The key concept that we want to convey in Lesson 2 is:
Focus on sensorimotor interactions rather than separating perception from action.
In Lesson 1, we saw that the agent's input data should not be confused with the agent's perception. Now, we need to better understand how exactly we should consider the input data. If input data is not perception, then what is it?
The sensorimotor paradigm suggests that input data should be taken in association with output data, by combining both of them into a single entity called a sensorimotor interaction. With the formalism introduced on Page 12, this gives i = 〈e,r〉: an interaction i is a tuple 〈experiment, result〉.
In his theory of mental development, Jean Piaget coined the term sensorimotor scheme to refer to a pattern of interaction between the agent and its environment. In our model, an interaction is a chunk of data that represents a primitive sensorimotor scheme.
From now on, we use the expression "to enact an interaction" to refer to performing the experiment and receiving the result that compose a given interaction. The expression "to intend to enact" interaction 〈e,r〉 means that the agent performs experiment e while expecting result r. As a result of this intention, the agent may "actually enact" interaction 〈e,r'〉 if it receives result r' instead of r.
See public discussions about this page or start a new discussion by clicking on the Google+ Share button. Please type the #IDEALMOOC021 hashtag in your post: