- Agent dicovering a minimal self
- Interaction between perception and reflection:
The cocenpts learned by the agents will be stored into the long term memory. We consider using the Semantic networks approach to store and connect the different concepts. Each learned concepts is represented by a node and those nodes are connected by links. The links might be shorter, for closely related concepts and or loger, for less related concepts.
It contains the concepts that are retrieved from the the Long term memory.
The angent needs to have perception mechanism to perceive the world surrounding him. At the begining, the agent doesn't know anithing about the world. We need to find a mechanism so that the agent learn new concepts (ex: self, Apple..ect) with all its features and caracteristics and store them in long term memory. if the agents see an know object, meaning that it does exist in the long-term memory, the this object(concept) will be retreived and stored in the working memory
We consider using the Spreading Activation Theory of Semantic Processing. Because concepts and ideas are connected together, if one concept is selected, then that will pull up other connected concepts along with it (Spreading Activation). For example, if the agent see an Apple in the scene, the that will activate the concept of Apple in the Long-term memory (we will need a search engine) and pulling up not only the Apple idea but also other related ideas such as "Sweet", "red", "cack"
