Portal de Periódicos da CAPES

The cerebellum and adaptive control, John S. Barlow, Cambridge University Press, Cambridge, U.K., 2002. Hardcover, No. of pages: xi+340. Price: $66.50. ISBN 0‐521‐80842‐1

2008; Wiley; Volume: 22; Issue: 1 Linguagem: Inglês

10.1002/acs.960

1099-1115

Yogeshwarsing Calleecharan,

Hearing, Cochlea, Tinnitus, Genetics

Scientists have made little headway understanding the human brain and how it processes information. The cerebellum, located in the hindbrain, while occupying a small fraction of the brain functions as a regulator of the timing of muscle movements and is fault-tolerant to some extent. This book by John S. Barlow tries to show the close link that exists between the cerebellum and an adaptive controller. Though the intended audience of the book are students and researchers of neurophysiology and electrical or computer engineers alike, those with a sole engineering background can find the book hard to digest when confronted to clinical terms. Seventeen chapters make up this book and are themselves arranged in four parts: firstly cerebellar anatomy and physiology, secondly cerebellar function, thirdly models and theories, and lastly the summary and conclusions. Part 1 of the book covers the anatomy and physiology of the cerebellar system and comprises six chapters. In Chapter 2, the author surveys the anatomy of the human cerebellum and compares it to that of vertebrate animals. Chapter 3 contains a good description of the anatomy and physiology of the cerebellar cortex. Major information pathways are introduced. In Chapter 4, the mossy fibre afferent system is presented with particular emphasis on its role to convey information to the cerebellum. In the succeeding chapter, the inferior olive cells and the climbing fibres are covered in more detail. This chapter also includes a few hypotheses on the functions of the inferior olive. Part 1 closes with Chapter 6 where voluntary motor learning and the way the cerebellum continuously updates its internal models are dealt with. Part 2 focuses on cerebellar functions. In Chapter 7, cerebellar memory and both long-term depression and potentiation are addressed and are illustrated by some experiments. Chapter 8 treats the vestibulocerebellum and the oculomotor system. Discussions centre on the vestibulo-ocular reflex (VOR) for its role to stabilize images on the retina and also for its functional simplicity as a control system. Chapter 9 gives an account of the discoveries pertaining to the cognitive aspects of the cerebellum and points out that this part of the brain is not limited to motor control. Chapter 10 looks at the precise timing and the spatiotemporal information processing capabilities of the cerebellum. Chapter 11 gives a synopsis of principal findings in cerebellar pathology. Finally, the last chapter of Part 2 focuses on cerebellum-like structures present in some fish. Part 3 examines models and theories of the cerebellum. The four chapters making up this part address principal non-adaptive and adaptive models of the cerebellum. The first chapter, Chapter 13, presents various non-adaptive models and theories such as Braitenberg's Tidal Wave hypothesis. The chapter continues with some cerebellar learning theories and then some early adaptive control models follow which set up the pace for coming chapters. A discussion of the cerebellum as a sensory processor ends Chapter 13. Chapter 14 covers neural networks and these self-adaptive systems are compared to their biological cerebellar counterparts. Various neural network models are covered including an interesting part on VOR first introduced in Chapter 8. Chapter 15 deals with adaptive controllers and adaptive signal processors. Interesting topics such as linear prediction and system identification form part of this chapter. The last chapter of Part 3 is devoted to adaptive control models of the cerebellum like Ito's General Function model. Several experimental results augment the contents. Part 4 reviews the important concepts and theories laid in the book and its only chapter, Chapter 17, considers the modelling of the VOR with an adaptive controller. This is where the (engineering) reader is finally rewarded for reading this book. The beauty lies in that the model is simple but yet functional. A discussion of the cerebellum as a three-layer neural network also makes up this chapter. In the final sections of this chapter, an insightful treatment of the cerebellum function as a linear predictor or as a linear combiner is given. On an overall note, this book with its wealth of information in addition to an extensive bibliography and an appendix on the implementation of an adaptive signal processor will certainly attract neuroscientists. It is good to remember that the genetic code was regarded too complex to be deciphered only some decades ago. This work by John S. Barlow does not end here, but surely an enormous contribution has been made that will, for example, help in understanding cerebellar lesions and finding possible remedies. Not all robotic systems require adaptive control or cerebellar models to accomplish useful tasks; yet the integration of cerebellar control systems be it feedforward or feedback is of more concern to engineers who wish to devise robust and autonomous robots 1. A recent feat has been the successful implementation of a self-replicating robot by researchers at Cornell University 2. On a fantasizing note: one ultimate goal might be creating a C-3PO droid of Star Wars.