All conscious actions made by animals, including human beings, are based on their Reward System. In other words, all they do is to seek pleasure and avoid pain. This fact is stated by the Pleasure Principle and earlier by the Hedonist philosophy.
Pleasure and Pain are mysterious phenomena present in most multicellular organisms. The true nature of pleasure and pain is not discovered yet by the current science, that's why we can't implement it on robots. But we still can simulate the process and create a fake reward system for a robot, which is only a number; no emotion is felt by that robot! Such artificial intelligence may be called a Philosophical Zombie and the method used in it is called Reinforcement Learning.
Reinforcement Learning might potentially be dangerous when implemented in robots! That robot might burst out of control and do anything possible in order to achieve its FAKE pleasure and avoid its FAKE pain!!! That is why I am reluctant to implement it in my ultimate robot, but I implement a simple structure of it, ONLY as a practice for manipulating senses and somehow machine learning.
This component contains the following modules:
-
Rewarder : the main module which contains the Fortuna value; a double value whose negativity means an overall pain and whose positivity means an overall pleasure, computed using weighted arithmetic mean from multiple other fortuna scores related to some CRITERIA. (the idea comes from my Android app related to pleasure and pain: Fortuna and the names of the both come from a medieval song called O Fortuna)
-
Criterion : base class for criteria used for calculation of the ultimate Fortuna score.
-
Expression : base class for ways of expressing pleasure and pain; namely a base class for emotions.