "Do you get energy from large crowds?"; etc. Add up the scores, 1=YES, 0=NO, and you will get a 0-10 sum for each scale. If they score 7, then round to Extrovert category (5+). That is how to create a test. (If you need more stringent validity you can mix in true/false questions that everyone answers the same to filter random choice, and you can randomize the questions also.)

\$\endgroup\$
  • \$\begingroup\$ Interesting idea, however the idea is to give the PC as much customisation as possible, limiting them to 16 premade characters would be arguably somewhat worse than having a relatively 'blank slate' character. I know how the Meyers Briggs thing works, my question was about how feasible it would be to integrate it into a game, so that the Player can project themselves on them, and if it would work. \$\endgroup\$ – Piomicron Aug 29 '17 at 8:38
  • \$\begingroup\$ @Piomicron But what if you have 16 archetypes, and through whatever personallity test you put your PC in one of the 16. it's all about the delivery. \$\endgroup\$ – Brian H. Aug 29 '17 at 9:22
  • \$\begingroup\$ @Brian H: The other aspects of the player will also be customisable. If you're taking that into account, then how different is it from just determining which of the 16 personalities through a quiz? \$\endgroup\$ – Piomicron Aug 29 '17 at 13:58
  • \$\begingroup\$ which other aspects? it may sound like a bad thing, but the "illusion of choice" is a very powerful tool for any designer if used properly, and having 16 different interaction branches seems like plenty to me. \$\endgroup\$ – Brian H. Aug 30 '17 at 9:59
  • \$\begingroup\$ @BrianH things like appearance, aversion to blood, voice, etc. \$\endgroup\$ – Piomicron Sep 6 '17 at 7:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.