How to explain utility function from personality, beliefs and ongoing emotions?

Question

Economists widely use utility function (and more broader approach called hedonistic regression) but are there explanations or derivations of utility function from the perspective of cognitive sciences or psychology which can take into account personality, beliefs and experience (memory) and ongoing emotions (affects)? Clearly one should start with the Maslov pyramid that determines goals and needs and then one can proceed to explain why different people value differently one and the same goods or services? Are there such research trends, some keywords, terms, notable researchers in this area?

I have read only about incorporation of altruism in the utility function of neoclassical economics, about hypberbolic discount function and so on. But I have not managed to find more practical explanation of utility function.

This question has the following context: I am trying to model the economic behavior or BDI agent - agent with Beliefs/Desires/Intentions. BDI is the most popular approach for modelling cognitive agents. Recent research (e.g. http://people.idsia.ch/~steunebrink/) has endowed BDI agents with formalization of emotions (KARO logic framework) and there is scarce but ongoing work to endow BDI agents with personality (e.g IEEE Transactions on Affective Computing and Five Factor Model from personality psychology). So, BDI agents have cognition, personality and affects and they can be used as models of human economic behavior. But there is missing link from economics and cognitive sciences - how humans act as economic agents? What are the models? What goals they can have, how behavior (decision to make purchase) can be infered from agent's goals, beliefs, personality and emotions? How human perception of utility (of particular attributes of particular products, particular designs, particular type of entertainment, etc.) is connected with his personality, experience, beliefs, affects? Utility function seems to be promising proxy in this research - psychology can explain utility function and economics can apply this function to particular transactions.

Is there some research along this lines?

This should be promising area of research. Because common approach to economic analysis and prediction is connected with Data Science and statistical methods (conjoint analysis and hedonic regression). What I am trying to do is to find logic, semantic, cognitive methods of explanation and prediction of customer behavior. Similar trend - from statistical to semantic methods can be found in other fields of Science as well - e.g. in natural language processing, in search engine engineering and so on.

Answer 1

How to integrate realistic cognitive models of human behaviour and ecomonic models is an open research question, however to tackle it, it might be helpful to break it into a few more easily answerable sub-questions:

• What models of human planning with emotional influence exist? There are a number of models in existence that try to capture emotion and personality. However, I don't know any that apply to planning specifically, given that modelling long-term planning is an extremely hard problem in it's own right! My lab-mate Peter Blouw made a rudimentary model of planning with Bryan Tripp that just got accepted to the CogSci2016 Conference. I will update this answer with the link once it is public ally available.
• What models of human social interactions integrate personality? Human social interactions are really hard to model, especially at a large scale, consequently most models that currently exist are rudimentary at best.
• What is the most detailed cognitive model of human behaviour used in economics? I don't know the answer to this off the top of my head, by I remember this being discussed in Khaneman's "Think Fast, Thinking Slow", where he talks about the diminishing returns that come with including more complicated cognitive factors in economic models.

To get more complete answers to these questions, feel free to ask them as separate questions on this site. I will happily upvote them!

Answer 2

The behaviour of agents can be predicted from utility or value based decision analysis models. Basically, you can model how the agent decides based on multiple objectives (or goals) by means of a multicriteria decision model. A simple multicriteria model consists in the weighted aggregation of value (utility) functions. Value (utility) functions represent how the agent values distinct levels of consequences on each objective.

Multicriteria decision analysis (MCDA) based on utility or value measurement are the fields that you should explore to construct such a model. MCDA has a sound foundational basis and it is the prescritive (normative + descritive) approach to make sound decisions.

A rational agent (decision-maker) faced with alternative courses of actions from which it must select one, it should always select the action with the highest overall expected value (or utility) as computed from the referred multicriteria decision model.

The distinct intrinsic beliefs, personalities, emotions, etc. are modeled by distinct value (or utility) functions and weights; so that each agent (decision-maker) behaves according to a specifc multicriteria model.