Computational neuroethology

Definitions Ethology is the study of behavior that allows animals to survive and reproduce in natural environments (i.e., adaptive behavior). Neuroethology is the study of the neural basis of adaptive behavior. Computational neuroethology is the use of modeling and simulation to study the neural control of adaptive behavior. In contrast, the broad area of computational neuroscience generally focuses on modeling microscopic properties of neurons (e.g., channel bio-physics, dendritic spines), single neurons (e.g., multiple conductance models of complex neurons and their morphology), neural circuits, and the dynamics of neural circuitry, without reference to biomechanics or environmental phenomena. History In 1990, both Randall D Beer and Dave Cliff independently called for the creation of a new field, computational neuroethology. Beer succinctly summarized the rationale for the new field: The working assumptions of computational neuroethol-ogy can be summarized as follows: (1) the ability to flexibly cope with the real world is a defining characteristic of intelligent behavior, and more fundamental than conscious deliberation; (2) adaptive behavior derives from a structural congruency between the dynamics of an intelligent agent's internal mechanisms and the dynamics of its external environment; (3) modeling the neural control of behavior in simpler whole animals will provide insights into the nature of the dynamics required for adaptive behavior, and eventually lead to an understanding of the successive elaborations of these mechanisms which are observed in higher animals. Recently, Cliff has published a review of this area that provides an excellent overview and a summary of key examples of computational neuroethology, with an emphasis on research that has focused on artificial agents. The conceptual framework of computational neuro-ethology draws on classical ethology, ecological psychology, biomechanics, and dynamical systems theory. The first key concept of the framework (see Figure 1) is that behavior is embodied, that is, it is not purely due to neural commands but emerges from the ongoing interaction of the central nervous system with a complex sensory and mechanical system. For example , in response to motion in the periphery of its visual system, a monkey will move both its eyes and head, thus using its motor system to reposition its sensory structures. As another example, the optimal angle for throwing a soccer ball from over the head is determined not simply by Newtonian mechanics but by the mechanical advantage that human arms can generate. More generally, embodiment implies that the central nervous system does not 'stand above' the …