Using robots to understand animal behavior

Publisher Summary: This chapter discusses the use of robots to understand the animal behavior. A fundamental difference of robotic implementations from computer simulations of animal behavior is that there must be explicit means of transducing relevant signals and of materially affecting the surroundings. Animals cannot respond to signals for which they have no sensors. While the nature of physical interaction importantly shapes behavior, it is nevertheless true that most interesting behaviors of animals, such as crickets are also dependent on nontrivial neural processing that connects sensory and motor systems. Robotic implementations of models have several benefits while adding the advantage of bypassing the need to simulate (or mathematically represent) the interaction with the environment, instead using real interaction to represent itself. Clearly, the biorobotic approach is not ideal for every problem. The areas where it is likely to be most productive are those where a complete behavioral loop can be closed, that is, with some knowledge or plausible hypotheses about the nature of each of the intervening mechanisms.