Chapter 1
Introduction

Ansgar Büschges1 and Scott L. Hooper2

1Biozentrum Köln, Institute für Zoologie, Universität zu Köln, Köln, Germany

2Neuroscience Program, Department of Biological Sciences, Ohio University, Athens, OH, USA

It is de rigueur in a review or book on motor control to quote Sherrington's (1924) statement that "To move things is all that mankind can do". Although strictly true, this quotation discounts the central role in human experience of such actionless phenomena as ideation, emotion, and consciousness. However, it is nonetheless true that movement is an absolute requirement for animal survival and reproduction and, as the only observable output of the nervous system, is the defining basis of behavior. Movement is also self-defining, and hence allows analyzing nervous system function on the objective basis of its performance alone without reference to experimenter defined classifications. Disorders of movement also have great clinical importance, and production of functional and robust movement is a central problem in robotics. Because movements must be chosen among, and because almost all motor networks receive sensory input and information about internal state and "decide" how to alter their output in response, studying such networks may also provide insight into how the networks underlying "higher" abilities such as ideation function.

Despite this, many researchers, as well as lay people, take the generation of motor behavior for granted, often rendering it as the outcome of simple and automatic neural processes that can be summarized with large arrows pointing "south" from an animal's brain accompanied by the words "motor system". Only when confronted with particularly outstanding motor performances, e.g., the graceful movements of a dancer or an acrobat, do we appreciate the complexity of generating motor output. This disparity was well captured more than 200 years ago in von Kleist's (1810) essay Über das Marionetten Theater (On the Marionette Theater): "He asked me if indeed I hadn't found some of the movements of the puppets.to be exceedingly graceful in the dances. I could not refute this observation", a recognition that led Kleist to elaborate further on the potential mechanistic background of this observation. This text highlights how the ordinariness of movement can prevent us from appreciating how difficult it is to produce (something of which roboticists are well aware), and thus how extraordinary it is that nervous systems can do so.

The last general textbook covering how nervous systems do so, at least with respect to locomotion, was Neural Control of Locomotion (Orlovsky et al. 1999). This exceptional book described the neural networks and mechanisms that generate locomotion in mollusks, insects, anurans, lower vertebrates, mammals, and man. This book was the first comprehensive comparative account of how nervous systems generate locomotion. Such an overview had been lacking for decades and its detail and depth made and make it exceptional.

However, the book's concentration on locomotion meant that it, by design, did not cover the full range of movements animals produce. More importantly, dramatic advances in motor science have occurred since it was published. These advances represent a sea change in that motor research up to the 1990s primarily involved ever more elegant and detailed application of classical anatomical and single unit electrophysiological techniques. In the last two decades, alternatively, a much broader palette of methods has become available or practicable, including multi-unit recordings, molecular neurogenetics, computer simulation, and new approaches for studying how muscles and body anatomy transform motor neuron activity into movement. This broadening of experimental options has been exceptionally fruitful. However, it also means that researchers in motor control must be multi-competent, sufficiently informed and trained to be able to select from these multiple methodological options the optimal approach for the research question at hand.

It is important to make this observation because human nature and the process by which researchers are typically trained (prolonged and intensely concentrated research on a narrowly-defined question in an individual mentor's lab) work against achieving such multi-competence. Instead, as with a person with a hammer seeing every problem as a nail, it leads to researchers using the methods they know in preference to ones that might be better, but about which the researcher only peripherally knows. This is not a new observation, and conscious efforts are being made in training programs to train new researchers across fields. Nonetheless, in our experience barriers still exist between molecular biologists, electrophysiologists, muscle researchers, modelers, biomechanicists, and roboticists. It is a truism that reducing such barriers would serve all well. The question is, how to do so?

This book, in part, is an attempt to contribute to this effort. Its intended audience is all workers in movement production, from molecular biologists to roboticists. Workers in each group will have most knowledge of fields nearest their own .thus an electrophysiologist from a biology program likely has greatest understanding of molecular biology, and perhaps least of robotics. A biomechanicist likely finds it easier to communicate with a roboticist than a molecular biologist. And in our experience, modelers, at least those whose training was in classical mathematics, always speak a foreign language.

We therefore begin this book with four chapters covering basic knowledge on electrophysiological techniques, methods for large ensemble recordings, neurogenetic and molecular techniques, and computer simulation. These chapters are obviously not intended for experts in the field (although we hope they will be useful for beginning students in their labs, and the molecular biology and simulation chapters include case studies that will interest even experts in the fields). Rather, we hope that these chapters will allow workers outside each chapter's field to better understand and critically assess the field's literature, to understand the later chapters in the book, and encourage workers to reach outside their comfort zone and consider applying different methodological approaches to their research. We believe that writing these chapters, with their at least partially pedagogical nature, was likely a considerable change from the more purely research oriented reviews the authors would be typically asked to write. We are therefore particularly grateful to the highly distinguished colleagues in the field of motor control who were willing to take on this burden.

Hooper and Schmidt cover classical (i.e., not multi-unit) electrophysiological recording techniques. The first sections of this chapter are practical, and provide the information necessary for readers to understand and interpret intracellular and extracellular recording in the contemporary literature without a detailed explanation of theory. It is very difficult for modern readers to appreciate just how difficult it was for these techniques to be developed. The authors therefore next provide a brief history of extracellular and intracellular recording. The authors end the chapter with a detailed explanation of the theory underlying both recording techniques, and potential pitfalls that can occur with them.

Lebois and Pouzat cover multi-unit recordings.recordings in which electrodes that record the activity of multiple neurons are introduced into nervous tissue. The ability to do so strongly depends on proper electrode design and use, which the authors therefore first cover. Given that these electrodes record the activity of many neurons, advanced techniques are required to identify the individual activities of the many neurons being recorded from. The authors explain these techniques in detail in the chapter's second part.

Schoofs, Pankratz, and Goulding cover the use of molecular genetic tools to study neural network topology and function. They begin with a detailed explanation of the techniques available in invertebrates and vertebrates to observe and alter neuron activity. They then provide four cases studies, two in Drosophila and two in mice, in which these techniques were used to make novel findings in motor control that would have been presently impossible to achieve with other methods.

Prinz and Hooper cover computational simulation. The authors first provide a relatively high level overview of both the great power, and also the potential pitfalls, of simulation, making use throughout of case studies relevant to motor control. Because computer simulation may not be a part of the training of many of the book's intended audience, the authors then provide a detailed and basic explanation of how simulation is performed and how it is applied to neurons, synapses, muscles, and biomechanics.

In planning this book we also aimed to reduce another set of barriers: those between workers in different experimental preparations, of which the greatest is between workers in invertebrates and vertebrates. Doing so is important on both historical and scientific grounds. First, many to perhaps most discoveries made in one of these groups have been later found to be also present in the other. Second, recent data suggest a deep homology between (bilaterian) invertebrate and vertebrate motor control structures. This observation suggests that the last common ancestor of these two groups (the urbilatarian) had a...