In Simulating Human Interactions I started pencilling in a wishlist for what I'd like to see in an agent framework. This blog post is going to sit down and actually reimplement the example from that post in that new system.

I will also be exercising an action/reaction system for interactions. Basically, an interaction will cause all of the parties involved to feel some sort of emotion afterward. Those emotions may be different for the parties involved. A peace officer reprimanding someone for an infraction feels anger, disgust, or some combination. The person being reprimanded can feel shame, indignation, anger, or some combination, in return.

I was working on a system based on psychology to properly map out human emotions. Unfortunately, they are still writing the book on the subject. Well, less "book" than pamphlets. There are theories, and models, and several systems to choose from. All of them have crippling limitations.

I'm not trying to say psychology is behind the curve. The human mind is just that complicated. Models developed by marketing people only have to deal with how to get people to buy stuff. Models developed for animation only how to deal with how to convince people they are watching other people. Models developed for academia are a work in progress by design.

Because I am writing a simulation, I know have the unenviable task of figuring out what level of detail is acceptable. As this is a game, and largely for entertainment purposes, I don't have to strive for hyper-realism. In fact, drama and comedy are both served by having slightly simplified characters with frequent boughts of randomness.

So my first rule is: keep it simple. If I have to choose between a giant stack of calculations with inscrutable output, or a dice roll, use a dice roll.

Second rule: there is no second rule. (When modelling people allow for inconsistencies and contradiction.)

Third rule: People don't bother reading the manual, don't bother writing one. Any system needs to provide enough feedback and have enough internal consistency that people can figure it out on their own.

Fourth rule: Magic isn't random. Magic systems provide certainty in an uncertain world.

Fifth rule: People are coming here for fantasy. Science is consistent. But it's not very entertaining.

Implementation

One model that does seem to be handy is the Valence/Arousal/Dominance model. It actually pairs nicely with my Mind/Spirit/Will system:

Mind - Soul - Will are still seats of consciousness. Valence is when Mind and Soul are stimulated. Dominance is the side effect of Mind and Will are stimulated. Arousal is when Spirit and Will are stimulated. Stimulation can also be positive, negative, cooperative or competitive. Mind and will suppressed creates submission. And most of the attributes we think of as submission are vaguely spiritual in nature. To quote Yoda:

“You will know (the good from the bad) when you are calm, at peace. Passive. A Jedi uses the Force for knowledge and defense, never for attack.”

The light side of the force, being one with the Universe, requires inaction. The Dark Side of the Force: power, requires direct action. They aren't directly opposites. They are two sides of an interference pattern.

So let us imagine that our emotions are an interference pattern from the workings of our brain. Some handy part of our brain analyzes that pattern, and unconsciously, produces facial expressions in order to clue in our fellow humans about how we are feeling. Sometimes that expression is a direct indicator of what exactly we are thinking. Sometimes it's a complex and misleading leak from a confused state. We all have a friend with a "Resting Bitch Face." He (though usually she), when distracted seems to project anger. When in fact she is simply not focused on one emotion or the other.

Marketing research shows that, at least in Western Culture, certain colors can evoke specific emotional reactions:

Color Feeling
Red Passionate, aggressive, important
Orange Playful, energetic, cheap, youthfulness
Yellow Happy, friendly, warning
Green Natural, stable, prosperous
Turquoise
Blue Serene, trustworthy, inviting
Purple Luxurious, mysterious, romantic
Pink Feminine, young, innocent
Brown Earthy, sturdy, rustic
Black Powerful, sophisticated, edgy
White Clean, virtuous, healthy
Grey Neutral, formal, gloomy
Beige (Nothing)

I will grant you that some of those sound more like traditions rather than intrinsic psychology. Purple is royal because purple dye was extremely expensive. Pink being "Feminine" is a relatively recent cultural phenomenon.

But marketing people aren't the only people who use color to describe emotional states. Psychics claim to be able to see people's Auras. Auras change depending on people's emotional state. (Ok, and they say a lot of other things, but lets leave it at that). I did my best to find one definitive link of which color goes with which state, but sadly the "science" is a bit of a mess. Here is the most internally consistent telling I could find:

Color Chakra Meaning
Red Root Passion, physicality, strength, determination, groundedness
Orange Sacral Excitement, vitality, thrill-seeking, leadership, confidence
Yellow Plexus Optimism, laughter, freedom, inspiration, creativity
Green Heart Healing, generosity, service, responsibility, love of nature
Turquoise
Blue Communication, self-expression, depth, intuition, clarity
Indigo Third Eye Wisdom, insight, sensitivity, spirituality, deep inner "knowing"
Purple crown Higher conciousness, visionary qualities, spiritual awareness
Pink Healers and Creatives
Tan Logical, analytical, focused individuals
Magenta Artists and Geniuses
White Trancendental Individuals

At first some of these combinations don't seem to make sense. But then we have to remember that colors mix in several ways.

When printing we use the CMYK model. Cyan, Magenta and Yellow are the primary colors. (K is pure black.) Red, Green and Blue are secondary colors. White is the absence of color. Black is all colors mixed.

When lighting we use the RGB model. Red, Green and Blue are the primary colors. Cyan, Magenta and Yellow are secondary colors. Black is the absence of color. White is all colors mixed.

I also demonstrate a third system: HSL. Hue is an angle on the compass. 0 is red, 120 is green, 240 is blue. Saturation of 100% means pure color. Saturation of 0% is grey. Lightness of 0% is black. Lightness of 100% is white. Pure reds, greens, and blues are found at saturation 100% and lightness 50%

I bring up HSL because we now have two additional factors to consider. Not just the original seat of the color (i.e. hue), but how "bright" the color is (i.e. lightness), and how clear is our signal (i.e. saturation.)

According to the folks who can see Auras, a distracted or ill mind tends to produce dull grey colors. On the other hand, a transcendent mind produces a pure white aura. Busy minds have auras that are mottled with different shades of bright colors.

So for our fictional game engine, let us assume that between marketing and pseudoscience emotional states can be related to a color. The color that we see is an interference pattern produced by the different parts of the psyche firing off. Our limbic system picks up that color and translates that to the emotion people read on our face. Emotion isn't the cause of thought, it's a side effect.

With such a system, I have an infinite number of emotions. In fact, I predict I will have a lot of seemingly similar emotions, but they vary in power. I will have some emotions that have clear opposites. I will have others that are vastly different, but require a different coordinate system (for instance HSL) to discern properly. People operating on a higher level of consciousness may not even see emotion, as the signal is pure white. People operating under high stress may not even see emotion, as set signal is mainly grey.

As infants we seem to emote 3 different emotional states: Happy, Distress, Neutral.

As children, the palette grows to 6: anger, disgust, fear, happiness, sadness and surprise.

As we mature, we learn more subtle shades of emotions. At this point science has preliminary evidence to support at least 27 distinct emotional states.

The problem is that conventional emotional classification systems also seem to read not just the global "feeling" of an individual, but context as well. Feelings like Schadenfreude. It's joy we feel (or think we should feel) guilty about, directed at the misfortune of another. There is a school of thought (namely mine) that says the only part that matters, from an emotional standpoint, is the Joy. The rest is context about why that Joy may or may not be appropriate. Or, perhaps at odds with another emotion: Empathy/Pity.

If you have studied psychology, you may be familiar with the Plutchik Wheel / Plutchik Dyads:

That diagram can be helpful in some cases, but it lashes itself to the wheel that there are only 3 primary emotions and 4 secondary emotions. And that every primary emotion must have an opposite. With my Hue/Saturation/Lightness based picture of emotion, we can imagine an infinite spectrum of emotions. And they don't have to fall into neat buckets. And some can simply be brighter or more confused expressions of other emotions. Anger, for instance, my be a red bordering on grey. But bright red could be self-confidence.

Gotta wrap this post up, but for next time I'll be working on a color mapping of common emotions to my Woods Wheel of Emotion.