I didn't find this model particularly compelling compared to the many that have been developed in the psychological and therapeutic literature, or in the esoteric spiritual traditions.<p><i>However</i> what I think <i>is</i> valuable, and is what the author is experiencing, is that thinking about emotion and how it works, and developing our own model of it that makes sense of your own experience, is a profoundly helpful thing to do. Developing your own personal model is fundamentally different from learning someone else's, wherever it comes from.<p>So, whether this model does or doesn't work for you, it may be worth figuring out why, and making your own better one.
> ...Bob may overestimate the probability that Alice lied intentionally and thus would experience more anger, because he will compute more possible outcomes where he gets hurt by Alice.<p>The author's theory kind of sounds like a "Parallel Universes / Many-Worlds Theory" model for psychology.<p>While it can be useful in many ways, the model itself might not be based on reality (i.g., how the brain works), and in the example - it might just be that the ego needs control, and when that control is shown to be false, it reacts the best way that particular brain's ego can...<p>A confident person would just blow what happened off. A stressed person might get angry. Etc.
This seems very similar to the old "Happiness = Reality - Expectations" equation. Great reality and low expectations? Large positive number. Poor reality and high expectations? Large negative number. If you want to be happy, then don't expect much, and don't expect often. Unfortunately for happiness, I don't know how to turn off the endless flow of "what if"s my mind produces - the flip side is that I'm always thinking of new things.
Similar to my model, which is itself quite intuitive to someone who likes economic modelling: I like to think of emotion as a function of the change in the expected discounted sum of lifetime utility.<p>First, utility is a measure satisfaction at any point of time, which can be a function of many variables, including relationships, food, wealth etc.<p>Discounted utility means that changes in soon satisfaction levels have a higher impact than changes in satisfaction that are still a long time away.<p>And the fact that I use the expected value of the function allows for biases abd inconsistencies, which are obviously very frequently observed.<p>[EDIT: typo]
What I don't get is how you reduce a variety of different emotions (hunger, anger, hope, etc) into a dichotomous "positive/negative outcome" spectrum. You don't really address that.
a) Hunger is not an emotional state. It may be an input to one's emotional state but I think that its impact is usually quite weak. b) Overly simplistic example-- do mature adults really have such wide ranging emotions during a conversation over whether there is pizza or not in the fridge?
A very nice summary of relevant scientific work is<p>www.fil.ion.ucl.ac.uk/~karl/Whatever%20next.pdf<p>Also, thinking fast and slow and other pop-sci books around decision making put a similar model forward. Personally, I am more focused on the active elements of consciousness and I try to use decision as the hermeneutics element rather than subject. Instead of looking at conscious beings like processors that model their environment and act upon that model, I am interested in the long term impact of decisions on behaviour and perception (i.e. my decisions impact the way I perceive the world.) Both models can co-exist. The article linked above has some very interesting links to much more researched models than my intuition.<p>Clarification: to reinforce my theory, I am wring about decisions when the author looks into emotions!
I wonder if a model like this one could be the basis for coding AI with emotions. Not that you'd really want a robot butler who experiences mood swings, but I'm sure there would be useful applications.
as the author mentions below, this is <i>his</i> effective internal model, and his sharing it simply offers the opportunity for anyone else to read it, think about it, and take and apply what works for him or her.<p>that said, for me, it's not so easily applicable to interactions with others, such as with Bob and Alice. for issues in my own peabrain, i can try and logically explore the bases for my emotions, my triggers, and possible alternate solutions, which i can apply at any point in a particular journey or task. i have the ability and choice to attempt to alter my own course, in terms of actions and reactions, any time i start to be plagued by and cognizant of my own discomfort. in interactions with others, however, the applicability is limited. unfortunately, it rests on having the ability to understand the complex nuances of someone else's emotional triggers, subconscious or unconscious, and having some sort of reliable heuristic to predict their reactions and digressions.<p>personally, i don't have an issue with the simplification to positive/negative outcomes. sort of reduces emotional processing and synaptic fires to binary code, or an electrical circuits with switches and AC current.
A former professor of mine wrote his PhD thesis on categorizing human emotion (I believe it was out of an AI department). Not sure how much help it will be, but it might have some ideas you can pull from.<p><a href="http://condor.depaul.edu/elliott/ar/papers/dis/elliott-phd.html" rel="nofollow">http://condor.depaul.edu/elliott/ar/papers/dis/elliott-phd.h...</a>
"This underestimate caused me to underestimate my ability to accomplish my goals"<p>I think the second "underestimate" is supposed to be "overestimate".