Anticipation by Analogy

This chapter claims that analogy-making, in particular the structural mapping ability is a very basic mechanism in human cognition and thus, it is also a basic mechanism for anticipation. A number of various everyday situations are analyzed that involve anticipation by analogy. Then the role of analogy is discussed in computational term to analyze why using analogy ensures flexible behavior and rapid learning based on a single case. A review of the existing analogy-making models is presented and their limitations are analysed. Then the specific AMBR model of analogymaking is presented, extended and integrated with mechanisms for visual perception and motor control. This is the first analogy-making model working in the real world extracting features from the environment using a camera, building its own representation, reasoning on them by analogy with episodes from long term memory, and acting in the real environment by modeling the process of motor control. The specific model of visual information processing implements a model of active vision with strong top-down pressures arising from the past experience. In this model there is a strong interaction between visual perception, analogical retrieval and analogical mapping they are working in parallel and influence and enhance each other. This model was implemented in a real AIBO robot and tested in the real world with simple examples. There is a lot to be done for extending the model and testing it with more complicated examples.