The goal of this portion of the project is to devise a personality model that meets the following criteria:
1. The personality traits are close to orthogonal.
2. They span the vector space of personality reasonably well.
3. They are operationally useful.
4.They are easily understood and applied by users.
The Three-Trait Standard Model
The
model that I have settled on after many years of trial and error
consists of three traits along with two extensions to each trait. The
three traits are Bad_Good, Faithless_Honest, and Timid_Dominant. The two
extensions are the perceived values and the accordance values for each
trait. The current nomenclature uses “p” as a prefix to indicate the
perceived value of the trait, and “a” as a prefix to indicate the
accordance value of the trait.
This model meets all four of the requirements listed above. I see no justification for removing or replacing anything in this model; at this point, the question to explore is the addition of other traits or other extensions.
Bad_Good: This meets all four criteria very well. Its p-value corresponds nicely to affection or love; it or its p-Value are frequently used in deciding a character’s inclination towards various options. Its a-value corresponds nicely to a character’s general friendliness.
Faithless_Honest: This trait is orthogonal to the others, and spans an important dimension of behavior, and is easily understood, but is not used often. Its application requires the inclusion of verbs that permit deceitful behavior; such verbs in turn get us into many difficulties involving the determination of truth.
Timid_Dominant: This trait overlaps somewhat with concepts of pride and humility. It is intermediate in utility between Bad_Good and Faithless_Honest, but is particularly confusing in its p-value and a-value. A character’s Timid_Dominant value correlates closely with their pTimid_Dominant value towards others, as well as their aTimid_Dominant value. This is because Timid_Dominant is intrinsically about a character’s relationship with others, so the p-value overlaps with the intrinsic value. [Question for discussion: should the nomenclature attach the prefix “i” to the intrinsic value of a trait to parallel the use of “p” and “a” as prefixes?]
Additional Traits
A number of additional traits have been proposed:
Ascetic_Hedonistic: This trait specifies the sensuality of a character. A high value denotes a character who enjoys good food, likes to dress up, is lustful, and enjoys entertainment. A person with a negative value of this trait cares not for any of these things. This trait meets criteria 1, 2, and 4 well; however, its operational utility is not as great as the previous three; it will be useful only in situations in which the character is tempted by a possible benefit. Moreover, its p-value does not denote any kind of relationship as with the first three. Sexual attractiveness, for example, does not in any fashion correlate with Ascetic_Hedonistic, so pAscetic_Hedonistic has no utility, nor does aAscetic_Hedonistic.
Stupid_Smart: This seems an obvious personality trait, but it falls flat on criterion #3; there just aren’t many behavioral options that would be decided by the intelligence or lack thereof of the character, because dramatic situations don’t hinge on intelligence.
There are a wide variety of personality traits commonly used in role-playing games, such as the old standards charisma, dexterity, strength, stamina, and so forth. These have no dramatic significance and are therefore useless for our needs.
We can work backwards from various relationship types to infer intrinsic traits. One such trait would be associated with the concept of respect or admiration. The problem here is that affection is not orthogonal to admiration; there’s considerable overlap. This is most easily perceived by asking, “Is it possible to like a person you don’t admire?” And “Is it possible to hate a person you admire?” Since the answers to both questions are mostly negative, we conclude that admiration is not independent of affection and therefore fails Criterion #1.
Cowardly_Brave: This trait is not completely orthogonal to Bad_Good, but is distinct enough to deserve consideration. It fails, however, on Criterion #3: there won’t be many verbs that require courage to opt for.
Could any of the Big Five (or “OCEAN”) personality traits used by psychologists be adapted for use in our personality models? Here they are:
Openness to experience (inventive/curious vs. consistent/cautious)
Conscientiousness (efficient/organized vs. easy-going/careless)
Extraversion (outgoing/energetic vs. solitary/reserved)
Agreeableness (friendly/compassionate vs. challenging/detached)
Neuroticism (sensitive/nervous vs. secure/confident)
If you apply Criterion #3 to each of these personality traits, you’ll immediately see the problem: none of them will be of much use in determining how a character will behave in a dramatic situation. That’s because dramatic situations consist almost exclusively of interactions with other characters, and few of these traits are directly pertinent to such interactions. Yes, they can be made to fit, but it’s a stretch. It’s better to have personality traits that can be directly and obviously applied to a character’s decisions.
Moods
A
somewhat different extension of personality traits comes from moods.
These are temporary emotional states. In the Storytron technology, I had
four moods:
Sad_Happy
Disgusted_Aroused
Fearful_Angry
Tired_Energetic
These were rarely used, but perhaps this reflected the complexity of the system. I am tempted to suggest that we create an extension of the existing personality traits for moods. mBad_Good would nicely correspond to Sad_Happy, and mTimid_Dominant corresponds to Fearful_Angry, but mFaithless_Honest doesn’t correspond to anything. mAscetic_Hedonistic might correspond to Disgusted_Aroused. Tired_Energetic would require a personality trait like Weak_Strong, which we already know is of little dramatic utility.
All in all, I think it best to dispense with moods at the outset of this project.