Agents with Personalities

Abstract

The creation of game characters with realistic behavior is an ongoing concern in the gaming and serious game industries. One way to achieve this is to find a way in which agent like characters can display human-like personality traits. The MBTI (Myers-Briggs Type Indicator) classification offers a way to classify human personalities using 16 types (e.g. INTP - Intraverted, iNtuitive, Thinking, Perceiving; ESFP - Extroverted, Sensing, Feeling, Perceiving; etc.) and 8 cognitive modes (NI - Intraverted iNtuition, SE - Extraverted Sensing, etc.). Each of the types consists of a combination of the 8 eight modes in a specific order. Although in psychology the personality types are considered descriptive and not prescriptive when they are studied in relation to job recommendations, for example, when this theory is adapted to autonomous agents, there is no distinction between these two approaches. In psychology, the specific order of the cognitive modes influences the behavior. In order to create a software model of this psychological theory, the importance of the cognitive modes in determining behaviors needs to be quantified. A quantified model is proposed and tested against pre-existing data (consisting of a list of job recommendations made for each of the 16 types). After it was determined that the proposed representation is valid, it was used to model the behavior of game characters in a game programmed in Unity using C#. By developing the game, I have shown that the BDI paradigm, specific to agent- oriented programming can be ported to an object- oriented language as C# and combined with psychological elements, in order to create game characters with more realistic behavior.