CSC594 Mini Personality Content Theory
Professor Clark Elliott

Overview:

Abbreviation: MBP = Myers-Briggs Personality Model

Scope: 1,500-3,000+ words.

Your job is to work out an original, computable, content theory of some aspect of personality based on the MBP scale (or a similar one—if you don't use the MBP, be sure to document the one that you do use). You must always ask, for each part of your theory: Is this computable? That is, are you describing objects, lists, categories, relationships, and etc. that can be represented in databases, data structures, program code objects, and can these data structures be manipulated by methods and procedures by following computer algorithms?

Examples:

Not [currently] computable, and there is zero chance you could write the program to execute it this quarter if asked to do so: Read the text of what someone says in a story and extract their personality type from it.

Note that the basic theory we start with is already quite algorithmic in nature: 4 dimensions, each with 2 possible values, giving us 16 personality types.

The task:

You might additionally have a (possibly major) component describing how such a computable theory of personality can be used with existing systems (such as tutoring systems, or helpdesk robots).

This is a "Blue Sky" assignment, in the early phases of development: Grounded creativity rules, along with common sense and general algorithmic tools. You need not justify the various steps of your proposed theory with references to the literature, though of course you are free to do so. In essence we are doing the good work of building something elegant, and later we'll decide how much we like it.

Furthermore, this is basic research. You don't need to justify why you are working on a particular style of content theory. In essence, we will design it first, and maybe build a running prototype of it, then later decide if we have a use for it, or some similar work that this version inspires.

Clear language and an organized exposition are required. You can use any form you like (e.g., formal prose, a hierarchical chart, a set of well-formed lists, etc.), but you must always be asking the questions: am I communicating my ideas clearly? Could a computer programmer follow my specifications to write a design, pseudo code and then a program to implement my work?

Some suggested content:

Ends of the bell curve:

1. Develop a very broad theory that covers all sixteen types at a surface level, with large granularity.
2. Develop a detailed theory working out the details of one of the scales, such as, e.g., thinking-feeling, with extremely fine granularity.

And then there is everything in between:

• (Note: for any of the following, you may wish to additionally constrain your work to a certain domain of knowledge or interaction, such as teaching about psychology for automated tutors, or non-player game characters only for first-person shooter games.)
• How MBP affects appraisal—that is, how do different personalities color the things we care about and how we see the world?
• How it affects expression (e.g., extend the work on the model for the AR temperament). How would these different MBP personalities affect how an AI agent expresses any emotions it has?
• What is the relationship between the types of personalitise described and the types of emotions they might favor?
• Social pros and cons of having a particular personality. (Computable!)
• Which types of personalities you would want for different jobs in a software development group?
• How do we make the descriptions of the eight personality elements (Extraversion, Introversion, Sensing, Intuition, etc.) computable?
• What are the if-then rules that apply to some aspect of each personality (with pattern variables)? [See the Rules / production systems lecture.] For example, give us 20 concrete situations and show how [some subset of?] the different personalities would react to them.
• For an automated tutoring system, how might pairs of personalities best work together? That is, if you are able to diagnose a personality for the student, what sort of complementary personality would you want to pick for the automated tutor? What is the rudimentary theory behind your choices? Even if we got the exact choices wrong for the moment, how would such a computational system that makes use of them work?
• (As precisely as possible) describe how such a personality model, added to software agents, might be useful for existing software, such as automated mentors or guides, or non-player game characters, or simple, but interesting interactive AI web personalities. (In the last case, please, no fantasy about how the AI agent will magically understand language.)
• (As precisely as possible) describe how such a personality model might allow software agents—such as automated mentors or guides—to adapt themselves after making predictions about the personality of the user with whom they are interacting (affective user modeling).
• How such personalities might help with AI generation of humor.
• People have personalities, and, for safety, we learn to detect them when driving, so we can make predictions about what a particular driver might do next. We might argue that self-driving cars that have no personalities might be confusing and unsafe. What set of personalities might be important for self-driving cars? What "car actions" might our own self-driving car use when making predicitons about the personality of the other driver?
• Reduce to 4 useful personality types (or 6, or?). Justify your choices as being correct for a particular domain (e.g., self-driving cars).
• What secondary personalities might we use to extend the theory? For example, we might need a primary type of the The Architect, combined with a secondary type of The Thinker, and so on. (Lots of room for work here: for example why would we want such added complexity? Where would it be useful?)
• If you decided on developing primary/secondary personality types, what would be the best such complex types? The essential such types? The useless types?
• What to expect of groups of personalities that have gathered together. What happens to the group when certain kinds of personalities join / leave?
• Describe the steps in a simple story (such as a folk tale, or a news story), then alter the details of the story by infusing the characters with different personality types. Computable please!
• Develop or borrow a set of major themes that categorize stories. Build a content theory of which personality types inhabit these themes. (E.g., judging vs. feeling might be important components of stories that have right and wrong as a theme, vs love as a theme.)
• Lists of phrases associated with descriptions of the different personality types in different (very specific!) domain contexts (e.g., how they speak, how they are described by others, within the domain of playing board games) Note: this is VERY hard to get anything useful. Once we introduce words, there is a tendency to assume computers understand the words and phrases, which they don't.
• ANY OTHER ASPECT of this personality domain that is of interest to you, and is computable.

Or, another example of ends of the bell curve:

1. A broad, shallow overview of how all pairs of personalities interact with one another (225 unidirectional relationships!)
2. An extremely detailed working out of how one pair of personalities interact with one another (e.g., Inspector / Crafter).

And then there is everything in between:

• How a select set of important pairs of personalities interact with one another.
• How can you label relationships in a meaningful way? How do you measure relationships between personalities?
• etc.

Throw-away implementation utilities:

Deliverables:

• Place your [ongoing?] work in a Google Document, shared for comments with the class group. WRITE YOUR NAME at the top of the document, always, and possibly a version number. Attempt to preserve the comments of others.
• We will add a new section to your Master Root Document: Right after your name, place a link to your new document labeled: Mini Personality Content Theory.
• Write a concise one-paragraph summary of what you have done in your mini content theory.
• Select a unique three-character code based on your initials and wrapped inside of parentheses such as (CDE), (TS2), (MCH), (1JK) and so on. I will search for this string to give you credit when grading. Use this same code throughout the quarter to identify your work.
• Once the work has begun to stabilize, write helpful or encouraging comments on the (possibly still developing) mini personality content theory for at least three other students (though more than three is always excellent!). Be sure to include your special, unique, three-character identifier in each comment you make.
• Write the names of the students on whose work you commented after your summary.
• As long as your work is your own (!) it is fine to get new ideas from what others are doing to help with your own content theory. For example, you may be working in a different area, but you like the way they describe categories. Cite everyone by name, if you have been inspired by their ideas.
• Use the D2L forums to discuss your ideas and your work on your content theory.