The Relevance of the Time Domain to Neural Network Models
A significant amount of effort in neural modeling is directed towards understanding the representation of external objects in the brain. There is also a rapidly growing interest in modeling the intrinsically-generated activity in the brain, as represented
- PDF / 11,028,287 Bytes
- 234 Pages / 439.37 x 666.142 pts Page_size
- 112 Downloads / 184 Views
Springer Series in Cognitive and Neural Systems Volume 3 Series Editors John G. Taylor King’s College, London, UK Vassilis Cutsuridis Boston University, Boston, MA, USA
For further volumes: www.springer.com/series/8572
A. Ravishankar Rao r Guillermo A. Cecchi Editors
The Relevance of the Time Domain to Neural Network Models
Editors A. Ravishankar Rao IBM Thomas J. Watson Research Center 1101 Kitchawan Road Yorktown Heights, NY 10598, USA [email protected]
Guillermo A. Cecchi Dept. Silicon Technology IBM Thomas J. Watson Research Center 1101 Kitchawan Road Yorktown Heights, NY 10598, USA [email protected]
ISBN 978-1-4614-0723-2 e-ISBN 978-1-4614-0724-9 DOI 10.1007/978-1-4614-0724-9 Springer New York Dordrecht Heidelberg London Library of Congress Control Number: 2011938345 © Springer Science+Business Media, LLC 2012 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)
Foreword
What is the relevance of temporal signal structure to the brain? We may gain some insight by comparing the brain to the computer. In the modern computer, signals are binary (have only two possible values), are made to change as quickly as technology permits, and temporal relations between signals are of central importance. The computer is driven by a clock through a quick succession of globally ordered states, while great care and effort is expended to make sure that no signal spills over from one state to the next. Ordered states are defined by commands in a program, each command specifying the setting of a large number of switches. At one time [1], this picture of a digital machine was taken seriously as a model for the brain, switches being identified with neurons. Digital machines are universal, meaning that any conceivable finite process can be realized in them, thus creating the vision that also the processes of the mind could be realized as processes in a physical machine. At the time, this idea was taken as the breakdown of the formerly perceived impenetrable glass wall between mind and matter. Unfortunately, the research program of Artificial Intelligence, which was built on this vision, has not given us intelligence in the machine yet. What is wrong with this vision of the brain as a digital machine? The succession of states in the computer is specified by prog
Data Loading...
