Memory and Brain Dynamics: Oscillations Integrating Attention, Perception, Learning, and Memory

Memory and Brain Dynamics Oscillations Integrating Attention, Perception, Learning, and Memory

© 2004 by CRC Press, LLC

Conceptual Advances in Brain Research A series of books focusing on brain dynamics and information processing systems of the brain. Edited by Robert Miller, Otago Centre for Theoretical Studies in Psychiatry and Neuroscience, New Zealand (Editor-in-Chief); Günther Palm, University of Ulm, Germany; and Gordon Shaw, University of California at Irvine, USA.

Brain Dynamics and the Striatal Complex edited by R. Miller and J.R. Wickens Complex Brain Functions: Conceptual Advances in Russian Neuroscience edited by R. Miller, A.M. Ivanitsky and P.M. Balaban Time and the Brain edited by R. Miller Sex Differences in Lateralization in the Animal Brain by V.L. Bianki and E.B. Filippova Cortical Areas: Unity and Diversity edited by A. Schuz and R. Miller The Female Brain by Cynthia L. Darlington

© 2004 by CRC Press, LLC

Memory and Brain Dynamics Oscillations Integrating Attention, Perception, Learning, and Memory

Erol Basar ç Brain Dynamics Research Center Dokuz Eylül University, Izmir Brain Dynamics Multidisciplinary Research Network of Turkish Scientific and Technical Research Council TÜBITAK, Ankara The International and Multidisciplinary Research Network: Brain Dynamics and Cognition Affiliated with the IDP/United Nations, New York

CRC PR E S S Boca Raton London New York Washington, D.C. © 2004 by CRC Press, LLC

TF1663_C00.fm Page 4 Friday, May 21, 2004 1:15 PM

Dr. Erol Basar ¸ Director, Brain Dynamics Research Center and Chairman, Department of Biophysics, Faculty of Medicine Dokuz Eylül University Izmir, Turkey Chairman, The International and Multidisciplinary Research Network: Brain Dynamics and Cognition Affiliated with the I.O.P./United Nations, New York Brain Dynamics Multidisciplinary Research Network of Turkish Scientific and Technical Research Council TÜBITAK Ankara, Turkey E-mail: [email protected] Web page: http://braindynamics.deu.edu.tr/basar.htm

Library of Congress Cataloging-in-Publication Data Basar, ¸ Erol. Memory and brain dynamics : oscillations integrating attention, perception, learning, and memory / Erol Basar ¸ . p. ; cm. — (Conceptual advances in brain research ; v. 7) Includes bibliographical references and index. ISBN 0-415-30836-4 1. Memory. 2. Electroencephalography. 3. Brain. 4. Oscillations. I. Title. II. Series. [DNLM: 1. Brain—physiology. 2. Memory—physiology. WL 300 B297m 2004] QP406.B366 2004 612.8'23312—dc22 2003069760

This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the consequences of their use. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage or retrieval system, without prior permission in writing from the publisher. The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from CRC Press LLC for such copying. Direct all inquiries to CRC Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe.

Visit the CRC Press Web site at www.crcpress.com © 2004 by CRC Press LLC No claim to original U.S. Government works International Standard Book Number 0-415-30836-4 Library of Congress Card Number 2003069760 Printed in the United States of America 1 2 3 4 5 6 7 8 9 0 Printed on acid-free paper

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 5 Friday, May 21, 2004 1:15 PM

Series Preface For over 30 years Erol Basar ¸ has been pursuing a distinctive line of research into brain function. In his approach, signals from spontaneous EEGs and those produced in response to stimuli (evoked or event-related potentials) are analyzed into their frequency components and these components are then taken as the elements for correlation with psychological variables. His work is distinctive not only in scientific terms, but is also guided by a distinctive philosophy. This book, the seventh in the Conceptual Advances in Brain Research (CABR) series, is his most complete exposition to date of this philosophical and scientific perspective on brain function. Basar ¸ ’s approach has its origins in part in physics rather than biology. This is hinted at in the early chapters of his book, and is made explicit in Chapter 11 where he refers to the strategy adopted by Isaac Newton in understanding planetary motion. This approach is very apt. A wellknown aphorism of Newton came to mind as I read this final chapter: “Hypothesis non fingo” (usually translated as “I do not feign explanations”). Newton formulated the concept of gravity and showed how this concept allowed one to understand planetary motion and various other things, but he did not try to explain gravity in terms of something more fundamental. Likewise, Basar ¸ shows how frequency-specific patterns of electrical activity in the brain help one understand psychological processes, but does not try to explain those frequency-specific patterns in terms of lower level phenomena (i.e., neurones). This approach is far from the focus of researchers schooled in single-unit electrophysiology. However, many of the oscillatory phenomena described by Basar ¸ can already be explained in terms of neuronal biophysics and related interactions between networks of neurones. In principle, there is no reason to doubt that they can all be so explained. Basar ¸ occasionally refers to single-unit studies, and there is no doubt that he does in fact accept that these rhythmic patterns of activity are derived from patterns of synaptic activation in single neurones. However, that is not the conceptual language he prefers. Instead he sees the different frequency-specific components of massed neuronal activity as the real units for understanding brain functions. He presents a great deal of evidence (especially his own) showing that frequency-specific electrographic activity, selectively distributed in various brain sites, correlates with the psychological aspects of the tasks his subjects are performing. If we accept the conceptual language used by Basar ¸ , he ably demonstrates that what can be described using this language is very substantial. Here are some of the many examples. Regularly occurring, accurately timed sequences of stimuli lead to phase locking of EEG rhythms that develop as the stimulus pattern becomes familiar. Such regularization of frequency-filtered EEG components is related to the difficulty of the task and factors such as task fatigue. Well-known, event-related potential (ERP) components such as the P300 can be analyzed as various frequency components that have different psychological correlates. The frequency components induced by stimuli depend on the frequency composition of the prestimulus EEG. Coherence of oscillations between different parts of the brain is increased by stimulation and the entropy (scatter of frequency components) of the EEG is decreased by stimuli. Different EEG frequencies appear to be of differential importance in different parts of the hemispheres. Familiar stimuli (like a photo of the subject’s own grandmother) induce EEG rhythms across the whole of the hemispheres, with different frequencies dominant in different regions. Unfamiliar faces produce patterns of resonance different from familiar ones like a subject’s grandmother. All these findings depend on the use of frequency filtering of the EEG or ERP signals.

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 6 Friday, May 21, 2004 1:15 PM

One of the great strengths of this book is what can be called “psychobiological holism.” On the psychological side, Basar ¸ refers continually to an alliance of attention, perception, learning, and remembering, emphasizing that they are not sharply separate functions. This is a far more realistic view of cognitive function than the traditional one consisting of a series of independent “black boxes,” for which one is always tempted to search, but in vain, for their strict anatomical localizations. On the biological side, the electrographic correlates of psychological function are oscillations capable of interacting over the whole of the brain. As a result, the sort of phenomena that become important in Basar ¸ ’s view are rhythms distributed somewhat selectively over the whole brain, correlations at one region between oscillations at different frequencies, and coherences between oscillations in widely separate locations at the same frequency. The psychological and biological sides of this holism fit together naturally and convincingly. In the main, the empirical evidence is dealt with in separate chapters or sections from the theory development. These two parts of the book are welded together using a carefully developed didactic style. The evidence will be a rich source for future researchers, both empirical and theoretical. The theory development comes at various stages of the book, the most substantial of which is Chapter 9. Overall this is a bold and forward-looking essay that explains the brain as a whole rather than artificially separating it into components. At times the author admits realistically that his formulations are somewhat tentative and in need of future revision. Books like this are exactly what the CABR series was set up to promote. Robert Miller

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 7 Friday, May 21, 2004 1:15 PM

Preface … a major task for neuroscience is to devise ways to study and to analyze the activity of distributed systems in waking brains, including particularly human brains, and to seek direct correlations and explanations of the relevant behavior in terms of those patterns of neural activity. V.B. Mountcastle, 1998

DYNAMIC AND SELECTIVELY DISTRIBUTED MEMORY IN THE WHOLE BRAIN This book aims to bridge the disciplines of neurophysiology, cognitive psychology and EEG–brain dynamics with the aim of describing how the brain represents mental events that are interwoven with memory. Memory is inseparable from all other brain functions and involves distributed dynamic neural processes. The analysis of concerted action of multiple oscillatory processes is a major key to understanding distributed memory. The role of the memory in human behavior cannot be overemphasized because no high level nervous function can operate successfully without memory contributions. Perception, cognition, problem solving, and decision making all rely on memory. Thus, a major task for neuroscience is to choose strategies to analyze the activated memory in the awake brain. Based on the explosion of neuroscience literature, the concerted actions of distributed multiple oscillatory processes (EEG oscillations) play a major role in brain functioning. New important strategies are introduced in this book, one of which treats the alliance of attention, perception, learning, and remembering (APLR alliance) by means of EEG oscillations. According to Baddeley (1996), working memory provides a crucial interface of perception, attention, memory, and action. During experiments involving learning and working memory processing, EEG oscillations manifest continuously evolving dynamics. Empirical results led to development of a model of the hierarchy of memories as a continuum and a theory covering the concerted actions of function and memory in the whole brain. Unique leitmotifs and strategies for this book include use of the expression dynamic memory to describe memory processes that evoke relevant changes in alpha, beta, gamma, theta, and delta activities. The concerted actions of distributed multiple oscillatory processes constitute major keys for revealing distributed memory. The notion of physiological or fundamental memory is introduced. This type of memory includes phyletic memory and reflexes. The evolving memory incorporating reciprocal actions or reverberations in the APLR alliance and during working memory processes is emphasized. A new model related to the hierarchy of memories as a continuum is introduced. The notions of longer-activated memory and persistent memory are proposed as replacements for long-term memory. A new strategy related to recognition of faces emphasizes the importance of EEG oscillations in neurophysiology and gestalt analysis. According to Damasio, memory depends on several brain structures working in concert across many levels of neural organization. Memory is a constant work in progress. The proposition of a brain theory based on supersynergy in neural populations is most pertinent for understanding this constant work in progress. The proposed basic framework called memory in whole-brain work

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 8 Friday, May 21, 2004 1:15 PM

emphasizes that memory and all brain functions are inseparable; they act as a whole in the whole brain. The model and EEG strategies introduced in this book may be relevant for analyzing pathological changes in Alzheimer’s disease, patients and psychiatric patients with attention, learning, and memory deficits. The results presented in this book are based on the application of frequency analysis to EEG records from human and animal brains. Emphasis is placed on event-related oscillations and/or function-related oscillations. Because neuroscientists have concluded that different brain regions must cooperate to accomplish all brain functions, the analysis of the relationships of different regions of the brain is becoming more important. Lashley (1929) proposed that memories are in fact scattered across the entire brain rather than concentrated in specific regions. The results described in this book demonstrate that the whole brain is involved in these processes and that the memory function is selectively distributed in the brain. Lashley did not indicate the selective distribution because adequate experimental techniques to reveal it were not available in the 1920s. Hebb's fundamental concept of cooperativity concepts opened the area of interactive and growing networks in cognition research (1949). This book raises more questions than it claims to answer. It opens many new windows although some remain closed. As I finished the writing of my 1980 book on EEG–brain dynamics, I raised many questions. Answers came from a large number of neuroscientists, and research in many areas continues to expand results. I hope that the new pathways described in this book will gain importance and that many unsolved problems will be solved by young scientists working on the dynamics of memory function.

COPERNICAN CHANGES IN MEMORY RESEARCH According to Fuster (1997), our thinking about the cortical organization of primate memory is undergoing a Copernican change, from a neurophysiology that localizes different memories in different areas to one that views memory as a distributed property of cortical systems. According to Steven Rose (1997), memory is not, as previously thought, a vast cerebral warehouse filled with rows and rows of neatly ordered filing cabinets. It is impossible to know where in the brain a particular memory is located. Memory is a dynamic property of the brain as a whole rather than of any specific region. Memory resides simultaneously everywhere and nowhere in the brain. Our long-standing experiments with the Brain Dynamics Research Program that started in the 1970s led us to conclude that memory function is selectively distributed in the whole brain because oscillatory processes evoked by memory load occur in a selectively distributed manner in the whole brain as concerted (coherent) actions.

WHAT JUSTIFIES WRITING ANOTHER BOOK ON MEMORY? Several treatises have covered the neural presentation of memory. Fuster (1995) asked, “Who needs yet another?” The time has come to build a framework surrounding EEG-related memory processes. As early as 1985, I used the dynamic memory to describe memory processes that evoked relevant changes in alpha, theta, and delta activity (Chapter 3). My 1980 monograph titled EEG–Brain Dynamics: Relation between EEG and Brain Evoked Potentials was not in the main line of brain research when it was reviewed in Trends in Neuroscience in 1981. Since then, analysis of functionrelated brain oscillations is one of the most important areas of neuroscience research. My motivation to write this book was triggered by increasing numbers of publications in this area and also my experience in several other areas. I belong to a small group of scientists working on oscillation phenomena in the brains of a broad variety of species ranging from Aplysia ganglia to humans. © 2004 by CRC Press, LLC

TF1663_C00.fm Page 9 Friday, May 21, 2004 1:15 PM

The brain dynamics concept is most relevant for determining how memory is distributed because measurements on a timescale from 1 to 1000 MS (impossible to perform with functional magnetic resonance imaging [fMRI]) can now be achieved. Single sweep analysis of oscillations enormously contributes to analyzing the interactions of attentive states, learning, and evolution of memory, i.e., how EEGs are altered during the development of working memory (Basar ¸ and Stampfer, 1985; Chapter 3, this volume). Experiments on oscillatory dynamics provide the only possibility of elucidating the process of memory evolving over a short time interval. Our group started such studies very early. Our work also included research on chaos, entropy, and comparative studies of EEGs and MEGs. Our experiments with implanted cats allowed us to analyze distributed processes of the brain stem and cerebellum. As a result, memory traces in the whole brain can be analyzed by using our experimental data. In order to establish experimental strategies to reveal cognitive processes and integrative brain function, the neurons–brain theory and the notion of superbinding instead the concept of the cardinal pontine cell may play a major role. The goal of this short book is not to be the most comprehensive discussion of experiments and the dynamics of electrophysiology. The descriptions of biochemical and electrophysiological micromechanisms that serve to store information in the brain are not within the scope of this book. The models presented cannot be perfect. They are intended as examples to help build a new frame for the so-called dynamic memory. Accordingly, I hope that this book fulfills its purpose of proposing a new framework in the new domain of EEG-related memory research.

ABOUT THE CONTENTS AND ORGANIZATION OF THIS BOOK Beginning in the 1970s, a series of experiments examined the oscillatory character of event-related potentials in animal and human brains. At that time, the understanding of memory correlates of the measured oscillations was not the main goal. Instead, we aimed to attack basic physiological components of event-related oscillations. The situation changed dramatically in the mid-1980s when event-related oscillations provided an important window for revealing descriptions of cognitive functions and memory. Although it may provide guidance for design and interpretation of experiments, this book is not a text. Apart from the general chapters in the first part, it cannot be read in the usual manner. It must be tackled piece by piece, step-by-step, and will hopefully advance your knowledge of the fertile terrain of brain dynamics. When the preliminary experiments related to working memory and memory-related oscillations were published in the 1980s, we could not have predicted that gamma, alpha, theta, and delta oscillations would provide core material for scientists working on memory-related research. Accordingly, we use in this book a strategy that should orient readers to assimilate the experimental and theoretical material in parallel by going back and forth between experimental and theoretical chapters. The book is divided in three parts. Part I covers foundations; it presents the introductory core material and the rationale for the book. The material is essential for understanding how the oscillatory approach reveals information about brain functions and memory. Part II details core experiments and their interpretations. Part III includes theoretical and modeling-oriented chapters. Although Chapter 7 has a more theoretical character, it contains experimental results that serve as a theoretical framework. It is included in Part II to explain why the grandmother experiments were initiated. Part I — Chapter 1 describes preliminary concepts and some frameworks initiated since the 1920s. Chapters 3, 6, and 8 provide empirical data obtained by application of these concepts. The new data, in turn, led to new theories or principles. The paradigm change in cognitive sciences put more emphasis on analyses of macrodynamics instead of microdynamics. Accordingly, the © 2004 by CRC Press, LLC

TF1663_C00.fm Page 10 Friday, May 21, 2004 1:15 PM

conjecture is now open to establish a new conceptual framework or theory of neural populations to extend or replace Sherrington’s neuron doctrine and also to include memory in its framework. Chapter 2 explains definitions and concepts related to different types of memories. For readers starting to learn the essentials of memory function, a reading of this chapter is obligatory before attacking more difficult descriptions in coming chapters. This chapter is kept short. Readers who want to acquire more knowledge from the literature are referred to other readings cited at the end of this section. Part II — Chapter 3 is a key chapter that describes experiments on dynamics of memory by means of EEG- and event-related oscillations during cognitive processes based on performance of working memory tasks. Analysis of single epochs prior to and following target signals led to the concept of dynamic memory at the EEG level. The preliminary experiments performed more than 20 years ago gave us the first hints of the dynamics of evolving memory and reciprocal activation of attention, perception, and memory during working memory tasks. The results of the experiments support the theory of reentrant or recurrent networks. After reading Chapter 9, readers will possibly return to Chapter 3 to review the experimental grounds of the theoretical treatise on transitions, evolving memory, matching processes, and the new model presented in Chapter 9. Chapter 4 has a similar character to Chapter 3. It deals with similar experiments with attention and working memory paradigms performed on freely moving and behaving cats. The results with behaving cats made it possible to find correlates to memory function, perception, and attention in the whole brain including the brain stem. These key results in the whole brain allow us to state that memory function is manifested with selectively distributed oscillations in the whole brain and that results of investigators working with limited locations of electrodes in the human brain should be interpreted with great caution. Memory functions cannot be localized as Lashley (1929) pointed out. Prestimulus EEG activity and its roles in brain responsiveness and short-term memory were already explained in Chapter 3. In Chapter 5, the relation between prestimulus EEG and brain responses is analyzed with detailed analytical and systematic steps, thus allowing the interpretation of prestimulus EEGs as important factors in the causality of brain responses. This causality is strongly related to endogenous brain activity and, in turn, related to its cognitive states. It is also an important controlling factor for the reciprocal activation of functions of the APLR alliance, as will be discussed in Chapters 7, 8, 9, and 11. The causality principle of Newton’s dynamics and quantum dynamics play an essential constructive and interpreting role in memory-related brain dynamics. In Chapter 6, function- and memory-related oscillations are treated systematically by starting with a chronological survey. Readers who have less knowledge about brain oscillations may jump to Chapter 6 after reading Chapter 1. Chapter 6 contains representative examples of gamma, alpha, theta, and delta frequency channels. It shows that integrative brain functions are manifested by multiple oscillations; about 50 functional correlates of oscillatory responses are discussed. The principle of superposition and its functional role are explained and accompanied by samples, The selectively distributed alpha, gamma, theta, and delta systems are described by showing that frequency responses are modality- and topology-dependent. Another important feature of this chapter is the analysis of long-distance coherences in the brain. This opens the issue of action in concert of selectively distributed frequency systems and superbinding in Chapter 7. The chapter serves as an intermediary one, orienting readers not yet informed about conceptual developments of the last 5 years. It aims to bridge the various theoretical steps in somewhat chronological order. The essential idea is to train experimenters in neuroscience to develop new frameworks for treating the electrophysiology of cognitive functions. EEG research scientists oriented to functional analyses in neuroscience suffer from a lack of rules and principles similar to those used in conventional neurophysiology, and they seek theoretical frameworks and new rules for proper understanding of EEG recordings. Table 7.1 is self-explanatory. It explains activities of distributed oscillatory systems and their relation to integrative functions and memory. © 2004 by CRC Press, LLC

TF1663_C00.fm Page 11 Friday, May 21, 2004 1:15 PM

Chapter 8 presents advanced steps in the analysis of cognitive processes outlined in this book by treating the enigma of the grandmother neuron — a prominent problem in the neurophysiology literature of the last century. A crucial point is the fact that both anonymous and known faces evoke oscillations that are clearly different from those produced by anonymous faces. This differentiation is absolutely impossible to make by means of conventional ERP analysis techniques. Accordingly, readers oriented to electrophysiological methods may find useful tools in this chapter. These experiments were the first involving recognition of pictures of the subjects’ grandmothers. Preliminary but tenable consequences of the so-called grandmother experiments are explained in Chapter 8. The most important issue is that the whole brain and all oscillations are activated during recognition or remembering of the faces of the subjects’ grandmothers and faces that were unknown at the beginnings of the experiments. The responses behave as a three-dimensional construct consisting of temporal, spatial, and frequency spaces. The responses to the faces were not represented solely by one location or unique frequency. The selectively distributed nature of multiple oscillations of the whole cortex clearly denies the possibility of a new version of the neuron doctrine of Barlow (1995) as an extension to Sherrington’s initial doctrine. The grandmother experiments and their implications resemble the tip of an iceberg and may be expanded into new versions including pictures of known episodes or pictures that induce emotions, thus enabling electrophysiological differentiation and transition between semantic and episodic memories. Part III — Chapter 9 is the heart of the book and discusses the essential model derived from the study of memory by means of EEG oscillations. Readers who are curious about the results emerging from studies of EEG oscillations may start with this chapter, then review earlier material based on cross-references throughout the book. Chapter 9 begins with schematic descriptions of selectively distributed alpha, theta, and delta response systems. One important reference is the work by Fuster (1995 and 1997) in which the notion of distributed memory in the cortex and the hierarchy of memories were anchored by relevant physiological findings. A new model presenting various levels of memory function and hierarchies of various types of memories (memory states) is proposed; it constitutes the core of Chapter 9. The role of physiological processes, their contributions to memory function, and the transitions between memory states are emphasized. The chapter also covers a new hypothesis based on frequency tuning and resonance between brain neural populations (multiple frequency matching). Interwoven with the proposed new model are questions related to equipotentiality by Lashley (1929), the reverberation hypothesis by Hebb (1949), and the reentry hypothesis of Edelman (1977). This chapter can be considered a real workshop. Readers may page back and forth to other chapters in order to assimilate and/or criticize the ideas or notions of the new model based on findings with EEG oscillations. We are open to interactions with readers and welcome their emails. Our homepage will include a presentation of this model ([email protected]; http://braindynamics.deu.edu.tr). Chapters 10 and 11 contain important information about new trends and emerging ideas discussed throughout all chapters of the book, but both chapters have different aims. Chapter 10 provides a type of concluding synthesis of the new trends in analysis of memory function by means of EEG oscillations. Since it combines results and ideas presented throughout this book, it is useful for gaining a general comprehension of the subject of brain dynamics. After reading Chapter 10, readers may return to previous chapters, possibly after acquiring a general orientation after the reading of this chapter. Chapter 11 focuses on the future. Readers who are interested in theories related to brain function may find in this chapter a theoretical framework to orient them to designing new experiments, devising new theoretical proposals, or possibly modifying this proposal by using some of its empirical foundations or basic principles. The epilogue points out the hope that the draft of the theory on whole-brain work will provide a new groundwork for understanding dynamic memory. The glossary contains some of the © 2004 by CRC Press, LLC

TF1663_C00.fm Page 12 Friday, May 21, 2004 1:15 PM

abbreviations, nomenclatures, definitions, and descriptions of terms used in this book. The appendix explains relevant mathematical procedures frequently cited in this book in order to provide continuity of the text for readers who are familiar with common mathematical tools.

ACKNOWLEDGMENTS The Scientific and Technical Research Council of Turkey TÜBITAK has provided the major contribution for the preparation of this book and the achievement of joint experiments with German and Turkish scientists in the last 10 years or so. In 2000 I moved from the Medical University Lübeck in Germany, where I was leading the Neurophysiology Research Group, to Dokuz Eylül University in Izmir. Before and during this transition period, The German Research Council DFG and The German Ministry of Education BMBF offered major support for the realization of joint experiments in Bremen, Lübeck, Ankara, Izmir, and joint publications with the groups in Moscow, Sofia, Buenos Aires, and earlier in Perth. The McDonnell Foundation in the United States further supported the cooperation between the Lübeck and Sofia groups. The interaction between Lübeck and Helsinki was supported also by DFG and the European research organization BIRCH. The fruitful cooperation of scientists from three continents could be realized due to the generous support of these foundations. An essential contribution for establishing the International Multidisciplinary Network “Brain Dynamics and Cognition,” which operates under the official legacy of the International Organization of Psychophysiology (I.O.P.) associated with the United Nations in New York, was achieved by I.O.P. President Professor Dr. C.A. Mangina. I express my appreciation to Prof. Dr. Mangina for his efforts to motivate scientists for joint cooperation and thus to enrich worldwide understanding and peace. Dr. Murat Özgören, M.D., Ph.D., coauthor of Chapter 8, and Dr. Adile Öniz made excellent contributions to the preparation of the difficult manuscript. Moreover, they prepared a number of illustrations and contributed to the organization and preparation of references and the glossary. Dipl. Psychol. Christina Schmiedt and Cand. Psychol. Ingo Fründ in Bremen read and corrected ˇ the manuscript. Mrs. Ahrens, secretary of the Bremen Institute, and Mrs. C. Yegin, my secretary in Izmir, greatly helped in the organization of international joint research programs. A special note of thanks is due my spouse and most important colleague, Professor Dr. Canan ˘ at the University Bremen and mother of our children Eren and Pelin. Since the Basar ¸ -Eroglu 1980s, she has performed in Germany the most important experiments that constitute the core of this book. In the last years we have been able to perform the intriguing experiments of Chapter 8 in her laboratory in Bremen. Accordingly, her work has been invaluable throughout my entire career and also in the development of the present book. I also express my deepest appreciation to Professor Dr. Sirel Karakas¸, my former graduate student in Ankara. She has had a major role in all my monographs for more than 30 years as well as in this book as coauthor of the last chapter. Her incessant questions, constructive suggestions and ability to predict the new emerging hypotheses were extremely helpful. Therefore, she has been my most important companion in the new avenue of memory and brain dynamics. I greatly appreciate the contributions of all these persons and foundations.

SUGGESTED READINGS In order to achieve maximum gain from reading this book, readers must be somewhat familiar with the principles of neurophysiology, psychophysiology, and the psychology of memory. Excellent references include: Baddeley, A., Wilson, B.A., and Watts, F.N. (1995), Handbook of Memory Disorders, John Wiley & Sons, New York. © 2004 by CRC Press, LLC

TF1663_C00.fm Page 13 Friday, May 21, 2004 1:15 PM

Basar, ¸ E. (1998), Brain Function and Oscillations, Vol. I, Brain Oscillations: Principles and Approaches, Springer, Berlin. Basar, ¸ E. (1999), Brain Function and Oscillations, Vol. II, Integrative Brain Function: Neurophysiology and Cognitive Processes, Springer, Berlin. Damasio, A.R. (1994), Descartes’ Error: Emotion, Reason, and the Human Brain, Grosset/Putnam, New York. Eichenbaum, H. (1999), The hippocampus and mechanisms of declarative memory, Behavioral Brain Research, 103: 123–133. Eichenbaum, H. (2000), A cortical–hippocampal system for declarative memory, Nature: Reviews in Neuroscience (U.S.), 1: 41–50. Fuster, J.M. (1995), Memory in the Cerebral Cortex, MIT Press, Cambridge, MA. Goldman-Rakic, P.S. (1988), Topography of cognition: parallel distributed networks in primate association cortex, Annual Review of Neuroscience, 11: 137–156. Goldman-Rakic, P.S. (1996), Regional and cellular fractionation of working memory, Proceedings of the National Academy of Sciences of the U.S.A., 93: 13473–13480. Goldman-Rakic, P.S. (1997), Space and time in the mental universe, Nature, 386: 559–560. Kandel, E.R., Schwartz, J.H., and Jessel, T.M. (1991), Principles of Neural Science, Elsevier, New York. Miller, E.K. (2000), The prefrontal cortex and cognitive control, Nature: Reviews in Neuroscience (U.S.), 1: 59–65. Miller, R. (1991), Cortico-Hippocampal Interplay and the Representation of Contexts in the Brain, Springer, Berlin.

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 15 Friday, May 21, 2004 1:15 PM

Author Professor Erol Basar ¸ was born in Istanbul, Turkey and studied at the Universities of Munich, Hamburg, and Hanover in Germany. He was educated as a physicist and physiologist and earned a Ph.D. in biophysics. He joined the staff of the Physiology Institute in Hamburg in 1965 and was appointed a postdoctoral fellow at the Nathan Kline Brain Research Institute in New York in 1968. In 1971, he was appointed associate professor and founding director of the Institute of Biophysics at Hacettepe University in Ankara, Turkey, where he performed basic research on brain oscillations and integrative brain function. In 1978, Professor Basar ¸ was appointed the Richard Merton Professor of the German Research Council at the University Kiel. He served as head of the Neurophysiology Research Group at the Physiology Institute of The Medical University in Lübeck from 1980 through 2000. During that period, he worked on several international projects with scientists from San Diego, California; Perth, Australia; Moscow, Russia; Sofia, Bulgaria; Istanbul, Turkey; Helsinki, Finland; Buenos Aires, Argentina; Copenhagen, Denmark; Shanghai, China; and Vancouver, Canada that merited considerable attention. His most important collaboration was the study of invertebrate ganglia with Professor T.H. Bullock, as a result of which both scientists have organized conferences and edited books. Since 1993 Professor Basar ¸ has served as president of the Brain Dynamics Research Network of TÜBITAK (the Research Council of Turkey). He was named a professor at Dokuz Eylül University in Izmir, Turkey in 2000 and currently serves as director of the Brain Dynamics Multidisciplinary Research Center and the Department of the Biophysics at the University’s Medical School. Professor Basar ¸ is strongly involved with the founding of a premier international research center in Izmir with the support of DPT, the governmental planning agency in Ankara. Professor Basar ¸ is currently the Vice President for Academic Affairs of the International Organization of Psychophysiology (I.O.P.) associated with the United Nations (New York). He is also the chairman of The International Research Network on “Brain Dynamics and Cognition” affiliated with I.O.P./U.N. (New York). Professor Basar ¸ has published 12 books (five of which are monographs) and approximately 200 other publications and has organized six international conferences. Since the 1970s, he has been one of the pioneers who noted the importance of oscillatory brain dynamics for integrative brain function and memory. His monograph titled EEG–Brain Dynamics: Relation between EEG and Brain Evoked Potentials published by Elsevier in 1980 is known as a milestone in the field of brain dynamics. ˘ Professor Basar ¸ is married to Professor Canan Basar ¸ -Eroglu, a staff member at the Institute of Cognition Research in Bremen, Germany. They have written several publications together.

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 17 Friday, May 21, 2004 1:15 PM

Contents PART I Foundations Chapter 1

Introduction and Core Philosophy

1.1

Landmarks: Memory Is Distributed, Memory is a Dynamic Property 1.1.1 Lashley’s Equipotentiality 1.1.2 Hebb’s Rules of Cooperativity 1.1.3 Hayek: Perceiving is Classification of Objects by Activation of Associative Nets 1.2 New Trends in Neuroscience 1.3 Copernican Changes in Memory Research 1.3.1 Distributed Networks 1.3.1.1 Distributed Memory: Findings with Functional Magnetic Resonance Imaging 1.3.2 Parallel Distributed Processing 1.4 EEG-Brain Dynamics 1.4.1 Importance of EEG Studies 1.5 Pioneering Studies of Brain Macrodynamics and Whole Brain Approach 1.5.1 Griffith: Statistical Mechanics in Biology and Physics 1.5.2 Rosen: Global Neurodynamics 1.5.3 Fessard: General Transfer Functions of the Brain 1.5.4 Edelman: Reentrant Signalling Theory of Higher Brain Function 1.6 Freeman, Katschalsky, and Haken: Preliminary Steps in Introducing Macrodynamics of Electrical Activity 1.7 Application of Principles of Biological System Analysis to Brain Research 1.7.1 Reasons for Establishing Programs for Brain Research 1.7.1.1 Program Steps 1.7.1.2 Mathematical Methods of Program 1.8 New Approaches to Brain Functioning at Macroscopic Level 1.8.1 Sherrington’s Neuron Doctrine Revisited 1.8.2 Renaissance of EEG Use in Search of Integrative Brain Functions 1.9 Neurons-Brain Theory: An Approach that Includes Whole Brain Organization 1.9.1 Topography of Cognition and Elements of Neurons–Brain Theory 1.10 Significance of EEG Brain Dynamics in Memory States and Integrative Brain Functions Chapter 2 2.1

2.2

Concepts and Theories

Memory Machineries 2.1.1 Dynamic Memory and APLR Alliance 2.1.2 Steps of Memory Processing 2.1.3 Encoding and Retrieval Fractionation of Memory

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 18 Friday, May 21, 2004 1:15 PM

2.3 2.4

2.5

2.6

2.7

2.8

2.2.1 Long-Term Memory versus Short-Term Memory 2.2.2 Working Memory Distinction between Implicit and Explicit Memory States Nondeclarative Memory 2.4.1 Phyletic Memory 2.4.2 Perceptual Memory 2.4.3 Procedural Learning 2.4.4 Priming 2.4.5 Evolving Memory Declarative Memory 2.5.1 Episodic Memory 2.5.2 Semantic Memory 2.5.3 Relationship of Episodic and Semantic Memories Neurobiology of Memory 2.6.1 Molecular and Cellular Bases of Memory 2.6.1.1 Hebb’s Proposal 2.6.1.2 Kandel’s Fundamental Results 2.6.1.3 EEG Oscillations in Aplysia and Helix pomatia New Scheme Based on EEG Studies for Categorization of Memory Levels 2.7.1 Physiological (Fundamental-Functional) Memory 2.7.2 Transition and Combination of Memory Stages (Evolving Memory) Longer-Acting Memory and Transition to Persistent Memory in Whole Brain

PART II Experiments and Their Interpretation Chapter 3 3.1 3.2

3.3

Shaping Dynamic and Evolving Memories by Reciprocal Activation of Attention, Perception, Learning, and Remembering

Essential Experiments Involving Dynamic Memory and Top-Down Activity Dynamic Memory Manifested by Induced Alpha Activity 3.2.1 Selective Attention 3.2.2 APLR Alliance 3.2.3 Importance of Internal Event-Related Oscillations 3.2.4 Coherent and Ordered States of EEG due to Cognitive Tasks 3.2.4.1 Preliminary Experiments 3.2.4.2 Preliminary Results 3.4.3 Global Trends of Pretarget Event-Related Rhythms: Subject Variability 3.2.5 Paradigms with Increasing Occurrence Probability 3.2.5.1 3.5- to 8-Hz Range 3.2.5.2 8- to 13-Hz Range 3.2.5.3 40-Hz Range 3.2.6 Experiments with Light Stimulation 3.2.6.1 Experiments with Varied Probabilities of Stimulus Occurrence 3.2.7 Experiments with Subject A.F. 3.2.8 Quasideterministic EEGs, Cognitive States, and Dynamic Memories 3.2.8.1 Dynamics of Time-Locked EEG Patterns Relations between Memory States and P300 Responses: EROs 3.3.1 Experimental Set-Up and Paradigms

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 19 Friday, May 21, 2004 1:15 PM

3.3.2

3.4

Frequency Analysis of ERPs: Preliminary Results 3.3.2.1 Comparison of EPs and ERPs 3.3.2.2 Comparative Analysis of Poststimulus Frequency Changes under Different Conditions and Their Contributions to Different Latency Peaks 3.3.2.3 Formation of Peaks 3.3.2.4 Comparison of ERP Responses to Regular and Random Infrequent Target Stimuli 3.3.3 Orientation Reaction and Learning during Repetitive Stimulation Requirement of Preparation Rhythms for Activation of Working Memory: Analysis of Pre- and Poststimulus Activity in Single Sweeps 3.4.1 Event-Related Theta Oscillations 3.4.2 Event-Related 10-Hz Oscillations 3.4.2.1 Interim Summary 3.4.3 Modulation of P300 Activity by Preparation Rhythms 3.4.4 Control of Learnable Sequences by Prestimulus EEG Activity or Building of Memory Templates 3.4.5 Varied Degrees of Augmentation and Prolongation: Gamma Oscillations in Memory Tasks 3.4.6 Action of APLR Alliance and Hyphothesis Concerning Reentrant Circuits 3.4.7 Habituation 3.4.8 Augmentation of Knowledge or Learned Material Is Reflected by Regular and Increased Alpha Activities

Chapter 4

Perception and Memory-Related Oscillations in Whole Brain

Canan Basar ¸ -Eroglu ¸ ˘ and Erol Basar 4.1 4.2

4.3 4.4

Relevance of Chapter Theta and Alpha Responses in Cat Brains during Cognitive and Memory-Related Tasks 4.2.1 Introduction 4.2.2 Methods and Paradigms Utilized for Obtaining P300 Responses from Freely Moving Cats 4.2.3 Systematic Analysis of Effects of Repetition Rate of Omitted Tones on ERPs Recorded from Cat Hippocampi 4.2.4 Utility of Analysis in Frequency Domain 4.2.5 Multiple Electrodes in Hippocampus 4.2.6 Hippocampal P300 and Cognitive Correlates: Theta Components in CA3 Layer Compound P300–40-Hz Response of Cat Hippocampus 4.3.1 P300–40-Hz Compound Potential Event-Related Oscillations in Cat Hippocampus, Cortex, and Reticular Formation during States of High Expectancy: Comparison with Human Data 4.4.1 Unit Activity and Behavior 4.4.2 Event-Related Potentials in Cortex and Hippocampus in a P300-Like Paradigm 4.4.3 Selectively Distributed Theta System: Involvement of Limbic, Frontal, and Parietal Areas 4.4.3.1 Integrative Analysis of Increased Theta Response 4.4.4 Interpretation of Changes in ERPs 4.4.4.1 Comparison with Human Responses

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 20 Friday, May 21, 2004 1:15 PM

4.4.5 4.4.6

4.4.7 4.4.8 Chapter 5 5.1 5.2 5.3

5.4

5.5 5.6 5.7 5.8

6.2

6.3

Causal Factors Controlling Brain Responsiveness and Memory: Prestimulus EEG Activity, Entropy, and Genetics

Introduction Relationship of EEG and ERP Algorithm for Selective Averaging 5.3.1 Dependence of EP Amplitudes and Waveforms on Prestimulus EEG: Vertex Recordings 5.3.1.1 Auditory-Evoked Potentials 5.3.1.2 Visual-Evoked Potentials 5.3.1.3 Topographic Aspects 5.3.2 Frontal Visual-Evoked Potentials 5.3.3 Inverse Relations of EEGs and Visual Responses Frequency Content of EROs from Different Locations: Major Operating Rhythms 5.4.1 Major Operating Rhythm (MOR) of Frontal Lobe: Theta? 5.4.2 MORs of Occipital and Central Region (Vertex) 5.4.3 Functional Significance of EEG–EP Interrelations Barry: Preferred States in Brain Activity 5.5.1 Creation of Preferred Brain States by APLR Alliance Causality of Brain Responses According to Changes in Oscillatory Networks Entropy as Causal Factor in Responses and Mechanisms of Super-Synergy Genetics as a Causal Factor in Delta and Theta Responses and Beta Rhythms

Chapter 6 6.1

Why Compare EP Results with Conventional Experiments? Structures Involved in States of APLR Alliance 4.4.6.1 Hippocampus as Supramodal Structure 4.4.6.2 Frontal Cortex 4.4.6.3 Global Function of Reticular Formation 4.4.6.4 Cognitive Functions of Cerebellum Secondary Alpha Response and Alpha Response with Delay Comparison with Human Brain Results

Correlation of Multiple Oscillations with Integrative Functions and Memory

Introduction 6.1.1 Aim of Chapter 6.1.1.1 Emphasis on Multiple Oscillations in Brain Research 6.1.1.2 Role of Oscillations in Memory Processing 6.1.1.3 Steps for New Synthesis and Binding Problem Survey of EEG Oscillations 6.2.1 Alpha Activity 6.2.1.1 Survey by Andersen and Andersson (1968) and Basar ¸ (1999) 6.2.1.2 Toward a Renaissance of Alphas 6.2.2 Earlier Experiments on Induced or Evoked Theta Oscillations 6.2.3 Gamma Frequency Range Selectively Distributed Oscillatory Systems: Distributed Multiple Oscillations 6.3.1 Concept, Definitions, and Methods 6.3.2 Oscillatory Responses in Invertebrate Ganglia 6.3.3 Gamma Oscillations in Sensory, Cognitive, and Motor Processes 6.3.3.1 Multiple Functions in Gamma Band 6.3.3.2 Important Causality Factor for Human Gamma Response

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 21 Friday, May 21, 2004 1:15 PM

6.3.4

6.4 6.5 6.6 6.7

6.8

Alpha Oscillations in Perception and Cognition: The Alphas 6.3.4.1 Sensory Components 6.3.4.2 Cognitive Components 6.3.4.3 Resonance in Brain Responses 6.3.4.4 Multiple Functions in Alpha Frequency Window 6.3.5 Theta Oscillations in Perception and Cognition 6.3.6 Delta Oscillations in Cognition 6.3.7 Klimesch: Multiple Oscillatory Activities in Alpha Band 6.3.8 Oscillations in Highest Frequency Window Superposition Principle and Superimposed Multiple Oscillations in Theta and Delta Frequency Windows in Cognitive Processes: Examples Selectively Distributed and Selectively Coherent Oscillatory Networks Interim Conclusions Distributed Oscillatory Systems and Distributed Memory 6.7.1 Event Processing in Distributed Systems 6.7.2 Multiple Functions of EROs and Multiple Functions of Memory: Convergence of Concepts 6.7.3 Human Memory Performance and Time-Locked Theta Responses EEGs and EROs as Information Codes 6.8.1 Frequency Coding at Different Levels 6.8.2 Most General Transfer Functions and Multiple Oscillations

Chapter 7 7.1 7.2 7.3 7.4 7.5 7.6 7.7

Are Integrative Brain Functions Shaped by Superbinding and Selectively Distributed Oscillations?

Rationale and Usefulness of this Chapter Binding Problem in Memory Processing and Gestalt Neurons–Brain Theory and Oscillatory Codes Description of Function–Memory Table Super-Synergy: A Spatio-Temporal and Functional Organization of Multiple and Distributed Oscillations Gedanken Model: Involvement of Selectively Distributed and Coherent Activities of Neural Populations in Grandmother Percept Neural Populations and “Feature” Cells 7.7.1 Sokolov: Feature Detectors

Chapter 8

Grandmother Experiments in Perception of Memory: Recognition of Gestalts

Erol Basar ¸ and Murat Özgören 8.1 8.2 8.3

Introductory Remarks Klimesch: Role of Theta and Alpha Oscillations in Memory and Attention Functions Grandmother Paradigm and Gestalt Experiments 8.3.1 Experimental Strategy 8.3.1.1 Electrophysiological Recording 8.3.1.2 First Data Recording (Random) Set 8.3.1.3 Second Data Recording (Regular) Set 8.3.2 Event-Related Oscillations Arising from Light, Anonymous Face, and Grandmother Face Stimulations 8.3.2.1 Topologies of Delta Responses 8.3.2.2 Topologies of Theta Responses 8.3.2.3 Topologies of Alpha Responses

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 22 Friday, May 21, 2004 1:15 PM

8.3.2.4 Distributed Beta and Gamma Responses Differentiation of Responses in Delta, Theta, and Lower and Upper Alpha Frequency Bands: Preliminary Statistics Recognition Memory and Gamma Oscillations What Does the Grandmother Paradigm Mean? Are Oscillations Distributed Templates in Memory Activation? 8.5.1 Selectively Distributed Enhancements in Whole Cortex 8.5.2 Efficiency of Grandmother Paradigm for Differentiation of Memory Components or States 8.5.3 Does Activation of Larger Neural Populations Indicate Reactivation of Episodic Memory? 8.5.4 Transition from Semantic to Episodic Memory: Distinctions between Semantic and Episodic Memories 8.5.5 Importance of Frontal Lobes and Other Brain Areas for Memory Processing and Perception 8.5.5.1 fMRI Experiments Related to Distributed Memory in Cortex 8.5.5.2 Critique of Experiments of Fernandez and Fell 8.5.5.3 Major Activation Areas of Semantic and Episodic Memories 8.5.5.4 Superbinding and Stryker’s Question about Oscillations 8.5.6 Do Grandmother Experiments Favor Hebb’s Hypothesis? Are the Descriptions of Gestalts and Emotions Related to More Complex Percepts Possible? 8.3.3

8.4 8.5

8.6

PART III Memory Function: Models and Theories Chapter 9 9.1 9.2

9.3

EEG-Related Models of Memory States and Hierarchies

Introduction of a New Construct on Memory Categorization Physiology of Selectively Distributed Oscillatory Processes 9.2.1 Connections of Sensory–Cognitive Systems 9.2.2 Activation of Alpha System with Light 9.2.3 Activation of Alpha System with Auditory Stimulation 9.2.4 Activation of Theta and Delta Systems Following Cognitive Inputs 9.2.5 Nonspecific Interactions Hierarchical Categorization of Different Levels of Memory 9.3.1 Fuster’s View of Memory Networks: A Milestone in Neuroscience 9.3.2 Tentative Model Related to EEG Activation 9.3.3 Inborn (Built-In) Networks (Level I) 9.3.3.1 Reflexes 9.3.3.2 Stereotypic Fixed Action Patterns 9.3.3.3 Phyletic Memory and Oscillatory Response Codes 9.3.3.4 Feature Detectors 9.3.3.5 Living System Settings 9.3.4 Physiological or Fundamental Memory 9.3.4.1 Changes of Sensory Memory: Spontaneous and Evoked Alpha Activity at Occipital Sites in Three Age Groups 9.3.5 Working Memory (Level II) 9.3.5.1 Perceptual Memory

© 2004 by CRC Press, LLC

TF1663_C00.fm Page 23 Friday, May 21, 2004 1:15 PM

9.3.6

9.4

9.5

9.6

Incorporation of Oscillatory Codes in Physiological Memory Consisting of Phyletic, Sensory and Perceptual Memories 9.3.7 What is Motor Memory? 9.3.8 Development of Procedural Memory throughout Life Dynamic Memory in Whole Brain: Memory States instead of Memories 9.4.1 Alpha, Theta, and Delta Oscillatory Processes during APLR 9.4.2 Are Dynamic EEG Templates Created during Processing of the APLR Alliance? Do They Build (Virtual) Short-Term Storage of Newly Learned Material? 9.4.3 Are All Brain Functions Linked with Memory? Complex Memory or Multiple Matching: Evolving Memory and APLR Alliance 9.5.1 Memory Activation: Working Memory and Hierarchical Organization of Memory States 9.5.2 Multiple and Complex Matching Processes: Reciprocal Activation of Alpha, Delta, Theta, and Gamma Circuits in Whole Brain 9.5.2.1 Reentry? 9.5.3 Prolonged Oscillations, Delays, and Coherent States during Complex Matching 9.5.4 Complex Matching 9.5.4.1 Matching of Multiple Oscillations in Whole Brain Longer-Acting Memory and Transition to Persistent Memory in Whole Brain 9.6.1 Evolving Memory: Multiple Level Functioning in CNS 9.6.2 Level III Activities Portrayed in Figure 9.7

Chapter 10 New Trends in Memory Dynamics: Concluding Remarks 10.1 The Emphasis of this Book: From a Research Program to a Theory on Whole-Brain Work 10.2 Distributed Memory in the Whole Brain 10.3 Correlation of Brain Oscillations with Multiple Brain Functions 10.3.1 Are All Memory States Tuned with Frequencies of EEG Oscillations? 10.4 Gestalts and the Grandmother Percept 10.5 Activated Memory Manifested by EEG Oscillations 10.5.1 Plausibility of Hebb’s Reverberating Activity Based on EEG Experiments 10.6 Model Related to Memory States 10.6.1 Active Memory and Reverberation Hypothesis 10.6.2 Memory State as Continuum 10.6.3 Multiple Matching with EEG Frequency Codes as an Essential of Recognition 10.6.4 Longer Acting Memory and Persistent Memory 10.7 Importance of EEG Analysis Chapter 11 Memory and Whole-Brain Work: Draft of a Theory Based on EEG Oscillations Erol Basar ¸ and Sirel Karakas¸ 11.1 Integration of Proposals Related to Whole-Brain Work 11.2 Whole-Brain Work Theory: How to Approach Brain Functions by Means of EEG Oscillations 11.2.1 Level A: Transition from Single Neurons to Oscillatory Dynamics 11.2.2 Level B: Superbinding of Neural Assemblies (Supersynergy) © 2004 by CRC Press, LLC

TF1663_C00.fm Page 24 Friday, May 21, 2004 1:15 PM

11.2.3 Level C: Integration, Alliance and Interplay in Memory 11.2.4 Level D: Causality and Brain Responsiveness 11.3 Newtonian Causality, Chaotic Dynamics, and Brain Language Epilogue:

From EEG–Brain Dynamics to Memory–Brain Dynamics

References Abbreviations and Glossary Appendix: Relevant Mathematical Methods

© 2004 by CRC Press, LLC

TF1663_book.fm Page 1 Friday, May 7, 2004 8:47 AM

Part I Foundations

© 2004 by CRC Press, LLC

TF1663_C01.fm Page 3 Thursday, May 13, 2004 2:24 PM

and Core 1 Introduction Philosophy 1.1 LANDMARKS: MEMORY IS DISTRIBUTED, MEMORY IS A DYNAMIC PROPERTY Memory is a functional property. Since brain functioning is based on dynamic processes, memory is also a dynamic process. Seeing even the simplest light signal is a memory process related to a fundamental inborn retrieval response. A baby perceives and shows reflex responses to light before he or she is exposed to more complicated learning processes. The response to light is probably a basic decoding process. Fuster (1997) stated that memory reflects a distributed property of a cortical system. Important components of higher nervous system functioning such as perception, recognition, language, planning, problem solving, and decision making are interwoven with memory. This author considers memory a property of the neurobiological systems it serves; it is inseparable from their other functions. Memory is a dynamic property of the brain as a whole rather than a characteristic of any single specific region; it resides simultaneously everywhere and nowhere in the brain (Rose, 1997). According to Antonio Damasio (1997), “Memory depends on several brain systems working in concert across many levels of neural organization. Memory is a constant work in progress.” How did neuroscientists arrive at such conclusions? The conclusions are based on a long evolution of thoughts and concepts originating with Karl Lashley, Donald Hebb, and F.A. Hayek in the first half of the 20th century. The following sections briefly explain the works of these pioneers.

1.1.1 LASHLEY’S EQUIPOTENTIALITY To study learning and memory concepts in mammals, Karl Lashley (1929) taught rats to successfully negotiate complex mazes. He then began incrementally removing thin slices of each rat’s cerebral cortex in an effort to pinpoint the memory locus for this task. No matter which sections of brain Lashley removed, the rats were still able to run the maze. Their performances diminished progressively as more brain tissue was excised, but Lashley found no single region whose ablation completely erased memory. In a landmark paper, Lashley proposed the theory of equipotentiality: memory is in fact scattered across the entire brain and is not concentrated in specific regions.

1.1.2 HEBB’S RULES

OF

COOPERATIVITY

Hebb`s rule (1949) implies that information processing requires functional cooperation by distributed neurons. More precisely, this rule postulates that groups of synapses that have a tendency to fire together and converge on a single neuron become strengthened as a group. This is known as the principle of cooperativity. Does some kind of modification of neurons or modification of connections between neurons occur as a result of learning? For example, when we learn to associate two stimuli (e.g., an unconditioned stimulus and conditioned stimulus as in classical conditioning), what happens in the brain to support the learning process?

© 2004 by CRC Press, LLC

TF1663_C01.fm Page 4 Thursday, May 13, 2004 2:24 PM

Early attempts to answer this question can be traced back to Donald Hebb who in 1949 proposed that the coactivation of connected cells would result in a modification of weights and when a presynaptic cell fired, the probability of firing by a postsynaptic cell firing was increased. Hebb said, “When an axon of cell A is near enough to excite cell B or repeatedly or persistently takes part in firing it, some growth or metabolic change takes place in both cells such that A’s efficiency as one of the cells firing B is increased.” This learning principle did not specify exactly what was meant by growth or metabolic change, but it served as a useful starting point and has become the widely cited heuristic for neurobiological investigations of learning and memory. The distributed nature of activations in cognitive tasks described in this chapter may explain why Lashley thought that the brain operated as a whole. The cooperation among distributed structures of the brain is also a factor because the coherences are selectively distributed. Analysis of oscillations in several neural populations of the brain in parallel and in various frequency windows brought a new refinement to descriptions of the whole brain and cooperativity: The whole brain is activated in all perceptual and memory-related mechanisms. The intensity of electrical oscillatory responses is selective in neural populations. The links or cooperativity, measured by means of coherences and phase differences, also show varied degrees of intensities. Accordingly, we may explore new interpretations of the statements of Lashley and Hebb by using new tools to analyze the electrical activities of the brain during sensory−cognitive activities. Hebb rejected the notion that stimulus−response relationships could be explained by simple reflex arcs connecting sensory neurons to motor neurons. It was necessary to postulate “a central neural mechanism to account for the delay between stimulation and response.” Hebb believed that sensory stimulation could initiate patterns of neural activity that were centrally maintained by circulation in synaptic feedback loops. Such reverberatory activities made it possible for response to follow stimulus after a delay. Seung (2000) claimed that the validity of Hebb’s theory remained uncertain. Although the existence of the Hebbian synapse is not in doubt, whether delay activity is thoroughly reverberatory is still unclear. (See also Section 2.6.1 in Chapter 2.) Electroencephalogram (EEG) studies recorded several delays and prolongations of responses (Chapter 3 and Chapter 4). Are the delays and prolongations candidates for reflecting Hebbian reverberatory mechanisms? Although no concrete answer can be provided, the possibilities will be discussed in Chapter 9. In the author’s opinion, the delays and prolongations of oscillatory responses reflect prolonged work of neural populations following difficult cognitive or memory tasks and their analysis can provide important hints for establishing learning and remembering models.

1.1.3 HAYEK: PERCEIVING ASSOCIATIVE NETS

IS

CLASSIFICATION

OF

OBJECTS

BY

ACTIVATION

OF

Associative networks play important roles in complex dynamics. Such networks are also considered essential building blocks in modern memory research. Hayek’s work (1952) was described in a very comprehensive and useful manner by Fuster (1995) who found Hayek’s work more important than Hebb’s related to describing memory function. Perceiving is the classification of objects by activation of the associative nets that represent them in memory. According to Fuster (1997), our thinking about the cortical organization of primate memory is undergoing a Copernican change — from a neurophysiology that localizes different memories in different areas to one that views memory as a distributed property of cortical systems. According to Fuster’s empirically founded hypothesis, the same cortical systems that serve us in perceiving the world also serve us in remembering it. Perceiving is the classification of objects by activation of the associative nets that represent them in memory. It is reasonable to assume, as Hayek did, that memory and perception share the same cortical networks, neurons, and connections to a large extent. To understand the formation and topography of memory, it is useful to think of the primary and sensory motor areas of the cortex that we may call the phyletic memory or the memory of the species. The primary sensory © 2004 by CRC Press, LLC

TF1663_C01.fm Page 5 Thursday, May 13, 2004 2:24 PM

and motor cortices may be considered funds of memory acquired by a species through evolution. We can use memory as part of the term because, like personal memory, the phyletic memory consists of information that has been acquired and stored and can be retrieved (recalled) by sensory stimuli or the need to act.

1.2 NEW TRENDS IN NEUROSCIENCE Between 1980 and 2000, seven important steps in neuroscience research advanced our understanding of brain dynamics and function: 1. The discovery of oscillatory phenomena at the cellular level based on the 40-Hz studies by Singer (1989) and Eckhorn (1988), measurements of 10- and 5-Hz oscillatory behavior at the membrane level, and extracellular single recordings (Dinse et al., 1997, Llinás, 1988). 2. The application of chaos theory to electroencephalogram (EEG) signals, demonstrating that the EEG is not only a noise signal (for reviews see Basar ¸ , 1990; Duke and Pritchard, 1991; Molnár, 1999). 3. Developments based on the acceptance of cognitive function analysis by the use of the EEG and event-related potentials (ERPs). 4. The use of the magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI) as complementary tools. 5. The development of fast laboratory computers and availability of sophisticated neurocomputing software that accelerated progress in all fields of research. 6. The binding hypothesis occupied an important place in conceptual discussions, although we strongly emphasized that it is not sufficient to explain the mechanisms of complex percept building. 7. Copernican changes in memory research, particularly as discussed in the publications of Fuster (1995 and 1997); Goldman-Rakic (1997); Mesulam (1990 and 1994); and Kandel (1982). (See also Section 2.6.1.3.)

1.3 COPERNICAN CHANGES IN MEMORY RESEARCH 1.3.1 DISTRIBUTED NETWORKS According to Fuster (1997), the classic terms (representation, retrieval, recall, recognition, shortterm memory, and long-term memory) remain valid, but need to be neurobiologically redefined. Arguably, the smallest memory network (netlet) is a cortical cell group or module representing a simple sensory response; a memory reflects a distributed property of a cortical system. It can be hypothesized that selectively distributed oscillatory systems (or networks) may provide a general communication framework and be useful for functional mapping of the brain (Mesulam, 1990 and 1994). Communications in these networks may contribute to the formation of specific templates belonging to objects and memories. According to a model of cognition, this formation occurs as selectively distributed processing with considerable specialization and in anatomically differentiated localizations (Mesulam, 1990 and 1994; for details about memory as a distributed property of a cortical system, see also Fuster, 1997). In particular, analysis of hypothetical distributed oscillatory systems may lead to fundamental functional mapping of the brain, complementary to morphological studies. Perceptual memory is acquired through the senses. It comprises all that is commonly understood as personal memory and knowledge, i.e., representation of events, objects, persons, animals, facts, names, and concepts. From a hierarchical view, at the bottom level are memories of elementary © 2004 by CRC Press, LLC

TF1663_C01.fm Page 6 Thursday, May 13, 2004 2:24 PM

sensations; at the top are abstract concepts that, although originally acquired by sensory experience, have become independent as a result of cognitive operations. Single neuron recordings in monkeys trained to perform working memory tasks have identified components of a working memory circuit in the prefrontal cortex. The neuronal processes related to task performance can be dissociated on the scale of milliseconds to seconds. During a working memory task, as the stimulus is sequentially registered and stored over a period of seconds and then translated into a motor response, specific neural populations respond in characteristic ways. One class of prefrontal neuron responds to a visual stimulus as long as the stimulus is in view. In contrast, other prefrontal neurons are activated at the offset of the stimulus and remain active as long as the monkey must remember the location or features of an object (Fuster, 1995; GoldmanRakic 1988 and 1997). As one can deduce from the work of Mesulam and Fuster, common codes for perpetual signal transfers between neural networks for parallel and serial processing and also for possible reverberation circuits and loops between neural network must exist. Oscillations in the brain may serve as adequate codes for this general communication by inciting networks to resonate. A more general view is that functional or oscillatory network modules are distributed in both the cortex and throughout the whole brain (Basar ¸ , 1999). We will now discuss an electrophysiological (EP) parallel between Fuster’s memory network and the distributed oscillatory systems mentioned earlier. When analyzing field potentials, it is difficult to define boundaries of brain nuclei and their electrical activities. Nevertheless, this approach is useful because great amounts of data can be collected and interpreted from several electrodes distributed in the brain. Furthermore, it is possible to perform measurements during continuously changing cognitive states. EP studies and EEG segments from the cortex, limbic system, thalamus, and cerebellum can be recorded and compared in waking and freely behaving animals. This type of recording during behavioral states cannot possibly be managed with single cell electrodes. Studies of functional correlates of structures like sensory cortices, hippocampi, and thalamic relay nuclei are based mostly on experiments using unit recordings. A major difficulty with interpretation of experiments made by single unit recordings (for example, experiments on corticothalamic information transfer) is that the results are limited to a few neurons. Accordingly, the author assumes that every hypothesis on localization of the thalamocortical circuit as a 10-Hz generator is restricted and not acceptable with regard to the results of experiments described in this book: the alpha, theta, and gamma generators are selectively distributed in the brain. 1.3.1.1 Distributed Memory: Findings with Functional Magnetic Resonance Imaging Cohen et al. (1997), and Courtney et al. (1997) used fMRI studies of humans to find parallels to the knowledge gained from single-cell recordings of animals. Courtney presented subjects with pictures of human faces and asked them to recall whether each picture was the same or different, from a picture presented 8 s earlier. Activations in the prefrontal areas correlated most strongly with delay periods, compared with activations in the visual areas that more strongly correlated with sensory stimulation. Cohen et al. presented subjects with single written consonants every 10 s and asked the subjects to judge whether each consonant was the same as a consonant presented one, two, or three trials back in the sequence. This task required subjects to remember the identities of the consonants and the order in which they were presented. The farther back in the sequence the consonant to be recalled appeared, the greater the load on the working memory. The authors showed that activations in the prefrontal cortex were maintained throughout the 10-s interstimulus intervals. The degree of prefrontal activation was higher for conditions with the greatest memory loads. By contrast, activations by the primary visual, somatosensory, and motor cortices and in several secondary regions were not sustained across the 10-s interval and were not related to memory demand. They © 2004 by CRC Press, LLC

TF1663_C01.fm Page 7 Thursday, May 13, 2004 2:24 PM

were probably responsive to sensory or perceptual stimuli, but did not aid the working memory in performing tasks. Based on the fMRI results of Courtney et al. (1997), early extrastriate visual areas demonstrated transient, relatively nonselective responses to complex visual stimuli and later extrastriate visual areas demonstrated transient, selective responses to faces. This indicated a more specialized role in the processing of meaningful images. Both extrastriate visual and prefrontal cortical areas demonstrated sustained activity during memory delays, indicating a role in maintaining an active representation of the face in working memory.

1.3.2 PARALLEL DISTRIBUTED PROCESSING According to parallel distributed processing (PDP) or the connectionist model (McClelland, Rumelhart, and PDP Research Group, 1986; Rumelhart, McClelland, and PDP Research Group, 1986) of cognitive psychology, information processing takes place through the interactions of a large number of simple processing elements. The connections between elements of information can be active at the same time and this enables the system to manipulate a large number of cognitive operations simultaneously. The PDP model conjectures that parallel distributed processing occurs through a network distributed across incalculable numbers of locations in the brain. The Goldman-Rakic (1988) hypothesis on parallel sensory−cognitive processing, Mesulam’s (1990) distributed processing in large-scale neurocognitive networks, Fuster’s (1995) cortical memory, and Basar ¸ ’s (1998 and 1999) theory of oscillatory neural assemblies are psychophysiological counterparts of the PDP model. Karakas¸ et al. (2000) commented on parallel processing: In the formulation of our research group, parallel distributed processing is based on the oscillatory activity, the EEG and event-related oscillations (EROs) of various frequencies; each oscillation represents multiple functions and, conversely, a given function is represented by multiple oscillations. The cognitive functions are represented by integrative activity of neuroelectric oscillations that occur in parallel (Basar ¸ , 1998 and 1999; Basar ¸ and Karakas¸, 1998; Quiroga and Schürmann, 1998).

1.4 EEG−BRAIN DYNAMICS Electroencephalogram−brain dynamics can be defined simply as measuring electrical activity (or magnetic fields) of the brain recorded from large numbers of neural populations selectively distributed in the whole brain by using large scalp or intracranial electrodes approximately 100 µm in diameter. It seems clear that this method is one of the fundamental approaches to understanding integrative brain functions. However, scientists working in the field of brain macrodynamics had a long way to go before the relevance of EEG and MEG studies for elucidating brain functioning became clear. Although scalp EEG activity was measured by Hans Berger in the 1920s and Lord Adrian (1942) initiated basic research with EEG oscillations, functional EEG research remained in the shadows of the single cellular level until the 1980s.

1.4.1 IMPORTANCE

OF

EEG STUDIES

Figure 1.1 illustrates new approaches and strategies in functional neuroscience. The utility of the ensemble of methods is emphasized because the application of a single method has severe shortcomings for elucidating integrative brain functions. The methods range from indirect means of measuring changes in cerebral blood flow in local regions of the human cortex with fMRI to measuring changes in electrical activity via EEG recordings of the human brain with multiple electrodes and surgically implanted multiple electrodes in primates. According to Mountcastle (1998), measurement of large populations of neurons is presently the most useful experimental paradigm used in perception experiments. However, fMRI has the © 2004 by CRC Press, LLC

TF1663_C01.fm Page 8 Thursday, May 13, 2004 2:24 PM

APPLIED DOMAINS _________ EVOLUTION OF BRAINS

Single Unit Studies ?

AGING PATHOLOGY (e.g., Alzheimer’s disease, Parkinson’s disease, epilepsy, multiple sclerosis) + Biochemistry Pharmacology

Mathematical and Technical strategies Chaos, Neural Networks

PET

EEG/ERP MEG/MEF [Event-Related Oscillations] EROs

BASIC FUNCTIONAL LEVEL __________ SENSORY DETECTION

fMRI

PERCEPTION COGNITION MEMORY MOVEMENT

Psychophysiology Attention, Perception Learning and Memory Paradigms and Tests

ATTENTION

FIGURE 1.1 New approaches and strategies in functional neuroscience.

disadvantage of low temporal resolution and long distance measurements with multiple microelectrodes cannot be yet performed. Therefore, measurements of macro-activities (EEG, ERP, and MEG) seem to be the most adequate methods of measuring the dynamic properties of memory and integrative brain function. Since neuroscientists have concluded generally that several different brain regions must cooperate to accomplish any brain function, the analysis of the relationships of different regions of the brain is becoming more important. We will now discuss the methods and strategies cited in Figure 1.1. Strategy is defined as combined (parallel or sequential) applications of several methods. Studies at single-cell level have been of great importance in eludicating the basic physiological mechanisms of communications between cells (Mountcastle, 1998; Eccles, 1973). However, the importance of these studies for understanding of integrative brain functions is questionable because the whole brain is involved during integrative processes, as Ross Adey (1960, 1966, and 1989) noted and the new trends in neuroscience clearly emphasize (see Freeman, 1999). Positron emission tomography (PET) is an invasive method that allows large temporal resolution within 30 min, but offers no possibility of dynamic measurements within microseconds. The methods incorporating analyses of EEG, ERP, and EROs with fMRI provide additional strategies to illuminate brain functions since they cover dynamic changes in the brain and morphological structures. MEG and studies of event-related magnetic fields (MEFs) greatly increase spatial resolution in comparison to EEG and ERP. Accordingly these methods show great promise in future applications. The new strategies are interwoven with relevant use of mathematical and psychophysiological strategies including: 1. Theoretical mathematical and systems approaches such as (a) chaos, entropy, (b) modelling with neural networks, (c) a frequency domain approach combined with wavelet analysis and spatial and temporal coherence (Bullock, 1989; Petsche, 1998; von Stein, 2000). 2. Psychological strategies involving behavioral paradigms and application of neuropsychological tests (Karakas¸ et al. 2002 and 2003). One important strategy not cited in Figure 1.1 is recording of data via surgically implanted intracranial electrodes in animal brains. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 9 Thursday, May 13, 2004 2:24 PM

To achieve relevant progress in functional neuroscience, it is essential to combine several methods (Freeman, 1999) although application of all strategies in every laboratory is impossible. Figure 1.1 illustrates levels of basic central nervous system (CNS) functions (right side) and applied domains (left side). Sensory detection, movement, and memory functions can be analyzed successfully by using individual methods or strategies in research domains that study evolutionary developments, aging, pathology, and pharmacology (use of drugs or pharmacological agents to treat pathologies). Applications of combined strategies at basic scientific levels and in all these specialized fields may reveal new horizons for understanding integrative functions of the brain and especially memory function. The importance of memory and its influence on behavior cannot be overly emphasized because few aspects of higher nervous functioning could operate successfully without some memory contribution. Perception, recognition, language, planning, problem solving, and decision-making abilities all rely on memory (Damasio and Damasio, 1994).

1.5 PIONEERING STUDIES OF BRAIN MACRODYNAMICS AND WHOLE BRAIN APPROACH The domain of mechanics that involves the motions of bodies without reference to the causes of motion is called kinematics; the domain that studies the resulting motions is called kinetics. These two domains constitute the field of dynamics, also known as Newtonian dynamics: The dynamic activities of the brain involve mutual influences of bodies as reflected by kinetics. The analysis of trajectories reflecting the activities of neuronal populations is somewhat similar to the analysis of motion. Accordingly, we will use brain dynamics to describe the causes or mechanisms that give rise to trajectories manifested as electrical signals from the brain. The early studies of Lashley (1929) and Hebb (1949) established important principles of memory and integrative brain function. Between 1960 and 1971, several biophysicists and neuroscientists including Fessard (1961), Griffith (1971), and Rosen (1969) published relevant studies indicating the transition from single neuron dynamics to the dynamics of neural populations. The finding of this transition by a group of scientists opened a new era of behavioral neuroscience based on the oscillatory dynamics of neural populations. The important issue of oscillatory brain dynamics at the functional level is explained in several books and papers (Basar ¸ et al., 2003).

1.5.1 GRIFFITH: STATISTICAL MECHANICS

IN

BIOLOGY

AND

PHYSICS

Griffith (1971) discussed the concepts of statistical neuron dynamics and tried to formulate a similarity between statistical mechanics and neurodynamics as follows: The situation is superficially very similar to that which is obtained in statistical mechanics, as it applies to the relation between macroscopic thermodynamic quantities and the underlying microscopic description in terms of the complete specification of the states of all the individual atoms or molecules …. These are, first, that we could not, even if we knew all the necessary parameters, actually solve in detail the 1010 or more coupled neuronal “equations of motion” necessary to follow the state of the system in detail as a function of time. Second, that there exists a simpler “macroscopic” level of description which is really our main ultimate object of interest so that we do not wish, even if we could, to follow the “microscopic” state in detail but merely wish to use it to understand the time development of the macroscopic state. One most important aspect of this is that we only wish to specify, at the macroscopic level, the initial conditions of any calculation we may make. This leads immediately to the problem of whether the fundamental assumptions of equal a priori probabilities and random a priori phases hold for nerve cell aggregates, and, if not, whether we can find anything to replace them.

Griffith’s remarks are more important now than they were 30 years ago because new trends in brain research clearly indicate the need to introduce new frameworks to analyze integrative brain functions by studying cell aggregates rather than single cells. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 10 Thursday, May 13, 2004 2:24 PM

1.5.2 ROSEN: GLOBAL NEURODYNAMICS In 1969, Rosen questioned the role of statistical mechanics in gas dynamics? The physics laws describing gas dynamics are based on an ensemble of molecules in an isolated system. One does not describe gas dynamics based on single molecules in an isolated system. After the laws were determined experimentally, attempts were made to correlate the macrosystem laws with dynamics at the microlevel, i.e., with gas molecules. In other words, the laws of gas dynamics were determined before they were correlated with molecular properties. This principle is complementary to Griffith’s statements cited above. Basar ¸ (1980 and 1998) commented on the works of Rosen and Griffith: In the analysis of brain waves, we are certainly interested to discover the particular properties of individual neurons and their relation to the gross activity. To further examine the problem of the correlation of single unit activity (microactivity) and gross activity (macroactivity), Rosen (1969) explained the concepts of statistical mechanics and physics and their relation to neurobiology: What is the micro-description? We know that here the fundamental state variables are the displacements and momenta of the individual particles which make up our system. According to Newtonian dynamics, the kinetic properties of the system are given by the equations of motion of the system, which express the momenta as functions of the state variables. The basic postulate of Newtonian dynamics is the following point: knowing the state variables at one instant and the equations of motion, we are supposed to be able to answer any meaningful question that can be asked about the system at any level. Statistical mechanics however identifies a macrostate with a class of underlying microstates, and then expresses the global state variables as averages of appropriately chosen micro-observables over the corresponding class of microstates.

1.5.3 FESSARD: GENERAL TRANSFER FUNCTIONS

OF THE

BRAIN

Fessard (1961) emphasized that the brain must not be considered simply as a juxtaposition of individual lines, leading to a mosaic of independent cortical territories, one for each sense modality, with internal subdivisions corresponding to topical differentiations. What are the principles that dominate the operations of hetero-sensory communications in the brain? The answer requires extensive use of multiple microelectrode recordings along with a systematic treatment of data by computers (Gray and Singer, 1989; Elkhorn et al. 1988). Fessard indicated the necessity of discovering principles that govern the most general — or transfer — functions of multiunit homogeneous messages through neuronal networks. The transfer function describes the ability of a network to increase or impede transmission of signals in given frequency channels. The transfer function represented mathematically by frequency ˘ et al., 1992) constitutes the main framework characteristics or wavelets (Basar ¸ , 1980; Basar ¸ -Eroglu for signal processing and communication. The existence of general transfer functions could then be related to a series of networks having similar frequency characteristics to facilitate or optimize signal transmission in resonant frequency channels in the brain (Basar ¸ , 1998). In an electric system, optimal transmission of signals is reached when subsystems are tuned to the same frequency range. Does the brain have such subsystems tuned to similar frequency ranges or do common frequency modes exist in the brain? The empirical results reviewed here imply a positive answer and provide a satisfactory framework to Fessard’s question. Frequency selectivities in all brain tissues containing selectively distributed oscillatory networks (delta, theta, alpha, beta, and gamma) constitute and govern mathematically the general transfer functions of the brain. To fulfill Fessard’s hypothesis, all brain tissues of mammalians and invertebrates would have to react to sensitive and cognitive inputs with oscillatory activities or similar transfer functions. The synchronies, amplitudes, locations, and © 2004 by CRC Press, LLC

TF1663_C01.fm Page 11 Thursday, May 13, 2004 2:24 PM

durations or phase lags vary continuously, but similar oscillations are most often present in activated brain tissues (Basar ¸ , 1999). The general transfer functions of the brain manifested in oscillations strongly indicate that frequency coding is one of the major candidates of brain functioning, as noted earlier. We will further discuss Fessard’s work in Section 6.8.2.

1.5.4 EDELMAN: REENTRANT SIGNALLING THEORY

OF

HIGHER BRAIN FUNCTION

G.M. Edelman (1978) posed important questions about brain functioning. Does the brain operate according to a single principle carrying out its high-order cognitive functions? That is, despite manifold differences in brain subsystems and the particularities of their connections, can one discern a general mechanism or principle required for the realization of cognitive facilities? If so, at what level — cellular, molecular, or circuit — does the mechanism operate? By means of EEG analysis we can try to determine general mechanisms or principles at the levels of neural populations or circuits of cells. We may also open other important avenues and find additional laws, as Rosen and Griffith did. More important is Fessard’s question about general transfer functions of signal communications in the brain. Mountcastle (1976) noted that the central problem of brain physiology was how to understand the actions of large populations of neurons, actions that may not be wholly predictable from properties of subsets. He also noted that the central problem of the intrinsic neurophysiology of the cerebral cortex was to discover the nature of neuronal processing with the translaminar chains of interconnected cells (in columns). Edelman (1978) transformed these statements by noting that the main problem of brain physiology was “to understand the nature of repertoire building by populations of cell groups.” As noted earlier, EEG oscillations represent primitive study methods and should be considered as building blocks for further techniques. Edelman (1987) developed a theory of neuronal group selection. This theory assumes a genetic endowment of neuronal groups such as the columnar modules of the cortex with inherent degrees of variability and plasticity in their connections. They constitute the units of selection of the primary repertoire. By exposure to external stimulation and a Hebbian mechanism, certain groups of cells that tend to fire together will be selected by stimuli insofar as groups respond to them, and thus their connections will be strengthened (Figure 1.2). Some of those connections will make recurrent or re-entrant circuits that are essential features of the model and of its theoretical and computational elaborations (Tononi et al., 1992). Groups not selected will be crowded out by the competition. According to Edelman, re-entry is dynamic and can occur via multiple parallel and reciprocal connections. It takes place between populations of neurons rather than between single units. Neurons within a group tend to be strongly connected. At higher levels, the integration of perceptual and conceptual components is required to categorize objects. See also Chapter 9 and Chapter 10.

1.6 FREEMAN, KATSCHALSKY, AND HAKEN: PRELIMINARY STEPS IN INTRODUCING MACRODYNAMICS OF ELECTRICAL ACTIVITY The mechanisms of self-organization via oscillations through various kinds of interactions in physical, chemical, biological, psychological, and social systems have been most deeply explored by Aharaon Katzir-Katchalsky (1974) and Ilya Prigogine (1980) in studies of dissipative structures and chaotic state transitions. According to Prigogine’s theory, no system is structurally stable; fluctuations lead to instabilities and two new types of functions and structures. The evolution of a dissipative structure is a self-determining sequence. Scheme 1.1 shows the relationships of function, structure, and fluctuation. This approach combines both deterministic and probabilistic elements in the time evolution of the macroscopic system. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 12 Thursday, May 13, 2004 2:24 PM

S1

S2

S3

R1

R2

R3

Si

Rk

Rj

FIGURE 1.2 Edelman’s principle of group selection. At birth, a primary repertoire of responses (R1, R2, R3, …) by cortical neuron groups can be potentially elicited by any of a series of stimuli (S1, S2, S3…). After learning or repeated experience, a given stimulus, Si, will elicit many or only one of those responses RK. (Modified from Edelman, G.M., Neural Darwinism, Basic Books, New York, 1987.)

Freeman’s viewpoint (1999) is that complex biochemical feedback pathways within cells support the emergence of oscillations at cycle durations of minutes, hours, and days and underline the recurrence patterns of normal cyclical behavior as well as epilepsies, mood disorders, and other pathologies. A large number of neurons form macroscopic populations under the influence of external and internal stimuli and endogenous neurohormones. Freeman’s opinion is that these populations are more closely related to the nerve cell assemblies conceived by Hebb (1949). The relationships of the neurons to the mass are explained by Haken’s synergetic theory (1977) whereby microscopic neurons contribute to the macroscopic order and then are “enslaved” in a manner similar to the containment of particles in lasers and soap bubbles. Freeman (1975) achieved an important step in understanding the dynamics of populations of neurons (macrosystems) and also finding correlations between the activities of single neurons and population responses by starting with induced gamma activity in the olfactory bulb of the rabbit. Our group tried to determine the dynamics of brain responses in an abstract way, then tried to show, based on existing neurophysiological data, what particular neural responses could give rise to the general transfer functions cited by Rosen and Fessard (see Basar ¸ , 1980 and 1999). Oscillatory responses and resonance phenomena in the alpha, beta, theta, delta, and gamma frequency ranges govern the brain dynamics as revealed by macroscopic brain activity. Resonance is the response that may be expected of underdamped systems when a periodic signal of a characteristic frequency is applied to the system. The response is characterized by surprisingly large output amplitude for relatively small input amplitude. We were looking for codes related to general dynamical rules and links between macrodynamics and microdynamics and between brain oscillations and functions. Our research and experimental foundations are described by Basar ¸ et al. (2004). The alliance of perception and memory based on concepts of Hayek (1952) is described with electrophysiological measurements and systems theory tools. The building of a general framework of macroscopic brain dynamics led to useful categorization of integrative brain functions. In order to find codes and general dynamic rules in the sense of Rosen (1969) and Fessard (1961), the biological systems analysis program was applied. This will be explained in the next section.

© 2004 by CRC Press, LLC

TF1663_C01.fm Page 13 Thursday, May 13, 2004 2:24 PM

1.7 APPLICATION OF PRINCIPLES OF BIOLOGICAL SYSTEM ANALYSIS TO BRAIN RESEARCH 1.7.1 REASONS

FOR

ESTABLISHING PROGRAMS

FOR

BRAIN RESEARCH

In the 1970s, Basar ¸ and coworkers tried to determine the dynamics of brain responses in an abstract way and named their approach a Program For Biological System Analysis. They tried to show, based on existing neurophysiological data, what particular neural responses could give rise to general transfer functions. The program was extended and modified in 1998 and designated the Brain Dynamics Research Program (Basar ¸ , 1976, 1980, and 1998). In the meantime, a number of other research groups applied some of the steps or the global concept of the program. In addition to the classical analysis tools of general systems theory, some supplementary experimental methods and methods based on the special natures of living systems are parts of this program. The program has three main classes of methods: (1) abstract methods of general systems theory, (2) specific methods for living systems, and (3) methods of thoughts and research principles. Figure 1.3 illustrates a more advanced version of a biological systems analysis and brain dynamics research program. The rationale for developing a research program was based on elucidating the black box (brain). Three basic quantities involved in biological investigations are the input (stimulus), the system, and the output (response). If the stimulus and response are known or are measured variables, it should be possible to estimate the properties of the system (the whole brain). The determination of the abstract frequency characteristics or transfer function of the system under study usually causes experimental and sometimes conceptual difficulties. This is partly due to rapid changes of the parameters measured. Mathematical representations merely help identify the frequency positions of all components without determining the exact natures of the components. At this stage, a researcher must elucidate the black box. Since the determination of mathematical characteristics alone did not allow concluBRAIN DYNAMICS RESEARCH PROGRAM I. ABSTRACT METHODS FOR SYSTEM ANALYSIS 1. a) b) c) d)

Methods to analyze brain states Power spectral density Cross correlation Cross spectrum Coherence

2. a) b) c) d) e)

Methods to analyze evoked brain activity Transient response analysis Frequency analysis Response adaptive filtering Combined EEG-EP analysis Evoked coherence

II. SPECIFIC METHODS FOR ANALYSIS OF THE BRAIN FUNCTION 1. Application of pharmocological agents

III. METHODS OF THOUGHT OR RESEARCH PRINCIPLES 1. Going into the system 2. Going out of the system

2. Selective blocking ot the system

3. New emerging methods to analyze eventrelated oscillations a) Wavelet analysis b) Wavelet entropy c) Single sweep wave identification d) Event related oscillations e) Study of nonlinearities and chaos approach

3. Reduction of the system into its passive response

3. Consideration ot the system as a whole

4. Application of various paradigms to influence the state of consciousness and alertness (attention, learning etc.) 5. Paradigms with complex gestalts, as “grandmother face”

A new framework to extend the Neuron Doctrine, considering the brain as a whole: “Neurons Brain Theory” related to whole brain work

FIGURE 1.3 Brain dynamics research program. (Modified and extended from Basar, ¸ E., Brain Function and Oscillations, Vol. 1, Brain Oscillations: Principles and Approaches, Springer, Berlin, 1998, p. 153.) © 2004 by CRC Press, LLC

TF1663_C01.fm Page 14 Thursday, May 13, 2004 2:24 PM

sions about the biophysical nature of the phenomenon, the difficult problem was to establish the biological systems analysis theory. The application of this program has followed the lines of thinking of Fessard, Griffith, and Rosen to develop transfer functions and the thinking of Hebb related to long distance coherence of macroscopic electrical activity of the brain. Investigators in the field of brain studies usually deal with gray boxes (partially elucidated black boxes). An apparatus or system is designated a gray box when it performs a defined operation and provides information about the structure or processes making possible (realizing) the defined operation. A gray box generates partial information concerning the structures and processes that realize input−output relations (Basar ¸ , 1998). In the context of the general framework of the brain dynamics research program, we developed certain research principles or strategies that allowed us to add to our knowledge about brain functioning. In fact, every neuroscientist has his own surroundings and develops his own definitions and classifications of signals studied. This approach has helped to expand our knowledge of global brain dynamics and global brain functions as reflected by EEG and oscillatory brain responses. 1.7.1.1 Program Steps The ensemble of abstract methods of brain state analysis shown in Figure 1.3 includes (1) power spectral density, (2) cross-correlation, (3) cross-spectrum, and (4) coherence. Combined EEG and EP analysis and wavelet analysis methods are also used to analyze brain activity. Basar ¸ ’s group first used conventional methods to apply abstract techniques to brain wave analysis. The group later performed studies on event-related oscillations (EROs) using abstract methods including long distance coherence and new methods such as wavelet entropy (Rosso et al., 2001; Quiroga et al., 1999). The third group of abstract methods shown in the figure includes emerging methods for analyzing EROs. The study of nonlinearities and chaos approach aims at understanding additional properties of the system. Specific methods for analysis of brain function included application of pharmacological agents and blocking of the system. Most importantly, the applications of different paradigms produced very interesting results and formed the frameworks of studies for complex gestalts such as the grandmother cell and similar techniques (Basar ¸ , 2003; also see Chapter 8, this volume). Treating the brain as a system means that the brain consists of a collection of components or subsystems arranged and interconnected in a definite way. One possible approach to understanding the brain system as an entity is to isolate the subsystems and study their specific properties. As a next step, one should understand how the subsystems are interconnected and which specific relations determine their integrative functioning. After determining subsystems and their interrelations, the next step is to try to model and reconstruct the whole entity. Abstract methods and their analogues used to analyze living systems aim at isolating distinct components. This approach is informative and defines a strategy generally called going into the system.* The conceptual framework provides us another, far more important, research strategy that cannot be realized by any of the analysis methods available or by their combined application. This strategy is called going out of the system and is defined as a method of thought. It is well known, for example, that a word is an abstract representation extracting the most essential attributive features from an enormous group of single concrete objects. In a similar manner, by using the method of thought, one can approach the essential principles of brain functioning by removing specific concrete representations and simultaneously extracting common building units. This can be achieved by going out of the system.** The principle of going out of the system is important for comparing the anatomy and physiology of the brains of humans and invertebrates, for example. This was the essential step undertaken by * This strategy is the search of the microstructure. ** This is the comparison of the properties of the analyzed system with those of other systems. Example: comparison of the circulatory system of the kidneys with the circulatory system of the brain.

© 2004 by CRC Press, LLC

TF1663_C01.fm Page 15 Thursday, May 13, 2004 2:24 PM

Darwin in establishing comparative biology. Going out of the system involves another important comparison. Analyzing the frequency responses of the cortex, hippocampus, and other structures within the same brain can provide important information regarding parallel processing, thus contributing to our knowledge of fundamental building blocks. We should also consider interpretation of results obtained from investigation of cellular and structural systems. Several research groups derived some of the most accepted consequences through application of the brain dynamics research program. The functional significance of oscillatory neural activity began to emerge from the analysis of responses to well-defined events (EROs phase- or time-locked to a sensory or cognitive event). It is possible to investigate such oscillations by frequency domain analysis of ERPs based on the following hypothesis. An EEG analyzes the activities of an ensemble of generators producing rhythmic activities in several frequency ranges. These oscillators are active usually in a random way. However, the application of sensory stimulation to these generators enables them to couple and act together coherently. This synchronization and enhancement of EEG activity gives rise to evoked or induced rhythms. Evoked potentials representing ensembles of neural population responses were considered the results of a transition from a disordered to an ordered state. A compound ERP manifests a superposition of evoked oscillations in EEG frequencies ranging from delta to gamma. Natural frequencies of the brain range include alpha (8 to 13 Hz), theta (3.5 to 7 Hz), delta (0.5 to 3.5 Hz), and gamma (30 to 70 Hz). (See Yordanova and Kolev, 1998 and Chapter 6, this volume.) These statements clearly indicate that the macrodynamics of the brain are governed by oscillatory EEG dynamics that provide important keys to understanding brain function. The concerted application of all three steps of the brain dynamics research program led to a new framework, namely the neurons−brain theory that will be explained in the next section. The development and achievements of the theory and the grandmother gestalt experiments discussed in Chapter 8 are based mainly on the implications of this program and extend the neuron doctrine to consideration of the brain as a whole. The brain dynamics research program methods provided conventional tools, continuously developing principles, and new applications. The program also provided a wide spectrum approach. It did not limit the research field to a single frequency (e.g., 40 Hz) window and allowed us to pursue the super-synergy concept. New methods such as wavelet entropy studies are also applicable within this framework (Quiroga et al., 1999 and 2001; Rosso et al., 2001). 1.7.1.2 Mathematical Methods of Program Mathematical methods are covered by Basar ¸ (1998) and Basar ¸ et al. (2001d). A vast amount of literature discusses the analysis of chaos, wavelets, and wavelet entropy. We introduce amplitude frequency characteristics in the Appendix at the end of this book because the method is not yet covered in the literature.

1.8 NEW APPROACHES TO BRAIN FUNCTIONING AT MACROSCOPIC LEVEL 1.8.1 SHERRINGTON’S NEURON DOCTRINE REVISITED Studies of Ramon Cajal (1911) related to neuron morphology and the physiological approach of Sherrington (1948) led the way to the single-neuron doctrine with the notion of one ultimate pontifical nerve cell that integrated CNS function. In this concept, the integration was related to motor activity; the functional mapping was a type of movement mapping. Memory and cognitive functions were not interwoven in the physiological descriptions. Horace Barlow (1972 and 1995) transformed and replaced the first interpretation by specifying a feature of an object, such as a line, color, or tone represented by the firing of a neuron. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 16 Thursday, May 13, 2004 2:24 PM

In the first half of the 20th century, the invention of the EEG was followed by an explosion of publications related to brain function. The invention renewed the hope of tapping a physical correlate of mental performance (psychic energy described by Berger, 1929). The principles introduced by Berger and experimentally supported by Adrian (1941) remained in the shadows of neurophysiology research limited to the single-neuron approach. Mountcastle (1992) said: Suddenly a paradigm change is upon us, considering slow oscillations as active agents for signal transmission … stimulusinduced slow wave oscillations are related to/are signs of/generate such complex brain functions as perception, execution of movement patterns, or storage in memory — in short what is called cognitive neuroscience. The new developments demonstrate that it is not possible to interpret the functional contributions of alpha, theta, and delta, and gamma responses with only the neuron doctrine originally proposed by Sherrington (1948). The generators giving rise to these frequency responses are extremely sensitive to the modalities of sensory and cognitive inputs. Tracking properties of functionally related distant single neurons is not yet possible because of technical limitations. Goldman-Rakic (1988 and 1997), in search of a topography of cognition, concluded: If subdivisions of limbic, motor, sensory, and associative cortex exist in developmentally linked and functionally unified networks, as the anatomical, physiological, and behavioural evidence suggests, it may be more useful to study the cortex in terms of information processing functions and systems rather than traditional but artificially segregated sensory, motor, or limbic components and individual neurons within only one of these components. These new developments followed the proposals of Griffith (1971) and Rosen (1969) and the concepts of unification of functional networks of Hayek (1952), and supported the renaissance of use of the EEG in functional neuroscience.

1.8.2 RENAISSANCE

OF

EEG USE

IN

SEARCH

OF INTEGRATIVE

BRAIN FUNCTIONS

A special issue of the International Journal of Psychophysiology (Basar ¸ , Hari, Lopes da Silva, and Schürmann, 1997) and a new book dedicated to Hans Berger (Basar ¸ , 1999) described a quiet revolution in neuroscience. In the previous decade, increasing numbers of brain scientists employed approaches using oscillatory components of event-related potentials, EEG, and MEG. When we recall the important remark made by Mountcastle (1998) about a possible turnaround in the analysis of the brain field potentials and their importance, we can imagine that this field will grow remarkably and the new millennium will be witness to a new era in brain research in which EEG oscillations together with complementary brain imaging techniques will become focal points for new discoveries. An important step in this revolution is the fact that the neuroscience experimenters started to consider the brain as an integrative system and no longer limited their analyses to results from a special structure or application of a single paradigm only. This change may have arisen from the availability of powerful computers and algorithms in the analysis of EEG events and new systems theory tools. Another important achievement is the parallel use of whole-cortex MEG and fMRI. The experimenter who chooses these algorithms to explore higher level nervous activities should essentially accept the EEG components as the most important functional building blocks of the brain both at the cellular and neural assemblies levels. We have probably witnessed more than a paradigm change in neuroscience as predicted by Mountcastle (1992). Neuroscientists are now able to attack a core problem: understanding brain functioning by means of its natural frequencies or EEG oscillations. The sudden paradigm change indicated by Mountcastle (1992) is interwoven with the discoveries of the activities of neural populations via new brain imaging techniques. These new techniques opened the way toward new ideas about renewing or extending the single neuron doctrine. Freeman (1975) designated the theory of using dynamics of neural masses as the new Sherrington doctrine, in which neural populations play a significant functional role. Roy John (1988) described a © 2004 by CRC Press, LLC

TF1663_C01.fm Page 17 Thursday, May 13, 2004 2:24 PM

hyperneuron consisting of neural populations as a functionally important entity of the brain. Barlow (1972) and Mountcastle (1992 and 1998) proposed modern views of Sherrington’s doctrine. Szenthágothai’s well-known illustration of a 300-µm diameter cortical module (1983) is one of the important examples of neural modules and local nerve circuits — an ensemble that plays a significant functional role. Mountcastle (1976) defined the basic function unit as a minicolumn approximately 30 µm in diameter and containing 100 to 300 neurons. Larger processing units called macrocolumns contained up to several hundred minicolumns. Mountcastle said: Prominent among them is the concept that the brain is a complex of widely and reciprocally interconnected systems and that the dynamic interplay of neural activity within and between these systems is the very essence of brain function. The large entities of the brain are composed of replicated modules. The linked sets of modules of the various brain entities comprise a distributed system (see Chapter 6). Another important trend in describing integrative brain activity and memory involving functionally and selectively distributed neural networks started with the publications of Goldman-Rakic (1997), Fuster (1995), and Mesulam (1990 and 1994). In line with their proposals, Basar ¸ et al. (2004) surveyed results of functional oscillatory activities compiled by more than 100 laboratories in the last 20 years at the cellular, field potential, and EEG−MEG levels. Depending on the regimes or states of the brain, the limbic system, brainstem, thalamus, and cortex are all involved with 2-Hz, 4-Hz, 10-Hz, and 40-Hz firing or with all of them. The new trends imply that the following are involved in integrative brain function: 1. Single neurons and also neural assemblies 2. Spikes of single neurons and also oscillatory activity of neurons and neural assemblies 3. Movements and also cognitive and memory processes Sherrington’s (1948) description of integrative brain activity preceded the empirical results that emerged in the past 20 or 30 years.

−BRAIN THEORY: AN APPROACH THAT INCLUDES 1.9 NEURONS− WHOLE BRAIN ORGANIZATION 1.9.1 TOPOGRAPHY

OF

COGNITION

AND

ELEMENTS

OF

NEURONS–BRAIN THEORY

In Mesulam’s model of cognition, the formation of specific templates belonging to objects and memories occurs as selectively distributed processing with considerable specialization. This functional selectivity exists in anatomically differentiated localizations. The view of Fuster (1995 and 1997) is that memory reflects a distributed property of cortical systems. Accordingly, it can be hypothesized that selectively distributed oscillatory systems may provide a general communication framework for morphology and may be very useful for functional mapping of the brain (Basar ¸ et al., 1999). The neurons−brain theory was based on the concepts mentioned above and empirical findings reviewed by Basar ¸ et al. (1999 and 2001), Bressler (1990), and Gruzelier (1996). It aims to partially replace and extend Sherrington’s neuron doctrine for exploring integrative brain functions manifested with activities of neural populations. New rules to describe brain functions by means of neural populations instead of single neurons were developed: 1. The neuron is the basic signaling element of the brain. Oscillatory activities of the brain (gamma, alpha, beta, theta, and delta) reflect natural frequencies and/or real responses (Basar ¸ et al., 2001). 2. Neural assemblies replace neurons in descriptions of integrative brain functions; this view diverges from Sherrington’s neuron doctrine. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 18 Thursday, May 13, 2004 2:24 PM

3. The EEG is a quasi-deterministic or chaotic signal; it is not always noise. Electroencephalogram oscillatory activities govern most general transfer functions in the brain. 4. Selectively distributed oscillatory neural populations are activated upon sensory stimulation or event-related tasks by manifestation of: a. Enhancements or resonances b. Delay of oscillations c. Blocking or desynchronization of oscillations d. Prolongation of oscillations e. Increases or decreases of coherences f. Increases or decreases of entropies 5. According to published data, parallel processing also functions selectively. Oscillatory systems showed varied degrees of coherence (Basar ¸ , 1980; Kocsis et al., 2001; Schürmann et al., 2000). 6. Types of neurons do not play a major role for frequency tuning of oscillatory networks since morphological different neural networks are excitable and communicate with the frequency codes of EEG oscillations. 7. Functions in the brain are manifested by varying degrees of superpositions of oscillations in EEG frequency ranges. Accordingly, neuron assemblies do not obey the all-or-none rule of the single neuron doctrine. 8. Although the existence of feature detectors has been demonstrated, integrative brain functioning needs the synergy of selectively distributed and selectively coherent neural populations. The role of the feature detectors is well described (Sokolov, 2001; Chapter 7, this volume). 9. Integrative functions in the brain are manifested by varied degrees of coherences. 10. A strong inverse relation exists between prestimulus oscillations and brain responses; spontaneous activity of the brain is effective as a control parameter. Do the super-synergy and superbinding concepts allow us to build a bridge to interpret manifestations of integrative brain functions? As a corollary to the brain−neuron theory, we recently (Basar ¸ et al., 2001 and 2003) we introduced the concept of supersynergy in brain electrical activity as an ensemble of at least six processes that act in synergy upon sensory−cognitive input. According to our hypothesis based on results of human and animal experiments, the electrical manifestations of integrative brain functions are shaped by: 1. The superposition of oscillations including the alpha, beta, gamma, theta, and delta bands. 2. Activation of two or more selectively distributed oscillations in gamma, alpha, theta, and delta bands upon exogenous or endogenous input; these activities are manifested with parameters such as enhancement, delay, blocking (desynchronization), and prolongation. 3. Temporal and spatial changes of entropy in the brain. 4. Temporal coherence between cells in cortical columns for simple binding. 5. Varying degrees of spatial coherence that occur as parallel processing over long distances. 6. Inverse relations of EEGs and EROs; prestimulus EEGs serve as control parameters. Recent experiments performed with the faces of a known grandmother and an anonymous person support the concept of EEG superbinding. We return to this concept in Chapter 7 and Chapter 8. This chapter mentions the neurons−brain theory and the supersynergy concept because they are almost prerequisites for a global reading of this book. However, the evolution of this framework can be better understood after carefully reading the experiments discussed in Chapters 3 through 6. Accordingly the new framework will be discussed analytically in Chapter 7. Chapter 11 aims to combine this chain of ideas into a theory. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 19 Thursday, May 13, 2004 2:24 PM

Suggestions and ideas by Fessard, Griffith, and Rosen; the principles of Hebb, Lashley, and Hayek; and the theory of Haken provided important tracks that converge in the supersynergy and superbinding concepts; the method developed by Hans Berger was the most important tool. Certainly, the cooperation between neural populations cannot elucidate phenomena at the synaptic level as described by Hebb. However the concepts presented in this book may bridge or possibly unify all previous views as a new proposal for scientists working at macroscopic levels of brain dynamics.

1.10 SIGNIFICANCE OF EEG BRAIN DYNAMICS IN MEMORY STATES AND INTEGRATIVE BRAIN FUNCTIONS The ERP is a compound neuroelectric signal that is rich in functional information (Bullock, 1993) and related to a large spectrum of activities ranging from single percepts to complicated memory processes. In the analysis of integrative brain functions, one must consider not only one specific ERP in a given brain structure, but must also take into account the interrelations of distributed ERPs due to strong parallel processing in the whole brain. Accordingly, it is necessary to analyze the entire brain in order to understand a specific function manifested by neurolectric activity of a given structure. For example, when we consider or analyze cognitive processes, the most marked ERPs are recorded usually in frontoparietal areas or in various association cortices. However, it is necessary to take into account recordings from other areas as well, e.g., from sensory cortices that may indicate parallel processing (Basar ¸ and Schürmann, 1994; Basar ¸ , 1998a and b). Remembering and memory are manifestations of various and multiple functional processes, depending on the complexity of the input to the CNS. The electrical response to a simple light flash is based on simple memory processes at the lowest hierarchical order. When we talk about a memory process — a short- or long-term one — we perceive a sensory input that is matched with information already stored in neural tissue. If a simple light evokes alpha and gamma responses, it is almost obligatory to assume that elementary oscillatory responses are also manifestations of several memory processes at different hierarchical levels. The topology of the memories, depending on the modality of the input, must be different (see examples provided by Basar ¸ and Schurmann (1994) and Basar ¸ (1998a,b) related to cross-modality experiments and measurements of cortical and subcortical structures). Such studies have rarely been performed and the results and their interpretations must be considered preliminary. Accordingly, multiple distributed memories cannot be treated in detail and classifications of all levels of distributed memories cannot be yet provided. In performing many complex tasks, it is necessary to retain information in temporary storage until it is needed to complete the task. The system used for this is called working memory (Baddeley 1996). Working memory is the temporary ad hoc activation of an extensive network of short-or long-term perceptual components that are, like any other perceptual memories, retrievable and expandable by new stimuli or experiences. Fuster states that working memory has the same cortical substrate as the kind of short-term memory traditionally considered the gateway to long-term memory. According to the functional descriptions above, a simple or complex light stimulation or a light stimulation or a light stimulation involving some task or event should evoke oscillatory responses with different time hierarchies. Our view is that functional or oscillatory network modules are distributed not only in the cortex but also throughout other parts of the brain. Several types of analyses are crucial to functional interpretation of ERPs: 1. The analysis of the stimulus. What can a stimulus evoke in the brain? It can evoke simple sensory percepts, complex sensory percepts, bimodal percepts, memory related functions, etc. © 2004 by CRC Press, LLC

TF1663_C01.fm Page 20 Thursday, May 13, 2004 2:24 PM

2. The analysis of ERPs should be performed in related or unrelated function-dependent areas. For example, if a complex semantic event or memory-demanding task is presented as stimulation to the brain, frontal and/or parietal recordings are considered to carry the most important information. In this case, it is very important to analyze ERPs recorded in the occipital cortex (an area thought to be less involved in high level cognitive processing). This shows what is missing in occipital ERPs in comparison to association areas or what is recorded additionally. These steps are analogous to the fMRI analysis mentioned earlier in this chapter. 3. Component analysis by means of EROs provides an advantage over conventional ERP analysis as, for example, the results of cross-modality measurements demonstrate. In occipital areas, auditory stimulation does not evoke 10-Hz responses, although ERPs are measured upon visual stimulation. This demonstrates the dependence of 10-Hz responses on visual perception. Accordingly, the spatial resolution of ERPs is highly increased. 4. Studies with single-cell recordings and fMRI indicate that memory networks are distributed. Although the ERPs and EROs do not provide the excellent spatial resolution of fMRI or the exact one-to-one locations of single-cell recordings, they have several outstanding advantages in memory research. 5. When compared with fMRI, the time resolution levels of ERPs and EROs are excellent, since it is possible to measure function-related neuroelectric changes within a few milliseconds. 6. In ERP studies, the neuroelectric or neuromagnetic recordings can be obtained in humans; this is almost impossible with single-cell recordings. Moreover, it is possible to apply simultaneous measurements with several recording electrodes in distant locations. This allows dynamic comparisons of various structures of the human cortex and diverse subcortical structures in animal brains. For example, immediate comparison of frontal theta or alpha activity with occipital activity is possible. 7. As Fuster (1997) noted, the brain has as many memory types as the number of percepts. The application of event-related oscillations for the analysis of working memory and for simultaneous analysis of perceptual memory is very useful as a complement to fMRI and single-cell studies. These remarks clearly show that the analysis of EROs fills an important gap in the analysis of selectively distributed percepts and memories. 8. Dynamic changes in the attention−perception−learning−remembering (APLR) alliance can be studied only with strategies involving EROs. This chapter describes concepts and frameworks developed since the 1920s, whereas Chapter 3, Chapter 4, Chapter 6, and Chapter 8 provide empirical data obtained by application of these concepts. The new data, of course, led to new theories and principles. The paradigm change in cognitive sciences emphasizes analyses of macrodynamics instead of microdynamics. Accordingly, the way is now open to establish a new conceptual framework or theory of neural populations to extend or replace Sherrington’s neuron doctrine and to include memory in the framework developed.

© 2004 by CRC Press, LLC

TF1663_C02.fm Page 21 Wednesday, May 12, 2004 7:51 PM

2 Concepts and Theories 2.1 MEMORY MACHINERIES 2.1.1 DYNAMIC MEMORY

AND

APLR ALLIANCE

This chapter will describe relevant research and categorizations that explain the theoretical frameworks and models by Baddeley, Damasio, Squire, Tulving, and especially Fuster. Definitions, classifications, and categorizations of memory are described in several books and reviews often with divergent opinions that usually reflect the experiences of the authors. In the present book, a model of dynamic memory is proposed. The dynamic changes arising during memory processes will be demonstrated by evaluation of EEG oscillations. The experimental onset of dynamic changes by means of reciprocal activation will be described in Chapters 3, 6, 7, and 8 and may lead to new perspectives for future research. Chapter 9 discusses the essential model derived from memory studies involving EEG oscillations. With this model, we aim to associate the evolving memory with reciprocal activations of the processes of attention, perception, learning and remembering that we call the APLR alliance.

2.1.2 STEPS

OF

MEMORY PROCESSING

According to Tranel and Damasio (1995), “The process of forming memory involves three basic steps: (1) acquisition, (2) consolidation, and (3) storage.” • • •

Acquisition is the process of bringing knowledge into the brain and into a first-stage memory buffer via sensory organs and primary sensory cortices. Consolidation is the process of rehearsing knowledge and building a robust representation of it in the brain. Storage is the creation of a relatively stable memory trace or record of knowledge in the brain.

In learning to recognize a new face, for example, an individual would consolidate information concerning its visual pattern and create a relatively permanent record of the pattern that would then be connected to other pertinent knowledge (the person’s name, the situation in which the individual met the new person, etc. (Tranel and Damasio, 1995)

2.1.3 ENCODING

AND

RETRIEVAL

Any system for storing information, whether biological or artificial, must be able to (1) encode or register information, (2) store it, preferably without much loss, and (3) subsequently access or retrieve that information. Baddeley noted that because the three stages are closely linked, it is difficult to isolate any phenomenon as exclusively occurring at a single stage. Nevertheless, this division into processing stages continues to be useful in helping explain the processing or operation of memory systems. Chapter 9 will show that the transitions between processing stages are dynamic processes. As experiments in Chapter 3 indicate, the attention, perception, learning, and remembering processes

© 2004 by CRC Press, LLC

TF1663_C02.fm Page 22 Wednesday, May 12, 2004 7:51 PM

are interwoven. Interplay among them is perpetual and they are not separable on a time axis (see also Baddeley, 1996). Human brains use dynamic records rather than static, immutable memory traces. For example, the record of a face an individual recognizes is a set of neuron circuit changes that can be reactivated rather than a “picture” stored somewhere in the brain. Dynamic records can be modified and in this way reflect evolving experience (Damasio, 1989 and 1994). Encoding is the initial processing of information to be learned or memorized. Immediate memory for arbitrary sequences of verbal material, such as sequences used in a digit span test, typically relies on encoding based on phonological or sound characteristics of the material. Retrieval is the process of reactivating knowledge in a way that will allow it to become an image in consciousness (as in recall and recognition) or translated into a motor output (movement of a limb, activation of vocal apparatus, autonomic activity). See Tranel and Damasio, 1995. New learning models using parallel distributed processing (PDP) or connection architectures that are assumed to closely simulate the parallel processing of the neural networks of the brain have once again raised the issues of interference effects and how the brain deals with interference (Ratcliffe, 1990; Rumelhart and McClelland, 1986). Parallel processing of distributed oscillatory systems with multiple frequency windows was first described by Basar ¸ in the 1980s by means of ERP experiments with animals. The experiments were later extended to humans (Basar ¸ , 1980 and 1999).

2.2 FRACTIONATION OF MEMORY Atkinson and Shiffrin (1968) developed the well-known scheme of fractionation of memory illustrated in Figure 2.1. According to their model, information from the environment enters a series of brief sensory registers that then pass on information to a short-term store. This temporary storage system plays a crucial role. Without it, information cannot be transferred into or from the third final component, the long-term store. Long-term storage is assumed to occur when information is transferred as described, with the probability that the transfer is a direct function of the duration of time an item resides in the short-term store.

2.2.1 LONG-TERM MEMORY

VERSUS

SHORT-TERM MEMORY

The classic explanation of encoding reflects the current view originally stated by Shiffrin and Geisler (1973): “The process of encoding is essentially one of recognition; the appropriate image or feature is contacted in long-term memory (LTM) and then placed (i.e., copied) in short-term memory (STM).” Complex cognitive processes such as speaking and thinking may also be described in terms of a close interaction between the working memory and LTM systems. The basic difference of the Shiffrin and Geisler model is that a sensory code is lacking and a code generated in STM, for example, during speaking, plans what the individual will say. The codes generated in STM trigger search processes in LTM to retrieve the relevant knowledge about the appropriate semantic, syntactic, and articulatory information. This idea is similar to Baddeley's (1986, 1992, 1997) concept of working memory comprising an attentional controller and central executive and subsidiary slave systems (Klimesch, 1999). Baddeley pointed out that despite its advantages, the described model rapidly encountered problems. The assumption that merely holding an item in a short-term system would guarantee learning proved difficult to sustain (Craik and Watkins, 1973). The addition of levels of processing seemed to give a better account of learning than the modal model. Even more problematic was the evidence for normal learning in patients with short-term store deficits (Shallice and Warrington, 1970). Such patients appeared to have remarkably few problems in coping with everyday life and this was interpreted as an argument against the importance of short-term memory as a crucial © 2004 by CRC Press, LLC

TF1663_C02.fm Page 23 Wednesday, May 12, 2004 7:51 PM

Environmental input

Sensory registers Visual Auditory Haptic

Short-term storage (STS) temporary working memory Control processes: Rehearsal Decision Coding Retrieval strategies

Response output

Long-term storage (LTS) permanent memory storage

FIGURE 2.1 Structure of memory. (Modified from Atkinson, R.C. and Shiffrin, R.M. (1968), in The Psychology of Learning and Motivation: Advances in Research and Theory, Spence, K.W., Ed., Academic Press, New York, p. 195.)

control center for cognition. Baddeley and Hitch (1974) therefore suggested abandoning the assumption of a unitary short-term store and suggested a multicomponent working memory system.

2.2.2 WORKING MEMORY Baddeley and Hitch proposed a division of the working memory model into at least three subsystems, as illustrated in Figure 2.2. An important part of the system is an attentional controller or central executive that forms an interface between long-term memory and two or possibly more slave systems. These subsystems constitute the capacity for the temporary storage of information with an active set of control processes that allow information to be registered intentionally and maintained within the subsystem. The visuo-spatial scratchpad or sketchpad specializes in maintaining visuo-spatial information; verbal information is held by the phonological or articulatory loop. The central executive is assumed to be responsible for the selection and operation of strategies and for maintaining and switching attention as the needs arise. It is assumed to be associated with the functional activities of the frontal lobes and thus sensitive to frontal atrophy and lesions (Baddeley, 1986; Baddeley and Wilson, 1988). According to Tranel and Damasio (1995), working memory is a transient type of memory processing on a time scale of seconds during which an individual can maintain “online” the relevant stimuli, rules, and mental representations required to execute a particular task (Baddeley, 1992; © 2004 by CRC Press, LLC

TF1663_C02.fm Page 24 Wednesday, May 12, 2004 7:51 PM

Phonological loop

Central executive

Visuo-spatial sketch pad

FIGURE 2.2 Working memory model. (Modified from Baddeley, A. and Hitch, G. (1974), in The Psychology of Learning and Motivation, Bower, G.A., Ed., Academic Press, New York, p. 47.)

Goldman-Rakic, 1987; Chapter 1, this volume). Working memory is used to bridge temporal gaps, that is, to hold representations in a mental workspace long enough to formulate appropriate responses to stimulus configurations or contingencies for which some or even all the basic ingredients no longer exist in perceptual space (Fuster, 1989; Goldman-Rakic, 1987). The concept of working memory in a way overlaps with assumptions about STM. Both are considered relatively transient and are thought to have limited capacities. Fuster noted (1995, 1997, and 2000) that working memory, also known as operant memory, is an operant concept of active memory and postulates (1995) that active memory is a state rather than a system of memory. Single neuron recordings in monkeys trained to perform working memory tasks identified components of a working memory circuit in the prefrontal cortex. In these studies, the neuronal processes related to task performance can be dissociated on a scale of milliseconds to seconds. During performance of a working memory task, as the stimulus is sequentially registered, stored for seconds, and then translated into a motor response, specific neural populations respond in characteristic ways. One class of prefrontal neurons responds to a visual stimulus as long as the stimulus is in view. In contrast, other prefrontal neurons are activated at the onset of the stimulus and remain active during the time the monkey must remember the location or features of an object (Fuster, 1995; Goldman-Rakic, 1988 and 1997).

2.3 DISTINCTION BETWEEN IMPLICIT AND EXPLICIT MEMORY STATES Another important categorization of memories (or memory states) is the distinction between implicit and explicit types. Squire (1992) distinguishes between declarative and nondeclarative memory — terms that more or less map onto the explicit and implicit terms (Figure 2.3). An early distinction was made between procedural and declarative learning. Procedural learning represented the acquisition of skills or “learning how.” Declarative learning involved the acquisition of facts or “learning that ….” (Squire, 1992; Baddeley, 1985). While many preserved learning capacities may be regarded as skills, regarding conditioning or stem completion as genuinely procedural seems to be stretching the term. According to Baddeley (1995), it has become increasingly clear that memory comprises not a single system, but consists of an alliance of interrelated subsystems. Empirical evidence for the distinction between long- and short-term memories began to emerge in the 1960s. Baddeley’s view (1995) bears repeating: One source came from the previously described evidence that immediate memory for verbal material appears to rely on phonological coding, while LTM appears to be semantically based. © 2004 by CRC Press, LLC

TF1663_C02.fm Page 25 Wednesday, May 12, 2004 7:51 PM

MEMORY

NONDECLARATIVE (IMPLICIT)

DECLARATIVE (EXPLICIT)

Facts (semantic)

Events (episodic)

Skills and habits

Priming

Nonassociative learning

Simple classical conditioning

FIGURE 2.3 Components of long-term memory. (Modified from Squire, L.R. (1992), J. Cognitive Neurosci., 4, 232.)

A second source of evidence came from the observation that certain tasks appear to have two components. If a subject is presented with a list of words for immediate free recall, there will typically be extremely good recall of the last few items presented (Glanzer and Cunitz, 1966). One interpretation of this result is to suggest that the last few items are held in a labile short-term store, whereas earlier items reside in LTM.

2.4 NONDECLARATIVE MEMORY Baddeley (1995) proposes a cluster of learning systems that have in common the fact that they are independent of episodic memory. This means that they are capable of accumulating information, but not of pulling out and identifying specific episodes. The episodic memory system has a remarkable capacity to associate previously unrelated events in a single trial. It can associate an event with a context, and hence locate it in time and place. In contrast, nondeclarative systems are specialized for accumulating information from the world, but incapable of keeping separate the individual episodes (Baddeley, 1994b). A number of different kinds of nondeclarative phenomena have been identified.

2.4.1 PHYLETIC MEMORY Perceiving is the classification of objects by activation of the associative nets that represent them in memory. It is reasonable to assume, as Hayek (1952) did, that to a large extent, memory and perception share the same cortical networks, neurons, and connections. To understand the formation and topography of memory, it is useful to think that the reaction ability of the primary sensory and motor areas of the cortex is called phyletic memory or memory of the species. The primary sensory and motor cortices may be considered a fund of memories that the species acquired during evolution. We use the memory term because, like personal memory, phyletic memory constitutes information that has been acquired and stored and can be retrieved (recalled) by sensory stimuli or the need to act.

2.4.2 PERCEPTUAL MEMORY Perceptual memory is acquired through the senses. It comprises all that is commonly understood as personal memory and knowledge, i.e., representations of events, objects, persons, animals, facts, names, and concepts. From a hierarchical view, memories of elementary sensations are at the

© 2004 by CRC Press, LLC

TF1663_C02.fm Page 26 Wednesday, May 12, 2004 7:51 PM

bottom; at the top, abstract concepts originally acquired by sensory experience have become independent via cognitive operations.

2.4.3 PROCEDURAL LEARNING Procedural memory is the representation of a series of actions or perceptual processing functions that occur unconsciously, and repetitions typically result in increased speed or accuracy. This relates to the acquisition of skills, whether perceptual motor skills such as riding a bicycle or driving a car or cognitive skills such as reading or problem solving. Skills clearly represent an important area of learning and serve as archetypal examples of procedural learning — learning how rather than learning that (Eichenbaum, 2000). See Figure 9.7 for a categorization of perceptual memory in the hierarchy of memory states. Skills are continuous (each component of the skill serves as a cue to the next as in cycling or steering a car) and discontinuous (a series of discrete stimulusresponse links are involved as in typing).

2.4.4 PRIMING Priming is the facilitation of recognition, reproduction, or bases in the selection of recently perceived stimuli (Eichenbaum, 2000). If a word has been presented, subjects are subsequently more likely to identify a noisy representation of the word or produce the word when faced with its stem or a fragment. As Tulving and Schacter (1990) pointed out, priming effects occur across a wide range of modalities and are typically dependent upon the repetition of the physical characteristics of the original stimulus; priming is typically much less sensitive to semantic or conceptual aspects of the primed material. It is assumed that priming is some form of neural residue that either enhances its subsequent speed of use (positive priming) or has an inhibitory effect (negative priming).

2.4.5 EVOLVING MEMORY The processes of evolving memory constitute the most important core of the presented empirical knowledge and serve as an important framework of this book (see Chapter 9). The process of formation of memory, which we denote also as evolving memory, probably constitutes the most important process during the transition from one memory state to another (see Chapter 8 for discussion of the transition between semantic and episodic memories).

2.5 DECLARATIVE MEMORY Declarative (or explicit) memory is the recall of events and facts; it is commonly known as personal memory. It constitutes two subsystems originally defined by Tulving (1972) as episodic and semantic memories.

2.5.1 EPISODIC MEMORY Episodic memory is a system that collects temporarily and spatially encoded events in a subject’s life, i.e., recalls of particular experiences or episodes. Remembering what was received as a birthday present last year and what was eaten for breakfast are examples of episodic memories. These types of memories are strongly influenced by the degrees of attention and organization that reflect the importance of the events in order to set up memory structures that are accessible to retrieval.

2.5.2 SEMANTIC MEMORY Semantic knowledge is an organization of factual information independent of specific episodes during which that information was acquired. Semantic memory, on the other hand, is knowledge © 2004 by CRC Press, LLC

TF1663_C02.fm Page 27 Wednesday, May 12, 2004 7:51 PM

of the world. Knowing the chemical formula for salt, the name of the French capital, and the number of inches in a foot are all examples of semantic memory (Baddeley, 1995). They are facts that through single or repeated mentions have come to categorize concepts, abstractions, and evidences of reality, although the subject may not necessarily remember when and where he or she acquired the information. Education can be regarded as the gradual growing and/or enriching of semantic memory, starting with perceptual knowledge of the physical world and progressing to language use, knowledge of society, and acquisition of detailed specialized information acquired via a trade or profession.

2.5.3 RELATIONSHIP

OF

EPISODIC

AND

SEMANTIC MEMORIES

Tulving's initial conceptualization (1972) proposed that semantic and episodic memories are based on separate memory systems but evidence indicates that they reflect the same system operating under different circumstances. Semantic memory stores information that may have originated from many separate experiences that are no longer individually retrievable (Fuster, 1995). The view of Baddeley is that semantic memory consists of the accumulation of many episodes. According to this author, a useful analogy is to think of a series of individual episodes piled on one another; episodic memory represents the capacity to pull one episode from the pile, whereas semantic memory reflects the capacity to look at the pile from above and draw out features that are common to many of the constituent episodes. In Chapter 8, we will analyze dynamic changes of theta responses after the transition of semantic knowledge to episodic knowledge. From the view of electrophysiological records, semantic and episodic memory systems probably share similar oscillatory activated neural populations or both systems are neurally unseparable. Section 8.5.4 is an attempt to describe the transition between semantic and episodic memory research and the electrophysiological manifestations of this crucial transition.

2.6 NEUROBIOLOGY OF MEMORY 2.6.1 MOLECULAR

AND

CELLULAR BASES

OF

MEMORY

2.6.1.1 Hebb’s Proposal Does some type of modification of neurons or of connections between neurons take place as a result of learning? For example, when we learn to associate two stimuli (e.g., an unconditioned stimulus and conditioned stimulus, as in classical conditioning), what happens in the brain to support this process (Tranel and Damasio, 1995)? Donald Hebb (1949) proposed that the coactivation of connected cells would result in a modification of weights so that when a presynaptic cell fires, the probability of postsynaptic cell firing is increased. Hebb stated: When an axon of a cell A is near enough to excite cell B or repeatedly or persistently takes part in firing it, some growth or metabolic change takes place in both cells such that A’s efficiency as one of the cells firing B is increased (p. 62). This learning principle did not describe what was meant by growth or metabolic change. However, this principle served as a useful pioneering idea, and has become one of the widely cited concepts for neurobiological investigations of learning and memory (see Chapter 1, this volume). Another important step forward arose from the work of Bliss and Lomo (1973). When the excitability of a postsynaptic cell was increased for hours, or even days or weeks, by stimulation with a high-frequency volley of pulses known as a tetanus (specifically, when the primary afferents of dentate granule cells in the hippocampus were exposed to a tetanic stimulus), the depolarization potential of the postsynaptic cell was enhanced, and this potentiation lasted for a long period. The effect is known as long-term potentiation (LTP) and it has become a very important model in modern conceptualizations of the cellular bases of learning and memory. Morris et al. (1982) © 2004 by CRC Press, LLC

TF1663_C02.fm Page 28 Wednesday, May 12, 2004 7:51 PM

demonstrated that the retardation of the behavioral learning curve in the performance of the water maze task was directly congruent with the extent to which LTP was blocked in the hippocampus. In other words, less LTP correlated with poorer learning and more LTP correlated with better learning. These results provided strong behavioral evidence supporting the role of LTP in the cellular basis of learning. 2.6.1.2 Kandel’s Fundamental Results Important advances in the understanding of learning and memory at the molecular level have come from the work of Eric Kandel and his colleagues (Kandel and Schwartz, 1982; Hawkins et al., 1983). Much of this work has been done with the Aplysia californica marine mollusk, which has a simple nervous system composed of approximately 10,000 neurons. The neurons are unusually large and easily identifiable, making Aplysia far more convenient for cellular level studies than vertebrates with far more complex nervous systems. Research by Kandel and colleagues provided the first direct evidence that alterations of synaptic efficacy play a causal role in learning. They discovered that behavioral habituation of the gill and siphon withdrawal reflex, a staple behavioral activity of Aplysia, was mediated by a reduction in transmitter release at a defined synaptic locus (Pinsker et al., 1970; Castellucci and Kandel, 1974). These results supported Hebb’s principle. Bailey and Chen (1983) later showed that habituation was accompanied by alterations in the morphologies of electrophysiologically identified synapses. These investigations provided direct evidence for forms of synaptic plasticity that may provide cellular and molecular bases for at least some forms of learning and memory. 2.6.1.3 EEG Oscillations in Aplysia and Helix pomatia Isolated invertebrate ganglia also show types of EEG oscillations in delta, theta, alpha, and gamma frequency windows. These results published by Schütt et al. (1992 and 1999) and Basar ¸ (1999) will be partly illustrated in Section 6.3.2. Do EEG oscillations manifest universal functional codes during the evolution of species? Does the phyletic memory or memory of species process EEG codes similar to those observed in the human brain? These questions cannot be answered in this book, but it is worth mentioning that correlations of oscillations with changes at the molecular level may provide essential material for tracking the molecular basis through possible electrical associations. The model by Kandel (1982) and his associates will play an important role in future memory research.

2.7 NEW SCHEME BASED ON EEG STUDIES FOR CATEGORIZATION OF MEMORY LEVELS This chapter discussing established theories and definitions briefly describes the new categorizations that emerged from experiments related to EROs. The model is new and must undergo a maturation process. However, based on experimental results cited in Chapters 3, 5, 6, and 8 describing the relevance of memory function-related oscillations, a new model is needed. Dynamic and reciprocal activations of integrative brain and memory functions can be more adequately described with oscillations because of their dynamic nature.

2.7.1 PHYSIOLOGICAL (FUNDAMENTAL-FUNCTIONAL) MEMORY We would like to introduce physiological memory as an alliance or collection of phyletic, perceptual, and procedural memories partially built in LTM storage or persistent memory. Throughout our lives, we acquire new physiological and cognitive facts and strategies. The ability to perceive red, for example, already exists in the phyletic memory. Knowledge about a flag, face, or the uniform © 2004 by CRC Press, LLC

TF1663_C02.fm Page 29 Wednesday, May 12, 2004 7:51 PM

of an enemy is acquired during life, and such images are stored easily in the LTM system, similar to skills like diving or driving. Accordingly, phyletic memory may be altered, extended, or formed into a physiological memory vital for functioning. Physiological functions that are vital for survival are genetically coded to a degree but are also partially acquired throughout life. Chapter 3 and Chapter 6 will show that all CNS functions are accompanied by or associated with some type of memory or components of whole brain memory. As noted earlier, physiological and cognitive functions are inseparable (Basar ¸ , 1999). According to our EEG-related physiology and cognition-related functions, we go a step further than Hayek (1952) and adapt the scope of Fuster (1995). We hypothesize that all functions of the CNS and memory are inseparable. Since the physiology expression is almost synonymous with function and acting, we also use the functional memory phrase. Although we have here a type of redundancy, this idea constitutes the leitmotif of this book. Accordingly, in Chapter 6 and Chapter 9, we explain the concept and functioning of physiological memory that serves as part of every memory action. A main point is that physiological memory also includes phyletic memory and (partially) procedural memory. It combines a collection of inborn (built-in) memories and newly developed and stabilized memory traces accumulated via everyday brain functioning. To see something, even the simplest light signal, is already a memory process related to a fundamental inborn or built-in retrieval process. A baby perceives a light and shows reflex responses to the light before going through learning processes.

2.7.2 TRANSITION AND COMBINATION OF MEMORY STAGES (EVOLVING MEMORY) According to Damasio (1997), memory depends on concerted work by several brain systems across many levels of neural organization. Memory is a constant work in progress and EEG oscillations evolve parallel to this constant work. Recording of function-related or memory-related EEG or MEG oscillations serves as an excellent measuring tool for transitions and other dynamic processes in human and animal brains. This is the only method that can analyze distributed dynamic processes during long experiments involving behaving and conscious brains. Sections 8.5.2 through 8.5.4 discuss an important example.

2.8 LONGER-ACTING MEMORY AND TRANSITION TO PERSISTENT MEMORY IN WHOLE BRAIN According to Section 9.5.5, event-related changes in EROs lead to substantial changes in the electrical manifestations of evolving memory. It will be shown that new learned material is transferred to LTM for longer time intervals in comparison to working memory. Astonishingly, the durations of time spent in working memory and in long-term memory storage are not defined clearly in the literature. In our opinion, longer-acting memory is a better description than long-term memory because it distinguishes between working memory and persistent memory. As a new proposition in memory categorization (Figures 9.7 and 9.12), fresh memory traces acquired in everyday experiences are temporarily stored in longer-acting memory, before reaching the persistent memory level. According to the description of memory levels introduced in Chapter 9, persistent memory combines built-in memory with physiological memory (an ensemble of submemories such as echoic memory, iconic memory, olfactory memory, etc.) and stabilized parts of longer-acting memory acquired throughout life (see Figure 9.7 and Figure 9.12). The answer to the question of how new information acquired during processes of memory evolution or memory building (manifested by multiple oscillations and enhanced coherence in the whole brain) is transferred and stored in persistent memory surpasses the scope of this book. However, it is important to note that the networks of persistent memory operate with the same oscillatory dynamics of evolving memory, i.e., they use the same basic oscillatory codes (alpha, beta, etc.). This indicates that frequency codes may be transferred to persistent memory or may play an essential role during the transition. © 2004 by CRC Press, LLC

TF1663_book.fm Page 31 Friday, May 7, 2004 8:47 AM

Part II Experiments and Their Interpretation

© 2004 by CRC Press, LLC

TF1663_book.fm Page 33 Friday, May 7, 2004 8:47 AM

Dynamic and Evolving 3 Shaping Memories by Reciprocal Activation of Attention, Perception, Learning, and Remembering “Attention” is the taking possession of by the mind, in clear and vivid form, of one out of what seem several simultaneously possible objects or train of thought. It implies withdrawal from some things in order to deal effectively with others. William James (1890)

3.1 ESSENTIAL EXPERIMENTS INVOLVING DYNAMIC MEMORY AND TOP-DOWN ACTIVITY Chapter 2 introduced the combination of attention, perception, learning, and remembering we called the APLR alliance. These processes are reciprocally active and overlapping constructs that are difficult to separate. Each term has a variety of meanings and implications. Attention, like memory, is a function of a system. Fuster (1995) noted that it made no more sense to speak of a special neural system for attention than it did to speak of one memory. Attention and memory are intimately interrelated. What we remember and how long we remember it depend on the selective function known as attention, as do the dynamics of active memory. Results of EEG oscillation studies clearly demonstrate that all integrative functions and memory are interrelated (Chapter 6). As we explained in discussing the tentative scheme related to levels of memory states, reflexes are also inborn memory components (Figure 9.7). This chapter will present experiments related to the dynamics of electroencephalogram (EEG) oscillations. Both prestimulus EEG and interstimulus activity EEG merit considerable attention because EEG activity prior to a sensory or cognitive input greatly influences brain responsiveness (Barry et al., 2003; Chapter 5, this volume). In order to understand the evolution of memory components in ERPs we must analyze the dynamic changes of prestimulus and poststimulus oscillations during cognitive paradigms that use sensory stimulations. In Chapter 1, we mentioned the importance of the hypotheses of Hebb (1949) and Edelman (1977) related to dynamic properties of the brain. Hebb assumed that brain morphology should be considerably changed after stimulations (excitations of cells creating new activated states in neural populations). Edelman mentioned the possibility of reentry, i.e., meaning that stimulation of the brain influences its responsiveness. Good research models based on these hypotheses are provided by combining prestimulus and poststimulus EEG segments; both are considered active or activated brain states. Combined EEG–event-related potential (ERP) experiments currently provide the only possibilities of studying dynamic changes for durations shorter than 1 s. The evidence of a minimal activation period of

© 2004 by CRC Press, LLC

TF1663_book.fm Page 34 Friday, May 7, 2004 8:47 AM

200 to 500 ms for awareness of a near-threshold stimulus (Libet, 1991) makes it possible to describe cognitive interactions occurring in this time interval only with EEG oscillations in behaving subjects. Top-down brain signals convey knowledge derived by prior experience rather than sensory stimulation. The analysis of changes in cognitive activation of the human brain, which is influenced by prior experience, i.e., processing of top-down signals from the brain, provides the core material of this chapter.

3.2 DYNAMIC MEMORY MANIFESTED BY INDUCED ALPHA ACTIVITY 3.2.1 SELECTIVE ATTENTION Selective attention, or simply attention, is a construct that has a rather broad but circumscribed set of meanings; selective attention is clearly distinct from nonselective central nervous system (CNS) processes such as arousal or alertness. Attentional processes are CNS functions that enable perceptual or motor responses to be made selectively to one stimulus category or dimension in preference to others. Irrelevant stimuli that are not required are partially or completely rejected from perceptual experience, entry into long-term memory (LTM), and control over behavior. Since attention refers to selective aspects of sensory processing, it follows that all experimental demonstrations of attention must measure the responsiveness of the organism to more than one category of stimulus. The differential response to attended versus unattended stimuli provides the operational basis for this construct. If an animal is tested with only one stimulus, one cannot be sure whether improvements in processing are the results of paying attention selectively or of an increased level of arousal or alertness that by definition influences a broader spectrum of sensory inputs or response propensities nonselectively (Hillyard and Picton, 1979).

3.2.2 APLR ALLIANCE Attention, perception, learning, and memory are interwoven processes. During experiments with working memory paradigms, short-term memory (STM) presents a continuously evolving interaction and reciprocal activation of each of these functions. During such experiments, memory processes are altered, enhanced, and evolved. Accordingly, dynamic experiments (single-trial oscillations) provide excellent opportunities for elucidating the processes of evolving memory during short time intervals, as we will describe in the following sections.

3.2.3 IMPORTANCE

OF INTERNAL

EVENT-RELATED OSCILLATIONS

Various neural populations in the brain can generate coherent states in which oscillatory 10-Hz activity and theta (3.5 to 7 Hz) activities are recorded. A light flash can elicit a 10-Hz enhancement in the brain if the brain shows disordered activity prior to stimulation. We can evoke a 40-Hz response with sharp onset light, acoustical stimulation, and other techniques. At this point an important question is whether we can find a way to put the brain in such coherent states of EEG activity without external sensory stimulation. Can we find a sensory-cognitive task to produce coherent internal evoked potentials, or better, internal event-related oscillations (EROs)? Petsche (1998) noted that in the past two decades inquiries involving spontaneous brain oscillations revealed new impulses and eventually led to a renaissance of alpha research (Basar ¸ , 1997). Petsche further stated: In 1994 Basar ¸ organized an influential symposium on this topic in which the multiplicity of phenomena in the alpha band was demonstrated. One of the agreements of this symposium was that alpha rhythms are not unitary phenomena but represent a large ensemble of integrative brain functions, the probable roles of which were observed (Basar ¸ et al., 1997a and b). It is for these reasons that research into possible reflections of cognition in the spontaneous EEG © 2004 by CRC Press, LLC

TF1663_book.fm Page 35 Friday, May 7, 2004 8:47 AM

has attracted psychologists. The rationale of planning the 1994 symposium was mainly based on the fact that memory-related alpha oscillations were considered as the most important signaling of integrative brain function. We would also like to mention the important work of Klimesch and his group (1996, 1997a, 1997b, 2000a and b, 2001a and b) on the memory functions of alpha activities (see also Petsche and Etlinger, 1998; Chapter 8, this volume).

3.2.4 COHERENT

AND

ORDERED STATES

OF

EEG

DUE TO

COGNITIVE TASKS

3.2.4.1 Preliminary Experiments The experiments were carried out with 16 healthy volunteers, mostly students 19 to 21 years of age. The EEGs were recorded in vertex, parietal, and occipital locations against references of ear lobes (Cz, P3, P4, and O1 in the 10–20 system). The EEG signals were amplified by using a Schwarzer machine. The subjects sat in a soundproof and echo-free room that was dimly illuminated. For stimulus preparation, evaluation of selective averaging procedure, and digital filtering, a Hewlett Packard 1000F computer was used. The filtering of EEGs and ERPs were carried out. The digital filters did not create any phase shifts. Auditory stimulation of 2000 Hz and 80 dB tones of 800-ms duration were applied at regular intervals of 2600 ms. Every third or fourth tone was omitted. The subjects were asked to predict and mark mentally the times of occurrences of the omitted signals. The EEG 1 s prior to the omitted stimulation was also recorded with the ERP. The light stimulator was a 20-W fluorescent bulb that was electrically triggered. The duration of the light step was also 800 ms. 3.2.4.2 Preliminary Results After the subjects learned and successfully followed the rhythmicity contained in the paradigm, they were usually able to increase their attention and rhythmic prestimulus EEG patterns could be observed. Most subjects reported that at the beginning of an experimental session with repetitive signals, they had difficulty predicting the time of occurrence of the stimulus omission. During the second half of the experiment, they were usually able to predict the time of the omitted signal. Accordingly, in our signal analysis we applied a selective averaging by grouping approximately the first 10 prestimulus sweeps at the beginning of the experiment and the last 10. Figure 3.1 illustrates comparatively the averages of the first 10 and last 10 prestimulus EEG epochs (digitally filtered between 1 and 25 Hz) recorded at the vertex of a subject who reported

–

10.00 µV +

–500

–400

–300

–200

–100

0.0

ms

FIGURE 3.1 Averages of the first (broken line) and last (solid line) 10 prestimulus EEG epochs of the experiment, filtered in the 1- to 25-Hz frequency band. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 36 Friday, May 7, 2004 8:47 AM

A

Filtered: 7–13 Hz

– 7.5 µV +

–500

B

– 400

–300

–200

–100

0.0 ms

–300 –200 –100 Omission of stimulus

0.0 ms

Filtered: 7–13 Hz

– 7.5 µV + –500

– 400

FIGURE 3.2 (A) Prestimulus EEG sweeps at end of experiment. (B) Prestimulus EEG sweeps at beginning of experiment. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

that he felt unsure and diffuse at the beginning of the experiment. Toward the end of the experimental session, he was more focused and performed his task better. The first 10 sweeps tended to follow the same rhythmicity, but the average was less regular and showed lower amplitudes. The average of the 10 sweeps at the end of the experiment depicted regular rhythmic behavior with large amplitudes. Rhythms similar to those illustrated in Figure 3.1 were observed with all subjects. The alignment and phase reordering were not the same in all the subjects. The exact times of regularity and phase reordering showed fluctuations from 0 to 700 ms prior to the event. Figure 3.2 shows 10 prestimulus EEG epochs from Subject C at the end (A) and at the beginning (B) of an experiment. Single sweeps were digitally filtered in a frequency range between 7 and 13 Hz according to the rhythmicity revealed in the wideband curve. It is easy to see the repeatable patterns at the end of the experiment in contrast to the lack of such patterns at the beginning. Are they recurrent networks or are the observed changes in alpha activity due to reentry following learning? Although the question cannot be answered with a clear yes, the descriptions of these results greatly favor the Hebb (1949) and Edelmann (1977) hypotheses. We will return to this question in Chapter 9.

3.4.3 GLOBAL TRENDS OF PRETARGET EVENT-RELATED RHYTHMS: SUBJECT VARIABILITY The subjects often decided responses in relation to their own set targets or mentally predicted targets. Some subjects could better predict the omitted signals after the second tones, others after the third tones. Most subjects showed better performance at the end of the experiment, but some were able to recognize the time of the occurrences early in the experiments. The reliability of the results was based on a comparison of phase-ordered EEG states with the subjects’ statements of whether they were able to mark the target signals mentally. It is well known that during long © 2004 by CRC Press, LLC

TF1663_book.fm Page 37 Friday, May 7, 2004 8:47 AM

TABLE 3.1 Comparison of Paradigms Type of Paradigm

Probability of Occurrence of Target

Every fourth or seventh stimulation randomly omitted (most difficult) Every third to fourth stimulation randomly omitted (intermediary, less difficult) Every fourth stimulation omitted (easiest; no randomness)

25% after third tone 50% after second tone 100% after third tone

Source: Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.

recording sessions, EEGs can show highly stationary behavior, random synchronization, and alignments. Our findings showed, however, that the phase-ordered patterns correlated fairly well with the subjects’ reports and, as will be shown below, fluctuations occurred with a number of subjects.

3.2.5 PARADIGMS

WITH INCREASING

OCCURRENCE PROBABILITY

To reduce the possibilities of recording randomly occurring coherent EEG signals, we extended our paradigm. In addition to paradigm 1, randomly omitting the third or fourth stimulation, we extended our measurements to include a more difficult paradigm 2 and thus decrease the probability of the occurrences of the target signals. In paradigm 2, the omitted signal could be changed from the fourth to the seventh randomly (the occurrence of a target signal was 25% when the subject already heard the second tone; see Table 3.1). During the same recording session, the subjects had to perform paradigm 3, which offered an easier way to mark the target signal mentally — every fourth signal was omitted. Paradigm 3 was the easiest because the probability of the target occurrence was 100%, after the subject heard the third signal. The comparison of the experimental results showed that the subjects could emit coherent and phase-ordered pretarget EEG signals almost in all cases with the easiest paradigm. The same subjects did not show the same good coherent and phase-ordered pretarget EEG responses with the most difficult paradigm. This section describes global results with mean value curves from experiments with 16 subjects. We will later present descriptions of experiments on single subjects. Figure 3.3 illustrates the comparison between the easiest and most difficult paradigms. The curves were wideband filtered (between 1 and 100 Hz); results were as follows: Every fourth to seventh signal omitted (the most difficult paradigm) — We observed no regular rhythmicity prior to target (or omitted) stimulation. We observed enhancement of the unfiltered EEG following the omitted stimulation (target signal). Target occurrence was rare; the surprise effect should be greater when the stimulation is omitted. We did not not analyze the variability of the P300 wave among subjects who showed relevant individual fluctuations and variabilities at the beginnings and the ends of the experimental sessions. The mean value curves from 16 subjects showed slight EEG enhancements with peaks around 300 to 400 ms. The latency changes of waves of the P300 family were large (see Galambos and Hillyard, 1981). Every fourth signal omitted (the easiest paradigm) — The pretarget EEG showed a rhythmicity around 9.5 Hz. No EEG enhancement was observed after the omitted stimulation. The mean value EEG signal looked like a continuation with a slightly slower rhythm. 3.2.5.1 3.5- to 8-Hz Range During the most difficult paradigm (every fourth to seventh signal omitted), enhancement following the omitted stimulation was observed in the theta frequency range, whereas no regular rhythmicity was noted in the pretarget EEG. In other words, the reaction was due to surprise (Figure 3.4). (Compare results with enhanced theta responses in the hippocampus cited in Chapter 4.) © 2004 by CRC Press, LLC

TF1663_book.fm Page 38 Friday, May 7, 2004 8:47 AM

FIGURE 3.3 Comparison of the most difficult (random omission of every fourth to seventh stimulation, top) and easiest (omission of every fourth stimulation, bottom) paradigms as mean value curve from 16 subjects. Filter limits: broad band, 1 to 100 Hz. Target signals at time 0. Derivation: vertex. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

FIGURE 3.4 Comparison of most difficult (random omission of every fourth to seventh stimulation, top) and easiest (omission of every fourth stimulation, bottom) paradigms as mean value curve from 16 subjects. Filter limits: 8 to 13 Hz (alpha frequency range). Target signals at time 0. Derivation: vertex. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 39 Friday, May 7, 2004 8:47 AM

FIGURE 3.5 Comparison of most difficult (random omission of every fourth to seventh stimulation, top) and easiest (omission of every fourth stimulation, bottom) paradigms as mean value curve from 16 subjects. Filter limits: 3.5 to 8 Hz (theta frequency range). Target signals at time 0. Derivation: vertex. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

3.2.5.2 8- to 13-Hz Range No coherent and ordered 10-Hz activity was observed preceding the target signal (Figure 3.5) during the most difficult paradigm. However, following the omitted stimulation, a 10-Hz enhancement was observed. During the easiest paradigm (every fourth signal omitted) a coherent 9-Hz rhythmicity preceding the target was observed. Conversely, a blocking of regular 10-Hz activity was observed after the omitted stimulus. 3.2.5.3 40-Hz Range Figure 3.6 shows the mean value curves of 16 subjects in the 40-Hz frequency range. During the easiest paradigm, we observed increased regular rhythmicity of 40-Hz activity just prior to stimulation (50 ms prior to omitted stimulation) in the mean value curve. The 40-Hz activity (or blocking) decreased following the omitted stimulation. During the most difficult paradigm (every third to seventh signal omitted), we noted no increased regular rhythmicity prior to target amd enhancement after the omitted stimulation (approximately 250 ms after stimulation). Results of enhancement or blocking of 40 Hz in this global analysis followed the same trend as the 10-Hz activity. Our analysis is not sufficient to describe whether the 10- and 40-Hz enhancements (or blockings) occurred simultaneously. Readers can compare these results with results in the cat hippocampus discussed in Chapter 4.

3.2.6 EXPERIMENTS

WITH

LIGHT STIMULATION

We will describe experiments during which repetitive light stimuli were used. Although the experiments involved small numbers of subjects, large numbers of experiments were carried out for each

© 2004 by CRC Press, LLC

TF1663_book.fm Page 40 Friday, May 7, 2004 8:47 AM

FIGURE 3.6 Comparison of most difficult (random omission of every fourth to seventh stimulation, top) and easiest (omission of every fourth stimulation, bottom) paradigms as mean value curve from 16 subjects. Filter limits, 30 to 50 Hz (40-Hz frequency range). Target signals at time 0. Derivation: vertex. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

subject. We discuss only studies carried out in the 8- to 13-Hz frequency range; detailed accounts of all the experiments at various frequency ranges are not feasible. We used the same experimental set-up and procedure described earlier in this chapter. The stimulation consisted of light steps of 800-ms duration. The light source was a 20-W fluorescent bulb that could be triggered with a short time constant. The intervals between stimuli were 2600 ms in duration. 3.2.6.1 Experiments with Varied Probabilities of Stimulus Occurrence J.K. is a medical student who quickly learned the goal of the experiments and was very cooperative during the experiments. Figure 3.7 illustrates samples of the filtered resting EEG as a control before an experiment with the cognitive task. There are three plots of the filtered EEG segments, with 10 sweeps in each plot. The three plots present samples from the same recording session. The mean correlation coefficient ( C ) of each ensemble of sweeps in a time range from -500 to 0 ms is also shown. The subject was instructed to be attentive to repetitive light stimuli. Every fourth light stimulation was omitted (the easiest paradigm). He reported at the beginning of the experiment that he could easily mark the target signal; however, after approximately 10 omitted signals or the first 40 sensory stimulations he could not concentrate as well; toward the end of the measurement, he had enormous difficulties in concentrating. Figure 3.8 shows the first 10 filtered sweeps together with the filtered mean values and wideband mean curves (1 to 30 Hz). Clear rhythmicity and good congruency are observed for most sweeps. In the following sessions of the experiments (B and C), the rhythms were less regular and the congruency among sweeps almost disappeared. Also at this stage, 10-Hz EEGs with larger amplitudes were observed in comparison with the resting EEG shown in Figure 3.7. At the beginning, when the subjects reported good performance, the correlation coefficient was high (0.38). It later diminished (0.13 and 0.01) and decreased drastically by the end of the experiment. © 2004 by CRC Press, LLC

TF1663_book.fm Page 41 Friday, May 7, 2004 8:47 AM

FIGURE 3.7 Resting EEG of subject J.K. Top: mean value results on ten sweeps. Bottom: 10 sweeps of EEG segments that were digitally filtered in the frequency range of 8 to 13 Hz. Time 0 was chosen arbitrarily. EEG samples were recorded at the beginning (A), middle (B), and end (C) of session. Correlation coefficients were evaluated from 3 ensembles of 10 sweeps. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

Figure 3.9 illustrates a similar experiment with J.K. a few months later. He again reported at the beginning of the experiment that he was able to mark the target mentally with ease; during the experiment, he lost his ability to follow the target. Near the end he again had better control in marking the target. Figure 3.9(B) shows the decrease in congruency and diminishing of the correlation coefficient. In Figure 3.9(C), the congruency is better ( C = 0.28). On the following day, we started the first experiment with the most difficult paradigm (every third to seventh stimulation omitted) and proceeded next to the easiest paradigm. During the most difficult paradigm, J.K. said he felt unsure whether he could follow the rhythmicity of the light signals at the beginning of the experiment. During the last two thirds of the experimental period, he was able to mark a larger number of target signals. Figure 3.10(A) shows the beginning and Figure 3.10(B) shows the middle stage. The amplitudes of the EEG increased during the experiments but the correlation coefficient did not. During another experiment with the easiest paradigm, J.K. reported that he had not performed well at the beginning (Figure 3.11). However, toward the end of the experiment, he definitely had better control in marking the target. Comparison of Figure 3.8 and Figure 3.11 shows that an opposite effect occurred. In the experiment shown in Figure 3.11, the congruency between single curves was better toward the end of the experiment and C increased from 0.00 to 0.16. In five subjects, the EEG measurements during the easiest paradigm using light signals were taken after application of the most difficult paradigm. During a session with the most difficult paradigm, congruency of single rhythms like the epochs of Figure 3.8 and Figure 3.11 were not observed. Further, the correlation coefficients calculated during the four stages of the experiment remained in all cases around 0.05; they never reached values around 0.4. Comparison of easiest and most difficult paradigms — We want to mention again why the comparison of results using the easiest and the most difficult paradigms for the same subject is important to formulate a judgment about event-related pretarget rhythms. It is possible for the same subject to increase the probability of the occurrence of the target by up to 100%. The increase in the EEG amplitude and the tendency to regularity and phase ordering are reflected in correlation coefficients. If the probability of the occurrence of a target were then decreased, one would expect a less good or even bad performance. In the latter case, it might be expected that the phase ordering © 2004 by CRC Press, LLC

TF1663_book.fm Page 42 Friday, May 7, 2004 8:47 AM

FIGURE 3.8 Pretarget EEG of subject J.K. (experiment 3) during the easiest paradigm (every fourth signal omitted). EEG segments were filtered in the frequency range of 8 to 13 Hz. Time scale from -1000 ms to 0 indicates 1 s recording time prior to target (omitted tone). (A) Ten single EEG samples at the beginning of experimental session (bottom). Mean value curves of 10 sweeps (middle). Broadband mean value curve from 10 sweeps (top). Filter range: 1 to 30 Hz. (B) Ten EEG samples in middle of session (bottom). Mean value curve from 10 sweeps (top). (C) Ten EEG samples at end of the session (bottom) Mean value curve from 10 sweeps (top). The correlation coefficients evaluated from 3 ensembles of 10 sweeps are shown at the top of each ensemble. (C) covers only the period from –500 to 0 ms., i.e., 500 ms prior to target. Subject's report: (A) = good performance; (B) and (C) = bad performance. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

of the EEG and the tendency to a repeatable pattern would finish. On the basis of this reasoning, we applied both paradigms to five subjects on the same day and always obtained comparable results that were similar to the results from subject J.K. The increase in correlation coefficient means an increase in similarity of single epochs. The fact that subjects who reported good performance produced mean correlation coefficients up to 0.4 shows that an EEG can attain good phase-ordered patterns; this is contrary to the cases of recordings with less probability of occurrence. We must also emphasize that the recording of almost repeatable EEG patterns during defined experiments with cognitive targets required a large number of experiments and good cooperation of the subjects. Different time windows — In Figure 3.7 and Figure 3.11, we consistently considered the time window between 500 and 0 ms prior to target signals. Although the EEGs of most subjects depicted phase orderings starting 1000 to 700 ms prior to target signals, the time scale of -500 to 0 ms is the most common one for a rough preliminary evaluation. To avoid errors of visual inspection, we started each analysis with some moving time windows prior to target. This means that we chose six time windows at various points along the time axis of -1000 to 0 ms. The narrowest window © 2004 by CRC Press, LLC

TF1663_book.fm Page 43 Friday, May 7, 2004 8:47 AM

FIGURE 3.9 Pretarget EEG of subject J.K. (experiment 15) during the same (easiest) paradigm. EEG segments were filtered in the frequency range of 8 to 13 Hz. Time scale from -1000 ms to 0 indicates 1 s recording time prior to target (omitted tone). (A) Ten single EEG samples at the beginning of experimental session (bottom). Mean value curves of 10 sweeps (middle). Broadband mean value curve from 10 sweeps (top). Filter range: 1 to 30 Hz. (B) Ten EEG samples in middle of session (bottom). Mean value curve from 10 sweeps (top). (C) Ten EEG samples at end of the session (bottom) Mean value curve from 10 sweeps (top). The correlation coefficients evaluated from 3 ensembles of 10 sweeps are shown at the top of each ensemble. (C) covers only the period from –500 to 0 ms., i.e., 500 ms prior to target. Subject's report: (A) = good performance; (B) and (C) = bad performance. Results show repetition after a few months. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

had a duration of 300 ms. Larger correlation coefficients were to be expected for the smaller windows. Let us consider the 10 EEG sweeps illustrated in Figure 3.8. For the 500 ms before stimulation, C = 0.40. As Table 3.2 shows, C takes different values depending on the length and position of the time window. For a time window of -300 to 0 ms before stimulation, C has the highest value; the window from -700 to -300 ms has a much lower value. Table 3.2 also shows correlation coefficients of control EEG sweeps from Figure 3.7. During the recording of EEG sweeps where the subject did not report good performance, the correlation coefficients were not much higher even by choosing narrow time windows (mean value of -0.05). The control EEG of the same subject (sweeps from Figure 3.7) did not show significant values of C even with narrow time windows. For all performed experiments, searches with different time windows were carried out; the results are similar to those in Table 3.2. They indicated highly increased mean values of correlation coefficients during good performance sessions compared with resting EEGs or bad performance sessions. Evaluation of all the subjects under study gave similar results, showing that with analysis of time, the correlation coefficient is always highest during the easiest paradigm.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 44 Friday, May 7, 2004 8:47 AM

FIGURE 3.10 Pretarget EEG of subject J.K. (experiment 16) during the most difficult paradigm (every fourth to seventh signal omitted). EEG segments were filtered in the frequency range of 8 to 13 Hz. The time scale from –1000 ms to 0 indicates 1 s recording time prior to target (omitted light). (A) Ten single EEG samples at beginning of the experimental session (bottom). Mean value curve (top). (B) Ten EEG samples in middle of session (bottom). Mean value curve from 10 sweeps (top). (C) Ten EEG samples at end of session. Correlation coefficients cover only the period from –500 to 0 ms. Subject's report: tried to do well. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

FIGURE 3.11 Pretarget EEG (8 to 13 Hz) of subject J.K. (experiment 19) during the easiest paradigm (every fourth signal omitted). (A) EEG samples at beginning of experimental session (bottom). Mean value curve (top). (B) EEG samples in middle of session (bottom). Mean value curve from 10 sweeps (top). (C) EEG samples at end of session. Subject's report: performance “bad” at beginning (A); increasingly good toward end of experiment [(B) and (C)]. (Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 45 Friday, May 7, 2004 8:47 AM

TABLE 3.2 Correlation Coefficients of EEG of Left Occipital Recording of Subject J.K. during Several Time Windows Experiment 3 Time Window (ms)

Experiment 15

Experiment 19

Bad Performance C

Good Performance C

Bad Performance C

Good Performance C

Bad Performance C

Control EEG C

–0.07 –0.04 –0.1 0.1 –0.07 0.03 –0.04

0.15 0.3 0.4 –0.07 0.14 0.46 0.23

–0.04 –0.04 –0.06 –0.07 –0.06 –0.06 –0.05

0.16 0.15 0.12 0.18 0.14 0.11 0.14

0.01 0.03 0 0.02 0.04 –0.01 0.01

–0.01 –0.05 –0.04 –0.01 –0.01 –0.02 0

–1000 to 0 –700 to 0 –500 to 0 –1000 to –500 –700 to –300 –300 to 0 Mean value (6 time windows)

Source: Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.

3.2.7 EXPERIMENTS

WITH

SUBJECT A.F.

Subject A.F. was a technical assistant in our research group. From the beginning of the experiments, he had a great interest in serving as a subject and provided detailed reports after every measurement session. The analysis of the single sweeps by visual inspection correlated highly with his reports in most cases. When he reported that he was able to mark the target signal mentally during a given measurement session, the single pretarget EEG curves usually showed good congruence. The agreement with his report was about 80%. Taking this degree of reliability into account, we performed seven experiments with A.F. over a period of about 3 months. The measurements and reports were as follows: The resting EEG prior to application of a paradigm was measured. The easiest paradigm with repetitive light stimuli was applied. After application of 30 light stimuli (and 10 omitted signals), A.F. wrote a report and describe the sessions as good or bad performances. The single sweeps of the pretarget EEG were plotted and the reliability of the subject's report was checked by means of analysis with correlation coefficients that revealed the degree of single sweep congruence. In six of the seven experiments A.F. reported that he had measurement periods with good and bad performance. For one experiment, he reported only bad performance. Table 3.3 shows C values in the frequency band of 8 to 13 Hz for resting EEGs and good and bad performance periods for the easiest paradigm. The subject of Experiment 32 could not achieve good performance in any measurement period. The correlation coefficient for the resting EEG had mean values no higher than 0.05 and they averaged <0.002. The periods with bad performance shows mean values of 0.02. The coefficients for periods with good performance varied between 0.04 and 0.3, with a mean of 0.16. The mean value showed high variability for one subject. It is noteworthy only as an important indicator of the trends of experiments with A.F. Experiments involving varied target probabilities, evaluation of correlation coefficient, and plotting of single EEG curves were performed with six subjects in similar sequences. Each subject was studied on two or more different days. All showed behavior similar to J.K.’s and partially similar to A.F.’s

© 2004 by CRC Press, LLC

TF1663_book.fm Page 46 Friday, May 7, 2004 8:47 AM

TABLE 3.3 Correlation Coefficients of 10 Single EEG Epochs in Left Occipital Recording of Subject A.F. during Easiest Paradigm and Control EEG Easiest Paradigm Experiment Number 24 27 32 37 42 47 52 Mean value

Control EEG

Good Performance C

Bad Performance C

Experiment Number

Correlation Coefficient C

0.16 0.3 — 0.04 0.13 0.1 0.24 0.16 (n = 6)

0.01 –0.02 –0.06 –0.05 0.01 –0.06 0.06 –0.02 (n = 7)

22 25 30 35 40 43 50

0 –0.04 0.02 0.02 –0.05 0.01 0.03 0.00 (n = 7)

Source: Modified from Basar, ¸ E. et al. (1989a), in Brain Dynamics, Springer, Berlin, p. 43.

3.2.8 QUASIDETERMINISTIC EEGS, COGNITIVE STATES,

AND

DYNAMIC MEMORIES

The experiments described in the previous sections show that it is possible during cognitive tasks to measure almost reproducible EEG patterns in subjects expecting defined repetitive sensory stimuli. During such coherent states, single sweeps of EEGs were time locked to target signals for at least 10 min* during which the subjects were able to mark cognitive targets mentally in a recurrent manner. We used the expression quasideterministic EEG to describe recurrently omitted, almost reproducible, EEG patterns. The specific brain function was related in this case to a type of short learning process and STM we tentatively denoted as dynamic memory (Basar ¸ , 1988). We have focused on the 10-Hz frequency range although other frequency ranges were briefly discussed earlier in this chapter. The dynamic memory expression could be extended to cover all relevant EEG frequency ranges and combinations of several patterns. The use of the quasideterministic EEG expression finds its legitimacy in results reported by Babloyantz (1988), Röschke and Basar ¸ (1998), Basar ¸ (1990), and Freeman and Skarda (1985), who demonstrated that an EEG reflects the properties of a strange attractor. Hillyard and Picton (1979) used selective attention or simply attention as a construct that has a rather broad but circumscribed set of meanings; its meaning is clearly distinguished from nonselective CNS processes such as arousal or alertness. Attentional processes are CNS functions that enable perceptual or motor responses to be made selectively to one stimulus category or dimension in preference over others. Irrelevant stimuli are partially or completely rejected from perceptual experience, entry into LTM, and control over behavior. These authors state that attention refers to selective aspects of sensory processing. Accordingly, all experimental demonstrations of attention must measure the responsiveness of an organism to more than one category of stimulus. The initial stages of sensory processing are generally thought to proceed inflexibly and consist of an initial afferent registration and feature analysis of incoming sensory data.** This information persists in

* A recording session with 10 target signals (10 omitted stimuli and 30 physical stimulations) requires at least 700 s or about 3 min. Some of the subjects can reach the same performances during 30-40 omitted stimuli, in other words, around 10 min. ** According to Sokolov (1975) there are expectation cells that fire according to the expected input and sensory-reporting cells that fire according to stimulus. See Chapter 4.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 47 Friday, May 7, 2004 8:47 AM

accessible form about 1 s in a high-capacity sensory memory designated the buffer store, iconic (visual) memory, or echoic (auditory) memory. During our experiments, the subjects had to pay attention to omitted stimuli. If they were sufficiently able to mark the targets mentally, they anticipated 10-Hz waves that were time locked to targets, showing almost reproducible patterns. Depending on a subject’s performance, the coherency time of such reproducible wave packets ranged from 300 to 1000 ms, most of it prior to the time the omitted stimulus was due. In earlier publications, Basar ¸ (1983a and b; 1990) tentatively assumed that the evoked potentials manifested bifurcation of the strange attractor EEG to a limit cycle attractor of short duration. A strange attractor is manifested by activity that appears to be random. However, the activity of a strange attractor is deterministic and reproducible if the input and initial conditions can be replicated. The experiments with human subjects described in this chapter showed that by increased certainty due to their expectation of repetitive sensory signals and accordingly by increased attention stages, the subjects seemed to generate internal cognitive inputs to the CNS. These cognitive inputs were probably the results of similar repetitive mental efforts. According to our results, if a subject cannot mentally predict the occurrence of an expected target signal (omitted stimulation), no averaged synchronization of the EEG occurs in the 10-Hz frequency range (see section on subject J.K.) or in the 40-Hz frequency range (only global findings were reported). The 10-, 40-, and 4-Hz EEGs proceeded from disordered states to ordered coherent states during defined cognitive inputs to the CNS, revealing a similarity to evoked potentials elicited by exogenous sensory stimuli that also showed transitions from disordered to ordered coherent states. 3.2.8.1 Dynamics of Time-Locked EEG Patterns Observations of changes in the amplitude of the EEG and enhancement or blocking of alpha activity during mental tasks (solving arithmetic problems or listening to music) is not new; relevant examples have been described in the literature (Creutzfeldt, 1983; Giannitrappani, 1985; Petsche et al., 1987). The new observation we reported is the fact that alpha waves can be time locked to a defined cognitive target and that in well-defined experiments, reproducible alpha patterns can be emitted from the human brain. In addition to carrying out conventional analysis as described in P300 wave studies by Sutton et al. (1965) and Galambos and Hillyard (1981), we encountered developments in the analysis of ERPs in relation to missing auditory stimuli (Nakamura et al., 1986; Gauthier et al., 1986, Takasaka, 1985; Friedman, 1984; McCallum, 1980; Simson et al., 1976). These authors’ results consisted mostly of descriptions of latencies of potentials following omitted stimuli and topographic distributions of the family of P300 waves. Gauthier et al. (1986) and McCallum (1980) reported contingent negative variation (CNV) changes under certain experimental conditions. Synchronized pretarget event-related rhythms, however, were not described by these authors because they did not use small segments of EEG or filter methods and their subjects were not asked to perform the same mental effort to mark targets as omitted stimuli. Lehmann (1989) stated that functional states might be recognized for relatively short epochs of EEGs depicting systematic relations with certain types of thought processes. According to his description, the duration of spatially stable microstates was about 500 ms. It is certainly attractive to think of brain microstates as building blocks of higher information processing that qualify as areas of consciousness only beyond certain durations. According to our preliminary results, microstates have various durations, depending on basic rhythmic activity. Sensory-cognitive microstates in the alpha frequency range may have durations of about 300 ms; theta microstate durations are about 500 ms. Conversely, 40-Hz microstates should have shorter durations of 20 to 50 ms. The results of long-standing experiments favor the hypothesis of Hebb. Electrical signals are changed after long-lasting experiments. Nothing can be said about changes at the synaptic level.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 48 Friday, May 7, 2004 8:47 AM

3.3 RELATIONS BETWEEN MEMORY STATES AND P300 RESPONSES: EROS Many studies have reported human brain potentials elicited by both unexpected and expected but absent sensory stimuli. Such emitted or endogenous potentials are generally regarded as signs of endogenous brain processes such as memory, internal time estimation, and other cognitive mechanisms (Ruchkin and Sutton, 1983). When a task-relevant stimulus occurs unpredictably and infrequently, it is often followed by a late positive potential at a latency of 300 to 400 ms after the point of information delivery. This wave has been designated the late positive component, P300, or P3, and constitutes a well-known example of an emitted or endogenous potential. Various studies suggest that the P300 change reflects the activation of brain mechanisms selectively concerned with the evaluation of stimulus relevance and probability, and that this change is relatively independent of stimulus modality, intensity, and presentation rate. Several authors have shown that the P300 wave can be related to the detection of an infrequent target stimulus that requires some cognitive response, e.g., counting target stimuli occurring randomly and infrequently in a train of nontarget stimuli (Desmedt, 1979; Hillyard and Picton, 1979). The studies cited, indeed most studies in this field, are based mainly on latency and amplitude measurements derived from averaged EEG data. Many authors discussed the limitations of the averaging technique in studies of sensory evoked potentials (EPs) and cognitive ERPs. Frequency domain analysis may hold greater potential for revealing the genesis of these potentials, especially if attempts are made to relate poststimulus changes to prestimulus EEG activity. Such an approach offers greater opportunities for the understanding of cognitive brain dynamics, since it is not possible to measure response variation in averaged data or investigate the relationship between the EP and ERP responses and prestimulus EEG activity. The P300 response obtained by means of “oddball” (OB) paradigms has been used successfully for psychological research and clinical diagnostics. Several excellent reviews of the methodological, psychological, and clinical aspects of this paradigm are available (Hillyard and Picton, 1979; Johnson, 1988 and 1990; Näätänen, 1988; Regan, 1989; Woods, 1990). An important aim in the analysis of physiological correlates of the P300 response is searching sources of generators giving rise to the P300 component of the ERP. Several investigators used intracranial electrodes to localize the sources in human recordings and animal models (Halgren et ˘ et al., 1991a). Several of these authors indicate the existence of multiple al., 1986; Basar ¸ -Eroglu generators including sources in the hippocampus (HI), parietal, frontal, and several other areas of the association cortex (Knight et al., 1981; Smith et al., 1990; Paller et al., 1988). This section covers an ensemble of paradigms and the combined analyses of EEG and EP applied to ERPs. The combined evaluation of results obtained with unconventional and new techniques is called the Brain Dynamics Research Program (see Chapter 1). This program, which includes the application of a chain of paradigms interwoven with the analysis techniques cited in this chapter, has been applied extensively in the laboratories at the University of Lübeck in Germany for the analysis of ERPs of both human and cat brains. (See Chapter 4 and Chapter 6. Readers should also compare the significant progress of Polich and coworkers (1995); their recent publications are not described in detail in this volume.)

3.3.1 EXPERIMENTAL SET-UP

AND

PARADIGMS

Electroencephalographic (EEG) activity was recorded at the vertex (Cz electrode site in the international 10-20 system); the reference electrode was placed on the left ear. A Schwarzer Encephysioscript EEG machine was used to record activity. The filter bandpass was set to 0.1 to 70 Hz. The EEG recorded continuously and was monitored during the experiments. The subjects’ responses were observed through a closed-circuit television system. Experiments were carried out on 12 healthy volunteers (4 females and 8 males) ranging in age from 25 to 45 years. © 2004 by CRC Press, LLC

TF1663_book.fm Page 49 Friday, May 7, 2004 8:47 AM

EEG

ERP

EEG

Tone (1500 or 1550 Hz)

variable interstimulus interval (1.7 to 8 s)

ERP

ETC

Tone (1500 or 1550 Hz)

Random distribution 1500 Hz Tone = 85% 1500 Hz Tone = 15%

FIGURE 3.12 EEG-ERP recordings. Thick horizontal lines indicate duration of stimulation. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

Stimulations — As auditory stimuli, 80 dB, 1500 and/or 1550 Hz tones with 0.5 ms rise time and 800 ms duration were presented binaurally. In each recording session, the first step was registering the spontaneous EEG for a few minutes to determine global characteristics of subjects' spontaneous activities and arousal states at the onset. This also helped the subjects to become familiar with the experimental conditions. Thereafter, auditory EPs, the omitted stimulus paradigm, and the oddball paradigm were applied on the first group of subjects with short resting periods between paradigms. The auditory EP experiments involved presentation of 1500 Hz tones. Selective averaging of EPs — The stored raw single EEG-EP or EEG-ERP epochs were selected with specified criteria after the recording session. EEG segments showing movement artifacts, sleep spindles, or slow waves were eliminated. -EP epochs — With every stimulus presented, segments of EEG activity Recording of EEGpreceding and the EP or ERP following the stimulus were digitized and stored on computer disc memory. This operation was repeated about 100 to 200 times. In recording EEG-ERP epochs, 1.6 s of EEG activity preceding each stimulus and 1 s of EEG activity subsequent to the stimulus were digitized, labeled, and stored on the memory of a Hewlett Packard 1000F data acquisition system. Paradigm 1: Oddball — This paradigm is illustrated in Figure 3.12. The tones were presented in a pseudorandom sequence with 1550 Hz tones used 20% of the time and 1500 Hz tones used 80% of the time. The intervals between tones varied randomly from 2.5 to 4 s with a mean value of 3 s as in auditory EP experiments. The subjects were instructed to keep mental counts of the numbers of 1600 Hz (nonfrequent target) tones. Paradigm 2: Oddball with increased certainty of alternating targets — For further analyses, the conventional auditory oddball paradigm used was modified to demonstrate the development and variation of prestimulus preparation changes. Repetitive frequent oddball tones were presented without informing subjects that the tones would be presented regularly. The 1550-Hz target tones were presented alternately with 1500-Hz nontarget tones. Subjects were asked to count the number of target tones and promised monetary rewards if they counted correctly. From a physiological view, the brain was stimulated repetitively at fixed intervals. From a psychological view, subjects were unsure when the target tones would arrive, although it was expected that they would recognize the regular pattern quickly. However, even when they felt certain that every second tone would be a target, they did not know whether the pattern would change. Figure 3.13 is a schematic representation of the paradigm. Repetitive frequent oddball tones were also presented in cases where subjects were informed that every second tone would be a target. Stimuli were presented as cited above and subjects were again promised rewards if they counted the correct number of target tones. The duration of tones was 800 ms and the fixed interval between tones was 1.65 s. Conventional auditory EPs were © 2004 by CRC Press, LLC

TF1663_book.fm Page 50 Friday, May 7, 2004 8:47 AM

EEG

ERP

EEG

ERP

Fixed Interval

1500 Hz Tone

EEG

ERP

ETC.

Fixed Interval

1550 Hz Tone

1500 Hz Tone

FIGURE 3.13 EEG-ERP recordings. Thick horizontal lines indicate duration of stimulation. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

obtained with randomly applied step function tones of 1500 Hz at 70 dB above threshold and duration of 800 ms. Those sensory EPs were recorded to establish a baseline to which ERPs in task-relevant experiments were compared. Subjects were seated comfortably in a quiet, soundproof room, and asked to keep their eyes closed. They were instructed to keep a mental count of the number of high (target) tones and promised rewards if they counted the correct number of target tones to enhance cooperation.

3.3.2 FREQUENCY ANALYSIS

OF

ERPS: PRELIMINARY RESULTS

3.3.2.1 Comparison of EPs and ERPs Figure 3.14 shows typical findings from one of our twelve subjects. Figure 3.14(A) shows the sensory (control) EP. Figure 3.14(B) shows corresponding ERPs in the OB experiment. Dashed line curves show the averaged ERPs elicited by frequently occurring 1500-Hz tones that subjects were asked to ignore. The solid line curves show the averaged ERPs elicited by infrequent 1550Hz tones the subjects were asked to count. Characteristically, the sensory EPs showed negative peaks in the region of 100 ms (N100) and positive peaks around 200 ms (P200). The N100 wave is was usually preceded by a small positive wave around 50 ms. (Readers are referred to Chapter 4 to compare the N100 responses from singleunit recordings of animal brains.) The inspection of Figure 3.14 indicates that the averaged responses to 1500-Hz tones show a configuration that is similar to their corresponding sensory-evoked potential. By contrast, the averaged responses to 1550-Hz target tones show a large positive deflection around 380 ms that represents the late positive component or P300. All twelve subjects showed similar types of deflections. Only examples for vertex recordings are shown. Chapter 6 will provide a comparative discussion of the results and interpretations of vertex, parietal, frontal, and occipital ERPs. Figure 3.15 shows the amplitude frequency characteristics (AFC) computed from the averaged EP and ERP results shown in Figure 3.14. The top curve was obtained from the sensory (control) EP, the middle curve from the ERP response to 1500 Hz (ignore), and the bottom curve from the response to 1550-Hz target tones the subjects were asked to count. Figure 3.16 shows the averaged responses and filtered components of the sensory-evoked potential (A), the ERP responses to 1500Hz nontarget tones (B), and responses to 1550-Hz target tones (C) for three subjects. The top traces are the averaged responses. The next group illustrates the respective frequency components obtained by filtering the averaged responses according to amplitude frequency characteristics. Figure 3.16 illustrates the different contributions of the three given frequency bands to peaks in the averaged curves. Assuming that these peaks were formed by the superposition of different frequency © 2004 by CRC Press, LLC

TF1663_book.fm Page 51 Friday, May 7, 2004 8:47 AM

A

– 3 µV +

0

200

400

600

800

600

800

B – 5 µV + – 3 µV + 0

200

400 Ignore

Attend

FIGURE 3.14 (A) Typical averaged auditory EP recorded from vertex. (B) ERPs from same subject. Solid lines represent ERPs to rare target tones; dashed lines represent ERPs to frequent nontarget tones. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

FIGURE 3.15 Amplitude–frequency characteristics (AFC) of vertex EP and ERPs of one subject. Curves were obtained by applying Fourier transform to averaged evoked potential (EP) and event-related potentials (ERP ignore and ERP attend). Along the abscissa is the frequency in logarithmic scale; along the ordinate is the potential amplitude or gain G(jw) in decibels. The curves are normalized such that the amplitude at 1 Hz is equal to 1 (or 20 log 1 = 0). (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

components, it can be seen, for example, that the N100 peaks were formed mainly by 3.5- to 8-Hz and 8- to 13-Hz activity; the P200 peaks were formed by the superposition of the first positive peak of 1- to 3.5-Hz activity and the first positive peak of 3.5- to 8-Hz activity. © 2004 by CRC Press, LLC

TF1663_book.fm Page 52 Friday, May 7, 2004 8:47 AM

FIGURE 3.16 Averaged response and filtered components of sensory-evoked potential (EP); ERPs to 1500Hz nontarget tones (ignore) and to 1550-Hz target tones (attend). (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

FIGURE 3.17 Filtered frequency components of sensory-evoked potential (EP) and ERPs (ignore and attend); curves are also shown in Figure 3.16 and are arranged differently here. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

3.3.2.2 Comparative Analysis of Poststimulus Frequency Changes under Different Conditions and Their Contributions to Different Latency Peaks Figure 3.17 shows the filtered frequency components of Figure 3.16 arranged to provide a comparison of each of the three frequency bands under different experimental conditions.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 53 Friday, May 7, 2004 8:47 AM

1- to 3.5-Hz activity (delta frequency) — The configuration of activity in this frequency band appears grossly similar to the EP, nontarget, and target ERPs except for a progressive latency shift to the right of the major positive deflection, which is maximally delayed in the responses to target tones. The delay between the EP and nontarget ERP was in the range of 20 to 30 ms, whereas the delay between the nontarget and target ERP responses was in the range of 70 to 90 ms. This maximum delay in target responses was associated with a tendency for greater positivity immediately poststimulus; the small negativity often seen in EP and nontarget ERP responses tended to be absent in target ERP responses. The maximum positive deflection to target tones was followed by a negative deflection, peaking around 600 s. This peak was usually larger for target tones than for EP and nontarget tones. Our analysis suggests that this N600 deflection should be considered a continuation of the positive deflection around 380 ms and a part of the mechanism that gives rise to changes in the 1- to 3.5-Hz range. We noted a tendency for N600 amplitudes to be larger if the P380 deflection was relatively small. 3.5- to 8-Hz activity (theta frequency) — Activity in this frequency range showed the greatest changes in the patterns of oscillation and damping. A “reverse” pattern of oscillation during approximately the first 400 ms can be seen in the ERP responses to target tones, and was observed in all twelve subjects. For EPs and nontarget ERPs, the maximum response amplitude (enhancement) is usually seen in the first poststimulus wave, whereas in the target ERP response, the maximum amplitude is noted in the second (Figure 3.18) and sometimes third wave. A longer duration of theta oscillation in target ERP responses indicates that this frequency band often makes significant contributions to N200 and P380 deflections. 8- to 13-Hz activity (alpha frequency) — Stimulus-elicited oscillation in this frequency band does not appear to make any direct or obvious contribution to the positive wave at 380 ms, and is usually maximally desynchronized at this latency. However, Figure 3.17 shows a frequently observed finding, namely that the second major oscillation is greater for target than for EP and nontarget ERP responses. This second oscillation contributes to the N200 wave (discussed below). Our findings suggest that while alpha activity does not appear to contribute to the P300 wave, it makes a contribution to task-relevant ERP changes. 3.3.2.3 Formation of Peaks N200 — A negative peak around 200 ms was a conspicuous feature of the averaged ERP responses to nontarget and especially target tones (see Figure 3.14 and Figure 3.16). Frequency analysis of the averaged responses shows that this peak resulted mainly from delayed theta enhancement and the more prominent second oscillation in the alpha band, although the delayed positive delta deflection was an indirect contribution. This peak cannot be observed in the sensory evoked potential, and it is suggested that it should be regarded as a part of the complex P300 changes. P165 — In the sensory EPs, positive peaks around 200 ms were considered to be formed by the superposition of 1- to 3.5-Hz, 3.5- to 8-Hz and, more variably, 8- to 13-Hz time-locked oscillations (see Figure 3.14 and Figure 3.16). For task-relevant ERPs, it appeared that a P200 peak occured earlier, in the region of 165 ms (P165; Goodin et al., 1978). According to our analysis, this P165 peak was formed by contributions of time-locked 3.5- to 8-Hz and 8- to 13-Hz activities that became “separated” from 1- to 3.5-Hz activity, whose major positive deflection shifted in latency to the right. Frequency decomposition of averaged data serves to illustrate how the peaks in the averaged data are formed by alternations in the time course of delta, theta, and alpha response oscillations. We performed a straightforward component analysis that describes how individual peaks in ERPs are shaped from the superposition of various oscillatory (and elementary) waveforms. Such an approach may lead to new perspectives that may facilitate the understanding of response mechanisms operating under different experimental conditions.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 54 Friday, May 7, 2004 8:47 AM

FIGURE 3.18 (A) Typical averaged ERPs obtained by applying alternating nontarget (dashed line) and target (solid line) tones. Recordings are from vertex. (B) Filtered components of averaged ERPs shown in (A) in three different frequency bands. (C) Comparison of filtered ERP components to randomly applied infrequent target tones (solid line) and averaged ERP responses to regular, frequently presented target tones (dashed line). Solid curves represent ERP components to target (attend) stimuli; dashed lines represent responses to nontarget (ignore) stimuli. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

3.3.2.4 Comparison of ERP Responses to Regular and Random Infrequent Target Stimuli Alternating tones — This section discusses increasing the probability of target occurrence after slightly changing the oddball paradigm. We applied a new step. Subjects increased their attention to detect a target stimulus, but not a rare one. The surprise reaction and decision making were eliminated. The issue was to determine how ERPs reacted when the probability of target occurrence increased to 100%. Figure 3.18(A) shows typical averaged vertex ERPs obtained under the paradigm of alternating nontarget and target tones. The subjects were not informed that target tones would be presented regularly. The dashed line represents the ERP to nontarget (ignore) tones and the solid line curve represents the ERP to target (attend) tones. The nontarget ERP is characterized by a negative peak around 100 ms (N100) and a positive one around 200 ms (P200). The target ERP has a similar shape except for a marked positive peak around 300 ms (P300 change). The amplitude frequency characteristics obtained from the ERPs of Figure 3.18(A) are shown in Figure 3.19. The dashed line represents responses to nontarget tones. The solid line represents responses to target tones. The amplitude frequency characteristics (AFCs) obtained from both curves show response peaks in the frequency ranges of 1 to 2.5, 2.5 to 4, 4 to 7, and 7 to 17 Hz. For the sake of simplicity, we confined our band-pass filter analysis to the first three major peaks of the AFCs shown in Figure 3.19. Figure 3.18(B) shows the filtered poststimulus oscillations within these frequency bands. The averaged N100 peaks of both target and nontarget ERPs were © 2004 by CRC Press, LLC

TF1663_book.fm Page 55 Friday, May 7, 2004 8:47 AM

FIGURE 3.18 (CONTINUED)

(C)

FIGURE 3.19 Amplitude frequency characteristics (AFCs) obtained from vertex ERPs of one subject to target tones (solid line) and nontarget tones (dashed line). The curves were obtained by applying the Fourier transform to averaged ERPs of Figure 3.18(A). Along the abscissa is the frequency in logarithmic scale; along the ordinate is the potential amplitude or gain G( jw) in decibels. The curves are normalized so that the amplitude at 1 Hz is equal to 1 (or 20 log 1 = 0). (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

formed mainly by 4- to 7-Hz and 7- to 17-Hz oscillations. The large positive peak around 200 ms (P200) was mainly formed by 1- to 2.5-Hz activity. The smaller positive peak around 180 ms was formed mainly by 4- to 7-Hz activity, while the small peak around 280 ms was formed solely by 7- to 17-Hz activity. The early negative peak around 50 ms (N50) was formed by 7- to 17-Hz © 2004 by CRC Press, LLC

TF1663_book.fm Page 56 Friday, May 7, 2004 8:47 AM

activity. The averaged ERPs in Figure 3.18(A) can be approximated closely by a superposition of the poststimulus, time-locked oscillations within the first three frequency bands derived from AFCs. Figure 3.18(C) shows a comparison between the filtered ERP responses to random, infrequently presented tones (taken from an analysis involving the same subject) and the averaged ERP responses to regular, frequently presented target tones. The subject did not know the target tones would be presented alternately. While the subject presumably paid close attention in both experiments to earn the rewards, it can be seen that the ERP change is much more pronounced in response to random, infrequent tones. Solid lines present the responses to random target tones and dashed lines represent responses to regular repetitive tones. The 1- to 2.5-Hz responses to random tones showed a latency shift to the right in the ERP response to random, infrequent target tones. The 4- to 7-Hz activity showed a prolongation of oscillation and a delay in maximum enhancement, or amplification, compared to 4- to 7-Hz activity of the ERP response to regular, frequent target tones. In the 7- to 17-Hz response, prolonged oscillation to random target stimuli was greatly reduced by application of regular target stimuli.

3.3.3 ORIENTATION REACTION

AND

LEARNING

DURING

REPETITIVE STIMULATION

Figure 3.20 shows averages of the first and last 20 sweeps for both target and nontarget tones from the experiment in which subjects were not informed that target tones would be presented regularly and alternately with nontarget tones. Minimal differences between the target and nontarget averages of the first 20 sweeps are readily seen, but clear differences can be seen between the respective averages of the last 20 sweeps. It appears, therefore, that despite the physiological regularity of the target stimuli, the cognitive uncertainty and possibly the incentive for gaining rewards resulted in similar P300 changes to both target and nontarget stimuli during the initial stages of the experiment. The average of the last 20 sweeps shows that P300 changes persisted in response to target tones and disappeared in the responses to nontarget tones. Upon questioning after the experiment, most subjects reported feeling confident about the regular stimulus pattern after 10 to 20 sweeps. As they became more certain of the stimulus pattern, they were able to identify and ignore the nontarget tones more confidently and this was reflected in the variations of the P300 responses to nontarget tones. Close inspection of Figure 3.20 reveals that the negative peak formed by theta and alpha oscillations around 200 ms (see above) is more pronounced in the average of the first 20 compared to the last 20 sweeps of target responses. Since this peak depends mostly on delayed theta (4 to 7 Hz) enhancement, it appears that this effect becomes less pronounced as subjects become more

FIGURE 3.20 Averages of first and last 20 sweeps for both target and nontarget tones from experiments in which subjects were not informed that target tones would be presented regularly and alternately with nontarget tones. Solid lines represent ERP to target (attend) tones; dashed lines represent ERP to nontarget (ignore) tones. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 57 Friday, May 7, 2004 8:47 AM

certain that every alternate tone will be a target. Hence, we have tentatively hypothesized that this change reflects a learning process following the initial orienting response (Pavlov, 1927). The progressive change can be seen quite clearly in the single sweeps.

3.4 REQUIREMENT OF PREPARATION RHYTHMS FOR ACTIVATION OF WORKING MEMORY: ANALYSIS OF PRE- AND POSTSTIMULUS ACTIVITY IN SINGLE SWEEPS Figure 3.21 shows single-sweep EEG-ERP epochs filtered in three different frequency ranges. The sweeps were recorded under various experimental conditions and subjects were asked to mentally mark the occurrences of target tones. Figure 3.21(A) shows 0.5- to 3.5-Hz activity in the responses to target tones (sweeps 43 through 61). The sweeps were recorded during the experimental paradigm in which subjects were informed that target tones would be presented regularly and alternately with nontarget tones. It can be seen that the general configuration of activity is very regular for both pre- and poststimulus segments. Furthermore, there is evidence of a stable phase reordering or preferred phase angle at the point of stimulation. The examples shown also evidence maximum negativity at the time of stimulation. Efforts were made to exclude methodological errors by recording baseline EEG epochs from all subjects. Eighty single-sweep EEGs were recorded with the same timing and data collection methods but without stimulation. Figure 3.21(B) shows a sample of baseline sweeps for comparison with Figure 3.21(A). Figure 3.21(B) shows no evidence of the pre- and poststimulus regularity seen in Figure 3.21(A). It should be emphasized that the percentage of single sweeps showing phase alignment at the point of stimulation can vary from 35 to 85% among subjects. The subject's cooperation and/or his capacity to focus attention may be important in determining this percentage.

3.4.1 EVENT-RELATED THETA OSCILLATIONS Figure 3.21(C) shows filtered theta (3.5 to 8 Hz) activity from single sweep EEG-ERP epochs recorded during an experiment in which target tones were presented randomly and infrequently (data taken from companion study). Prolonged regular oscillations and delayed enhancement can be seen after sweep 49, i.e., the 10th target tone that also marks a reduction of prestimulus theta amplitude. The onset of these changes varied among and within subjects and under different experimental conditions. Further studies are required to explain this variation in more detail. For present purposes, we wish merely to draw attention to the more obvious changes in pre- and poststimulus activity during the courses of various ERP experiments. One general observation was an increase in theta amplitude in experiments in which subjects were presented with regular, frequent target tones. The regular, frequent pattern is associated with magnitudes of 20 mV, whereas the random, infrequent pattern is usually associated with magnitudes below 10 mV. The theta responses to regular and frequent stimulation diminished progressively during the experiment. The increased prestimulus theta activity associated with regular, frequent target stimulus is unlikely to be an example of the driving phenomenon described by John (1967). In our experiment, the stimulation intervals were 10 to 12 times longer than the intrinsic theta frequency. Hence, it seems unlikely that our findings can be explained by a driving phenomenon.

3.4.2 EVENT-RELATED 10-HZ OSCILLATIONS Figure 3.21(D) shows filtered 8- to 13-Hz (alpha) activity from single sweep EEG-ERP epochs recorded during an experiment in which subjects were not informed that target tones would be presented regularly and alternately with nontarget tones. It can be seen that the magnitude of the prestimulus alpha activity increased progressively during the experiment and was associated with © 2004 by CRC Press, LLC

TF1663_book.fm Page 58 Friday, May 7, 2004 8:47 AM

FIGURE 3.21 (A) Single-sweep EEG–EP epochs obtained in response to repetitive regular target tones filtered in the 0.5- to 3.5-Hz frequency band. Sweeps 43 through 61 are shown. (B) Filtered EEG baseline epochs (no stimulation applied during recordings). The curves were pass-band filtered in the 0.5- to 3.5-Hz frequency band. (C) Filtered EEG–ERP epochs following randomly applied target tones filtered in the 3.5- to 8-Hz frequency band. Sweeps 2 through 80 are shown. The prolonged and enhanced 3.5- to 8-Hz (theta) deflection is observed after sweep 49. (D) Filtered EEG–ERP epochs to repetitively applied rate target tones. Filter limits are 8 to 13 Hz. Groups of sweeps (3 through 13 and 65 through 79) are separately illustrated to show relevant changes in EEG and ERP activities. (From Stampfer, H.G. and Basar, ¸ E. (1985), Int. J. Neurosci., 26, 181.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 59 Friday, May 7, 2004 8:47 AM

increasing phase alignment at the point of stimulation. These changes were evident in most subjects by approximately the 20th target tone although, again, variations in this regard were noted within and among subjects. Although similarities in the changes within different frequency bands during any one experiment were apparent, the timing or onset of these changes could vary among the different frequency bands. This suggests a degree of functional independence among different frequencies and will be an issue investigated in future studies. 3.4.2.1 Interim Summary The general conclusion drawn from these results is that when subjects perform tasks in relation to regular, frequently presented target stimuli, the regular pattern of stimulation induces more stable, predictable pre- and poststimulus activity. The onset of these changes can show within- and amongsubject variations as well as variations among the different frequencies in any single experiment. Informing subjects of the stimulus pattern appears to augment the regularity and synchronization of their EEGs, as though the knowledge somehow influences EEG rhythmicity. Figure 3.21 provides evidence for our hypothesis of an active functional relationship between prestimulus EEG activity and ERP changes manifested at N100, N200, P200, and P300 peaks in the averaged curves; the experiments also illustrate the progressive regularity of pre- and poststimulus activity and increasing phase alignment at the point of stimulation. According to Ruchkin and Sutton (1983), the P300 ERP change is determined more by the subject's expectancy concerning an event and the information provided by it, than by the physical characteristics of the stimulus that signals the event. The results in this chapter support this conclusion. We will provide more information about the mechanisms underlying pre- and poststimulus response variations under different experimental conditions in order to better explain the brain dynamics involved.

3.4.3 MODULATION

OF

P300 ACTIVITY

BY

PREPARATION RHYTHMS

Stampfer and Basar ¸ (1985 and 1988) found evidence of specific changes in theta responses to unexpected target stimuli, namely, delay of poststimulus enhancement and prolongation of oscillation (see previous sections). These findings are supported by results presented in this section. Both paradigms show that the above frequency components are mainly responsible for the development of the P200 waves seen in the averaged responses to nontarget stimuli and in sensoryevoked potentials. It would appear, therefore, that the same sensory mechanism is involved in both the P200 and P300 peaks, except that the reaction time is delayed and prolonged in the case of unexpected target stimuli. Unexpected target stimuli do not elicit “new” responses; they appear to trigger additional information processing in the brain that is reflected by delayed and prolonged sensory responses. If target stimuli are expected, the reaction time is not delayed and the response is less prolonged. This was demonstrated by experiments in which subjects were informed that target tones would be presented regularly and alternately. In these experiments with repetitive stimuli, we noted evidence of preparation rhythms in prestimulus EEG activity, namely regularity of activity within different frequency bands and alignment of phase angles at the time of stimulation. It would seem that the expectancy and experience of regular target stimuli lead to a physiological prestimulus synchronization that in turn reduces poststimulus delay and prolongations that form P300 waves. Bullock (1993) discussed the expectancy physiological function. The nervous system shows adaptation by preparedness in a variety of ways. Preparedness can be passive or active; it comes from genetic factors and can be learned as a response to normal and unimportant input. According to Bullock, associative learning always requires an expectation. He further states that expectations are at least as widespread in the animal kingdom as habituation and associative learning. See Chapter 4, Section 4, this volume. © 2004 by CRC Press, LLC

TF1663_book.fm Page 60 Friday, May 7, 2004 8:47 AM

According to Chapter 1, sensory-evoked potentials are generated by enhancement and frequency stabilization of EEG activity. The findings presented here extend this concept; rare, unexpected stimuli also give rise to enhancement and frequency stabilization but the enhancement is delayed and prolonged. This suggests that the rare target stimuli exposed further functional dimensions of the enhanced EEG oscillations. The dimensions are the delay and duration of the enhanced EEG synchronization within delta, theta, alpha, and higher frequency bands. See Chapters 6, 7, and 11, in this volume. For new types of experiments involving preferred states, readers are referred to Barry et al. (2003) and Chapter 5.

3.4.4 CONTROL OF LEARNABLE SEQUENCES BUILDING OF MEMORY TEMPLATES

BY

PRESTIMULUS EEG ACTIVITY

OR

It has become increasingly difficult to give a clear definition of the P300 wave or change. The original P300 component described by Sutton et al. (1965) has been succeeded by a variety of late positive components (Donald, 1983). It has not yet been established that all the late components possess the same properties. For example, it has been suggested that responses elicited by certain online decision tasks (Kutas et al., 1977) and responses elicited by highly novel stimuli (Squires, 1975) are outputs that originated from different mechanisms. In a review of several relevant studies, Donald (1983) concluded that in order to explain how stimulus rarity could affect P300 amplitude, it would be necessary to place the P300 wave at the output stage of a memory-based system that contains a record of prior stimulation. One possible version of such a memory record —the echoic trace — would register a template of the most recent stimulus features. This template would then feed back into a comparator that also received each new input. A mismatch would be registered whenever a low probability stimulus occurred because an echoic match to that stimulus would be unlikely to exist in storage. One result of a mismatch may be the P300 wave. This model provides a theoretical explanation for the P300 wave elicited both by presented and omitted OB stimuli. Donald (1983) suggested that the choice mismatch model was compatible with observations that the P300 wave is very large after the first stimulus in a new series (Ritter et al., 1968; Vaughan and Ritter, 1970) because no memory template of the first novel stimulus would elicit a mismatch response. We presented evidence that learning the pattern of stimulation affects the P300 change. The findings of Donchin et al (1973) are similar in this regard. They presented subjects with different sequences of target stimuli possessing the same probability of occurrence but varied the degree of regularity (Donald, 1983). Learnable sequences in which stimuli alternate in some predictable order produced smaller P300 responses than irregular sequences that were unfamiliar and unpredictable. The findings of our studies and those of Donchin et al. (1973) suggest that feed-forward from memory can influence P300 amplitude. If memory correctly predicts the input, the P300 response is reduced; if not, a mismatch is registered and a large P300 wave develops. Chapter 6 discusses memory-related oscillations. Based on the findings of this chapter, we can ask whether various EEG frequencies (delta, theta, alpha, and higher) function as templates of memory records. The development of prestimulus regularity and phase alignment may reflect STM templates that may feed into a comparator. When the target signal is predictable, the preparation waves are aligned and have a predictable phase alignment at the time of stimulation. In such cases, the P300 wave is moved back to the P200 position noted for nontarget responses and sensory-evoked potentials. It appears that the abolition of the P300 responses in single sweeps and the reduction of the P300 amplitude in the averaged response are the results of regularization of the prestimulus EEG. Furthermore, if the EEG components prior to stimulation show large amplitudes or if the input does not coincide with the preferred phase angle, the P300 wave and other ERP components are diminished. Based on the range of EEG frequencies and the various ERP changes, it is possible that several templates of different frequencies are involved in the evaluation of input. The learnable sequences are manifested with preferential states in the brain. We again refer to Barry et al. (2003a). © 2004 by CRC Press, LLC

TF1663_book.fm Page 61 Friday, May 7, 2004 8:47 AM

3.4.5 VARIED DEGREES OF AUGMENTATION OSCILLATIONS IN MEMORY TASKS

AND

PROLONGATION: GAMMA

As we will explain in Chapters 4 and 6, gamma oscillations are correlated with multiple functions and act as universal operators or codes of brain functional activity. In addition to the topolology of the gamma activity, other parameters are important for analyzing oscillatory correlates of brain function. These parameters are enhancement, time locking, phase locking, delay of oscillations, ˘ et al., 1996). One aspect related to the functional and prolongation of oscillations (Basar ¸ -Eroglu correlation of the gamma band is attention (Tiitinen et al., 1993); some of the other aspects are perception, STM, binding, and associative learning (Miltner et al., 1999). The analysis of one of the most complex cognitive functions revealed that frontal gamma activity increases during mul˘ et al., 1996a and b). tistabile visual perception (Basar ¸ -Eroglu Important progress in the investigation of cognitive processes has been made in the past two decades, especially in characterizing the neural substrate of working memory (WM). Human lesion studies suggest that WM depends on the activity of a number of cortical regions, primarily the prefrontal cortex (Frisk and Milner, 1990; Owen et al., 1996). Gevins et al. (1997) suggested that WM is a function of distributed systems with both task-specific and task-independent components. ˘ et al. (2000 and 2002) was to investigate the The aim of a new analysis by Basar ¸ -Eroglu relations of gamma responses and incrementally increasing working memory (WM) loads in a sensory-motor working memory paradigm consisting of three tasks. The first task was a simple choice reaction that functioned as a control task. The second task involved low WM demand and the third was based on high WM demand. The first task contained no element of WM function. The second (easy) WM task required the subjects to keep in mind the movement of a previous response (one trial back). The third and most difficult task required the subjects to remember the last two movements (two trials back). Figure 3.22 presents grand averaged ERPs and averaged gamma responses of 12 subjects to easy and difficult WM tasks. Gamma activity increased without exception in all subjects. Prestimulus (F [1.4, 15.1] = 7.91, p = 0.003) and poststimulus (F [1.4, 15.2] = 7.28, p = 0.004) gamma activity revealed highly significant effects of task but no other significant effects or interactions. In the prestimulus time window (p = 0.03) and poststimulus phase (p = 0.03), gamma activity was substantially higher for the difficult WM task than for the control task. ˘ et al. (2000 and 2002) clearly showed the parallel increase of The results of Basar ¸ -Eroglu gamma activity with increasing memory demands in all the studied locations. These findings confirm the distributed nature of gamma activity in cortical areas during WM tasks. The results showed four different levels of changes in gamma activity during WM performance: 1. During two difficult task paradigms, the level of ongoing gamma was increased; average rms value increased from approximately 0.1 mV. to approximately 0.03 mV. 2. The enhancement did not change drastically following stimulation. 3. The distribution of the gamma activity in the whole cortex occurred in a manner that was not observed upon simple sensory stimulation, in less difficult oddball paradigms, or in observation of ambiguous figures. 4. During this WM paradigm, the selectively distributed gamma system reached a high gamma activation state both between tasks and in task responses; the higher activation state gradually increased with increasing task difficulty. The observed prolongation of gamma oscillations suggests that the brain works longer during difficult WM tasks. This is similar to the findings with alpha and theta oscillations discussed in previous sections. Gamma activation is not the unique component in this WM paradigm. The gamma activation is superimposed with low frequency oscillations as Figure 3.22 shows.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 62 Friday, May 7, 2004 8:47 AM

FIGURE 3.22 (A) Wideband ERPs from three types of stimuli with increasing task difficulty. Electrode locations are Fz , Cz , Pz , and Oz . (B) ERPs filtered in gamma frequency band from three types of stimuli with ˘ increasing task difficulty. (From Basar ¸ -Eroglu, C. et al. (2002), Int. J. Psychophysiol., 45, 36.) © 2004 by CRC Press, LLC

TF1663_book.fm Page 63 Friday, May 7, 2004 8:47 AM

3.4.6 ACTION OF APLR ALLIANCE CIRCUITS

AND

HYPHOTHESIS CONCERNING REENTRANT

The results described in the foregoing sections imply that during sensory–cognitive processing, the oscillatory networks in neural populations are activated as recurrent or reentrant circuits. This behavior has been measured with various types of exogenous or endogenous stimuli as plasticity and evolving behavior of oscillatory prestimulus or poststimulus activity in the experiments described earlier. Seemingly, this reentry is conveyed or brought to resonant states by the reciprocal activation of the APLR alliance. Reciprocal activation is usually increased during learnable sequences (Section 3.4.4). The reentry principle proposed by Edelman (1977) and Tononi et al. (1992) will be discussed in Chapter 9.

3.4.7 HABITUATION Several chapters of this book demonstrate that if prestimulus activity at various frequency bands shows high amplitude regular rhythmic behavior, the response amplitude bears an inverse relationship to the prestimulus amplitude (Basar ¸ et al., 1976a and b, 1979a and b, 1980, 1983, and Chapter 5, this volume). In terms of our experiments, habituation implies a temporary storage of specific information about previous stimulation that results in a decline of the response to repeated stimulation (Donald, 1983). Temporary storage may be a function of or may be reflected in the regularized or synchronized prestimulus EEG activity. Our results do not allow us to draw a definite conclusion. However, we can conclude that the disappearance of the P300 changes is causally related to the development of regular, synchronized prestimulus EEG activity in various frequency bands and that the mechanism of habituation is somehow connected to this development. We will return to this discussion in Chapter 9.

3.4.8 AUGMENTATION OF KNOWLEDGE OR LEARNED MATERIAL IS REFLECTED REGULAR AND INCREASED ALPHA ACTIVITIES

BY

During and after activation of the APLR alliance, clear changes of expectation alpha and of P300 theta were observed. This indicates a need to introduce the concept of longer-acting memory in contrast to the persistent memory discussed in Chapter 9. Chapter 5 covers the infuences of prestimulus activity on brain responsiveness and will show that spontaneous EEGs can reflect intensive cognitive activations and serve as important causal factors controlling the manifestation of ERPs. We conclude that the oscillatory elements of the EEG can be considered as basic building blocks of dynamic or immediate memory.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 65 Friday, May 7, 2004 8:47 AM

and Memory4 Perception Related Oscillations in Whole Brain ˘ and Erol Basar Canan Basar-Eroglu ¸ ¸ 4.1 RELEVANCE OF CHAPTER This key chapter shows that cognitive and memory-related processes are distributed throughout the whole brain, based on cat experiments using chronically implanted electrodes and applying paradigms similar to those used for humans in different focused attention states, as described in Chapter 3. The cats were stimulated with repetitive auditory stimuli to induce expectancy states that were registered as changes of event-related potentials (ERPs) recorded in various intracranial structures. Multiple electrodes were placed in various layers of the hippocampus (HI), cortex, and brain stem. Two of the aims of this chapter are to focus the attention of readers on changes in neuronal activity of the hippocampus, reticular formation (RF), and sensory cortex, and to correlate them with the theta components that exist in ERPs at the single-cell level. We have chosen examples from the most pertinent experiments with regard to the studies of ERPs described in this book. The results presented in this chapter indicate the necessity for studying P300 response differences in a given structure (such as the HI) with methods of systems theory and a larger number of simultaneous recording electrodes. The crucial point is that only one of many generators may exist in the HI and others may exist in other brain structures such as the cortex and brain stem. The importance of frequency analysis should be taken into account because it is possible that distributed generators may have different weights, based on our analysis of frequency responses. Latency and amplitude of ERPs do not perfectly describe the entire coding of the responses. Without analysis of frequencies, cross-correlation, and coherence functions, the relations among generators cannot be properly established. The experiments of Sokolov (1975) citing the presence of comparator cells in the HI are pertinent to all studies related to memory and matching processes described in Section 7.2.6 and the complex matching that will be proposed in Section 9.5.4. The roles of feature detectors are cited in Figure 9.7 and the proposal contained in Chapter 11. A further aim of this chapter is to emphasize the problems related to localization of memory traces in human brain recordings with limited numbers of electrodes (Section 4.4.8).

4.2 THETA AND ALPHA RESPONSES IN CAT BRAINS DURING COGNITIVE AND MEMORY-RELATED TASKS 4.2.1 INTRODUCTION Chapter 3 described results from application of the brain dynamics research program on the human brain. This chapter demonstrates comparative data obtained from experiments with cat brains. We will describe ERP recordings in the auditory cortex or gyrus ectosylvian anterior (GEA), mesenphalic reticular formation (RF), and dorsal hippocampus (HI) in unrestrained and behaving cats.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 66 Friday, May 7, 2004 8:47 AM

The interpretation of these results is based on analysis of ERPs in the frequency domain. For an overview of measurements of P300-like potentials in animals, see Paller (1988).

4.2.2 METHODS AND PARADIGMS UTILIZED FROM FREELY MOVING CATS

FOR

OBTAINING P300 RESPONSES

The experiments described in this chapter were performed on freely moving female cats after surgical implantation of stainless steel electrodes 100 mm in diameter in the GEA, HI, and RF. Figure 4.1 shows the omitted stimulus paradigm utilized to elicit ERPs. Every session consisted of three experiments: (1) recording of electroencephalograms (EEGs) for control purposes and comparison with ERPs obtained in response to omitted stimuli; (2) recording of auditory-evoked potentials; (3) P300 (omitted stimulus paradigm). The tones were presented repetitively so as to produce anticipation of the times of occurrence of the omitted stimuli. A 2-KHz tone with intensity of 80 dB SPL, duration of 1 s, and stimulus interval of 2.5 s was used. The cats were allowed to move freely in cages placed in a dimly illuminated, soundproof, and echo-free room. The EEGs were monitored throughout the experimental sessions. Epochs containing movement artifacts and stages in which the EEGs of cortex recordings showed sleep spindles or slow waves in the cortex were eliminated offline after the recording sessions. The cats were naive; they had not been previously exposed to conditioning or training trials. Long experimental sessions were avoided to eliminate the effects of adaptation and fatigue that might develop. EEG

EEG

EEG

EP

EEG

EP

EEG

EP

EEG

EP

EEG

EP

SPONTANEOUS EEG

AEP

S EEG

S EP

EEG

S EP

EEG

ERP

P300

S

S

OS

3rd tone

4th tone

5th tone

S = Stimulus 80 dBA, 2000 Hz OS = Omitted stimulus

1s

1s 0.5 s

FIGURE 4.1 Omitted stimulus paradigm. Sequence of experiments used in triggering event-related potentials (ERPs) in cats. Every session consisted of three experiments: control (spontaneous) EEG (top); acoustical electrophysical (AEP) studies (middle); and ERPs with omitted stimuli (bottom). Every fifth stimulus is ˘ omitted. (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 67 Friday, May 7, 2004 8:47 AM

In order to analyze the ERPs in the time domain, a combined analysis of EEGs and ERPs was ˘ and Basar used. For methods, see Basar ¸ -Eroglu ¸ (1991). In order to elicit ERPs, omitted stimuli were used as targets (paradigm previously explained).

4.2.3 SYSTEMATIC ANALYSIS OF EFFECTS OF REPETITION RATE OF OMITTED TONES ON ERPS RECORDED FROM CAT HIPPOCAMPI A systematic study was performed to analyze the influence of repetition rate on ERP by varying the omission rates of stimuli. In the first series of experiments with five cats, we applied a paradigm in which every third, fifth, eighth, or tenth tone was omitted. Because this type of experiment requires long recording and analysis periods, only five cats were used. The hippocampal ERPs indicated that the procedures using every fifth tone or every eighth tone gave the most satisfactory results, as can be seen in Figure 4.2. Figure 4.3 shows the amplitude frequency characteristics (AFCs) computed from spontaneous EEG and ERP recordings in the hippocampus. A marked change of AFCs was the existence of a maximum around 5 Hz in comparison to theta frequency range. A less distinct peak was recorded between 10 and 20 Hz. Another prominent peak appeared at 40 Hz. Figure 4.4 presents results of experiments on the GEAs, RFs, and HIs of nine cats. The grand averages obtained from transient ERPs are shown at left. Standard deviations of the grand averages are shown at right. The most important information in this illustration is the characteristic frequency response coding revealed in the amplitude frequency characteristics (middle column). The marked

FIGURE 4.2 Effect of the rate of omission on ERPs recorded from the hippocampus. Grand average of five ˘ cats. Filter range: 1 to 30 Hz. (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 68 Friday, May 7, 2004 8:47 AM

FIGURE 4.3 Amplitude frequency characteristics of the hippocampus. Thick line = ERP. Thin line = control EEG. Abscissa: frequency in Hz. Ordinate: amplitude in relative units and decibels (dB). (Modified from ˘ Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

prominent frequency response of the hippocampus is at 5 Hz (theta response). The cortical and reticular formation responses, however, produced the most prominent results in the 10-Hz frequency range. In other words, although a P300 response in the transient ERPs was, as a rule, measured in the GEA, RF, and HI, frequency contents were not the same: The HI showed resonant properties in the theta frequency range, and the RF and GEA responded at the 10-Hz frequency range. Figure 4.5 shows latency and amplitude histograms of ERPs obtained in three experimental sessions with each of eight cats. On the left side is a latency histogram; the vertical axis represents the number of experiments performed in three sessions. In the GEAs of the cats, we observed ERPs with flat distribution of latencies between 250 and 400 ms. Most latencies of the RFs were centered around 300 ms. Because of the short latencies, high amplitudes, and excellent reproducibility of the P300 response, the HI produced the most distinct results among three brain structures investigated. Although all these results did not demonstrate that neural sources may be located in the hippocampus, they showed that P300 responses in cats may be moderated strongly by the activity of the HI.

4.2.4 UTILITY

OF

ANALYSIS

IN

FREQUENCY DOMAIN

Chapter 1 emphasized the importance of the Brain Dynamics Program including, the computation of AFCs and digital filtering. Using these procedures in the present analysis clearly indicated that various brain structures differ in the latencies and amplitudes of the peaks in the ERPs and also with respect to their frequency components. These components are preferred frequency channels of brain structures in response to omitted stimuli. Depending on the studied brain structures, different frequency components such as theta (3 to 8 Hz) and alpha (8 to 15 Hz) contributed to the P300 responses. Stampfer and Basar ¸ (1985) and Basar ¸ et al. (1984) demonstrated with both oddball (OB) and omitted stimuli-experiments that the P300 responses in humans were characterized by large increases in theta response components as revealed by the AFCs (see also Chapter 3). Accordingly, the frequency analysis of P300 responses presented in this chapter provides a promising model for the comparison of human and animal ERPs.

4.2.5 MULTIPLE ELECTRODES

IN

HIPPOCAMPUS

Event-related potentials (ERPs) of the cat hippocampus were investigated by using multiple implanted electrodes. The largest N200–P300 complex was recorded near the CA3 region of the © 2004 by CRC Press, LLC

TF1663_book.fm Page 69 Friday, May 7, 2004 8:47 AM

FIGURE 4.4 Averaged ERPs recorded from auditory cortices, hippocampi, and reticular formations of nine cats (left), amplitude frequency characteristics computed from ERPs shown in left column (middle), and ˘ corresponding standard deviations (right). (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

HI. In the last decade, the HI and/or limbic system were postulated as neuronal generators of ERPs in cats (Wilder et al., 1981; O'Connor and Starr, 1985) because of a polarity reversal of P300 responses in the HI. The purpose of the experiments described in this section is to determine whether the P300 responses varied in latency, amplitude, or polarity according to hippocampal layers. A multiple electrode array with four tips was placed in the right HI of each of eight cats. The electrode diameter was 25 mm and the distance between the tips of electrodes was 0.7 mm. Figure 4.6 shows the position of the array in the right HI. The electrodes were labeled HI1 through HI4. The first electrode was located in the upper pyramidal layer of the HI (CA1), the second between the upper pyramidal © 2004 by CRC Press, LLC

TF1663_book.fm Page 70 Friday, May 7, 2004 8:47 AM

Auditory cortex

A N 10

B

N 15 10

5

5 0 250

300

350

400 ms

0

25

50

75

µV

25

50

75

µV

25

50

75

µV

Occurrence

Hippocampus 10

15 10

5

5 0 250

300

350

400 ms

0

Reticular formation 10

15 10

5

5 0 250

300

350

400 ms

0

FIGURE 4.5 Histograms of ERPs of auditory cortex, hippocampus, and reticular formation. (A) Latency histogram. The vertical axis represents the number of experiments (three sessions with each of eight cats). The horizontal axis time shows time in milliseconds. (B) Amplitude histogram. The vertical axis again shows the number of experiments. The horizontal axis shows amplitude values (mV). (Modified from Basar ¸ ˘ Eroglu, C. et al. (1991b), Int. J. Neurosci., 60, 239–248.)

FIGURE 4.6 Cross-section of hippocampus showing locations of multielectrodes. CA1: upper pyramidal ˘ layer. CA2: lower pyramidal layer. (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

layer and the dentate gyrus, and the third and fourth in the lower pyramidal layer (CA3 and CA4). ˘ , 1990). For details about the accuracy of the electrode positions (Basar ¸ -Eroglu Figure 4.7 illustrates the ERPs in various layers of the cat HI with the configuration of electrodes as described in Figure 4.6. Figure 4.7A illustrates the grand averages obtained from eight cats. Each cat performed the experiments three times (n = 3 ¥ 8 = 24). The ERPs were digitally filtered between 1 and 30 Hz. Figure 4.7B shows characteristic results from one cat. In all the hippocampal positions, ERPs showed waves around N200 and P300. However, the HI3 and HI4 responses in the vicinity of CA3 of the HI were most marked. The grand average curves (n = 24) showed no marked N200 response in the upper layers of the HI and a flat P300. The clearest responses are again in the HI3 and HI4 positions. © 2004 by CRC Press, LLC

TF1663_book.fm Page 71 Friday, May 7, 2004 8:47 AM

FIGURE 4.7 Event-related potentials of hippocampus between CA1 and CA3 (labeled HI1 through HI4, respectively). A: grand average ERPs (eight cats, three experiments each). B: typical averaged ERPs of a ˘ single cat. (Modified from Basar ¸ -Eroglu, C. et al. (1991b), Int. J. Neurosci., 60, 239–248.)

Figure 4.8 shows the histogrammatic results based on Figure 4.7. Low pyramidal layers of HI (HI3 and HI4 corresponding CA3) showed the largest amplitudes. Figure 4.9 shows ERPs from a distant electrode placed outside the HI in the lateral thalamus (5 to 7 mm deeper than HI4 electrode). Neither a marked ERP nor a P300 component was observed. These results emphasize the possible existence of a strong P300 generator in the HI, especially near the CA3 layer. Figure 4.10 presents the AFCs of the HI computed from the curves shown in Figure 4.7. It clearly shows that the magnitudes of the frequency responses of the ERPs in several layers also depict significant differences. In the upper pyramidal layer (CA1) of the hippocampus (HI1 and HI2), the theta frequency response components were not prominent. The most marked theta eventrelated responses were observed usually in the CA3 layer. Again, higher frequency components (the so-called beta–gamma 25- to 40-Hz frequency range) were mostly higher in CA3 (HI3 and HI4) positions than the upper layers. The response in HI4 position centered at the 10-Hz frequency range with a shoulder in the theta frequency range. The HI4 response was mostly a mixture of theta and alpha components, whereas the HI3 response was an almost homogeneous theta response. © 2004 by CRC Press, LLC

TF1663_book.fm Page 72 Friday, May 7, 2004 8:47 AM

N 15

HI 1

10 5 0 10

20

30

40

50

15

60 µV HI 2

10 5 0 10

20

30

40

50

15

60 µV HI 3

10 5 0 10

20

30

40

50

15

60 µV HI 4

10 5 0 10

20

30

40

50

60 µV

FIGURE 4.8 Histograms of amplitude values (mV, peak to peak) of P300 response in the four hippocampal ˘ layers. (Modified from Basar ¸ -Eroglu, C. et al. (1991b), Int. J. Neurosci., 60, 215.)

FIGURE 4.9 ERP of distant electrode implanted at a level 5 to 7 mm deeper than CA3 of hippocampus (n ˘ = 3 cats). (Modified from Basar ¸ -Eroglu, C. et al. (1991b), Int. J. Neurosci., 60, 239–248.)

The grand average as an important methodological approach provided valuable parameters that globally described the occurrence of potentials in various brain structures. The grand averages demonstrated differences among various layers. The comparison of HI3 and HI4 responses demonstrated that marked differences even within a given structure with interelectrode distances of merely 0.7 mm may be obtained. The results of digital filtering also confirmed the results derived from the AFCs (see Figure 4.11).

© 2004 by CRC Press, LLC

TF1663_book.fm Page 73 Friday, May 7, 2004 8:47 AM

FIGURE 4.10 Amplitude frequency characteristics obtained from the averaged ERPs of the hippocampus. Abscissa: frequency in Hz. Ordinate: amplitude values in dB. A: results from eight cats, three experimental ˘ sessions each. B: results from one cat, one experimental session. (Modified from Basar ¸ -Eroglu, C. et al. (1991b), Int. J. Neurosci., 60, 239–248.)

4.2.6 HIPPOCAMPAL P300 AND COGNITIVE CORRELATES: THETA COMPONENTS IN CA3 LAYER Event-related potentials (ERPs) were elicited by means of omitted stimuli in various hippocampal layers of the cat brains. They showed increasing amplitudes toward deeper layers. In earlier evokedpotential (EP) studies (Basar ¸ , 1980), the frequency response depended on the electrode location in the HI. In the AFCs from the upper pyramidal layers, the 4 Hz and 40 Hz maxima were dominant in comparison to other structures. The studies of the dynamics of P300 in the cat brain described in this chapter demonstrate a differentiation of ERPs in the time and frequency domains in various hippocampal layers. In the time domain, the amplitudes of P300 waves increased toward the lower pyramidal layers. The © 2004 by CRC Press, LLC

TF1663_book.fm Page 74 Friday, May 7, 2004 8:47 AM

FIGURE 4.11 Filtered grand average ERPs of hippocampus computed from data shown in Figure 4.7. A: ˘ filtered in theta range (3 to 8 Hz). B: filtered in alpha range (8 to 15 Hz). (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

largest P300 was observed in the CA3 region. Neurophysiological theories suggest large amplitude and/or inverted polarity of P300 as criteria for generator processes in certain brain structures. By means of frequency analysis, a direct relationship between the amplitude of enhanced theta response and amplitude of P300 may be shown. We further suggest the theta response as a major subcomponent that plays an important role in hippocampal P300. Furthermore, evidence that the P300 theta response generators are not homogeneously distributed in various neuronal populations within the HI could be sought. An excellent review of neural generators in CA3 is given in Buszáki (1985). Cognitive correlates of hippocampal electrical activity have been discussed in the literature. Sokolov (1975), Vinogradova (1975), John (1967), and Vinogradova and Dudaeva (1972) pointed out that the cells in the hippocampus act like comparator cells and fire only to novelty and soon habituate; they return to life if the expected stimulus is altered. For an interesting report of the comparative neurology of expectation, see Bullock (1988a and b). Sokolov’s model (1975) concerning orienting response mentions expectation cells that fire according to the expected input; sensory reporting cells that fire according to actual stimulus; and comparator cells that fire whenever a discrepancy between the other two types of cells arises. Smith et al. (1990) gave evidence for a neocortical P300 generator in studies of the intracranial topography of P300 elicited with the oddball paradigm and noted that related activity also occurs © 2004 by CRC Press, LLC

TF1663_book.fm Page 75 Friday, May 7, 2004 8:47 AM

in the HI and probably in the frontal cortex. These authors conclude that activity generated in the HI and frontal lobe may only make minor contributions to scalp recordings and that a scalprecordable P300 response may be the most readily observable aspect of synchronous activity occurring across a widely distributed and highly integrated cortical network supporting cognitive activities. Pineda et al. (1987) discussed a late positive component in the frontal cortex of the monkey brain. The differences observed between squirrel monkey and human ERPs in terms of scalp topography and recovery periods are most likely consequences of differences in brain structure as well as possible differences in function. The authors also concluded that the marked similarities observed in morphology, responses to specific stimulus parameters, and the presence of analogous subcomponents suggest that a nonhuman primate model of P300 may be useful in investigating the anatomical structures and physiological mechanisms underlying human P300 activity.

4.3 COMPOUND P300–40-HZ RESPONSE OF CAT HIPPOCAMPUS 4.3.1 P300–40-HZ COMPOUND POTENTIAL The methods used were explained in Section 4.1.2. The cats were stimulated with tones of 2000 Hz and 80 dB. Every fifth stimulation was omitted. The omitted stimulation was the target signal. Figure 4.12 shows 10 single digitally filtered EEG epochs recorded during a P300 experimental session. The filter limits were in the range of 30 to 50 Hz. The averaged curve of 50 sweeps (also filtered between 30 and 50 Hz) from the same experimental session and the wideband-filtered ERP are shown at the bottom of the figure. The wideband-filtered averaged curve shows a marked ERP in the region of N200 and P300. Without filtering, it is not possible to recognize the 40-Hz components that may be masked by the low frequency activity. However, the filtered single epochs and the averaged curve show a 40-Hz burst between 270 and 300 ms following the omitted (target) stimulation. In a large number of single sweeps, 40-Hz wave packets can be seen. The maximal peak-topeak amplitudes of 40-Hz wave packets usually did not exceed 50 mV; as a rule, results were in the range of 20 mV. Time locking was weak. At the beginning of the experiments (the first 10 to 20 sweeps), single sweeps depicted weak or no phase locking. Better and nearly strong time locking was observed toward the end of the recording sessions. Figure 4.13 illustrates results of experiments from eight cats; curves were filtered between 30 and 50 Hz. All cats showed significant 40-Hz packets around 300 ms in the CA3 region of the HI. The 40-Hz oscillation packets showed jitters along the time axes. Close inspection of the responses at P300 latency from the four electrode locations showed that the responses with the highest amplitude were found in CA3 (HI3 and HI4) Figure 4.14 shows grand averages of wideband-filtered ERPs and 40-Hz responses. The grand averages were obtained from the ERPs of eight cats and were digitally filtered at 30 to 50 Hz. This illustration also shows that the slow wave activity (N200–P300 response) is marked mostly in the CA3 region, whereas the response in the region of the first or upper hippocampal electrode (HI1) ˘ et al., 1991b). was less significant (see Sections 4.1 and 4.2 and Basar ¸ -Eroglu The waves recorded in the first and second hippocampal locations (HI1 and HI2) did not show prominent N200 responses. The P300 waves and the accompanying 40-Hz oscillations were not as marked as those in the CA3 region. At the moment of the occurrence of omitted stimulation (the first 100 ms), small 40-Hz wave packets were also seen in a large number of sweeps, but usually they were not as high as the 40-Hz waves at the P300 location. The first observations of 40-Hz resonance to acoustical stimuli were reported in the cat GEA (Basar ¸ , 1972) and HI (Basar ¸ and Özesmi, 1972; Basar ¸ and Ungan, 1973). The 40-Hz resonance phenomenon (or enhancement) was further analyzed as a distributed property of various brain © 2004 by CRC Press, LLC

TF1663_book.fm Page 76 Friday, May 7, 2004 8:47 AM

FIGURE 4.12 Event-related potentials of lower pyramidal layer (CA3) of hippocampus (one cat). Top: single ERP sweeps (epochs) filtered at 30 to 50 Hz. Middle: averaged ERP filtered at 30 to 50 Hz. Bottom: unfiltered ˘ ERP, average of 50 artifact-free epochs. (Modified from Basar ¸ -Eroglu, C. and Basar, ¸ E. (1991), Int. J. Neurosci., 60, 227–237.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 77 Friday, May 7, 2004 8:47 AM

FIGURE 4.13 ERPs of eight cats recorded from lower pyramidal layer (CA3) of hippocampus. Waveforms filtered between 30 and 50 Hz. Each filtered ERP is the average of 50 to 60 artifact-free single epochs. ˘ (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

structures including the cerebellar cortex and it has been indicated as a principal component of the middle latency responses of the auditory pathway-evoked potentials (Basar ¸ , 1980). The P300–40-Hz response is very important from the cognitive process viewpoint. This gamma response is not phase locked. It is time locked and it appears after considerable delay, similar to the gamma responses observed by Singer and coworkers (Gray and Singer, 1987; Gray et al., 1989) and Eckhorn et al., 1988.

4.4 EVENT-RELATED OSCILLATIONS IN CAT HIPPOCAMPUS, CORTEX, AND RETICULAR FORMATION DURING STATES OF HIGH EXPECTANCY: COMPARISON WITH HUMAN DATA Bullock (1993) discussed the physiological function of expectancy and stated that the nervous system shows adaptation by preparedness in a variety of ways. The expectancy state can be passive or active; it originates from genetic factors and can be learned as a response to both normal and unimportant input. © 2004 by CRC Press, LLC

TF1663_book.fm Page 78 Friday, May 7, 2004 8:47 AM

FIGURE 4.14 Grand averages (mean values from eight cats) of ERPs in various layers of hippocampus. Top: locations of multielectrodes. CA1 and CA3 correspond to HI1 and HI3/HI4, respectively. Middle: unfiltered ˘ ERPs. Bottom: filtered ERPs (30 to 50 Hz; OS = omitted stimulation). (Modified from Basar ¸ -Eroglu, C. et al. (1991a), Int. J. Neurosci., 60, 215.)

Learned expectancy includes short- and long-term memory activation. According to Bullock, expectation refers to an inferred state of the memory system arising from a number of behavioral or physiological factors that are more or less specific. Sensory input is anticipated as a familiar event. A surprise may involve a familiar unanticipated movement or an unfamiliar stimulus. Expectation is often a form of learning. Associative learning always requires an expectation. Bullock further stated that expectations are at least as widespread in the animal kingdom as habituation and associative learning.

4.4.1 UNIT ACTIVITY

AND

BEHAVIOR

Single-unit recording from the hippocampi of intact behaving animals began with the pioneering studies of Vinogradova (1970), Olds and associates (Olds et al., 1969; Segal and Olds, 1972), O'Keefe and Dostrovsky (1971), and Ranck (1973). All these researchers noted that learning appeared to exert powerful influences on hippocampal unit activity. Among the most fascinating results are O’Keefe’s studies of spatial units in the HI (O’Keefe, 1976 and 1979; O’Keefe and Nadel, 1978). These authors described two types of responsive hippocampal units in freely moving rats: displace and place units. The displace units are Ranck's theta cells; during hippocampal theta activity, they show increases in overall firing rates in relation to theta waves. Hippocampal theta activity in rats is associated most commonly with behavioral movements of an exploratory nature (Vanderwolf, 1969; Vanderwolf et al., 1975; Black, 1975). Place units, on the other hand, respond when a rat is in a particular place in a maze. Both cell types are found in fields CA1 and CA3 of the HI (Ammon's horn), but place cells are not found in the dentate gyrus (O’Keefe and Nadel, 1978). © 2004 by CRC Press, LLC

TF1663_book.fm Page 79 Friday, May 7, 2004 8:47 AM

The characteristic of the second class of units in the CA1 region of the dorsal HI is that firing off is dependent on the rat's location in its environment. The part of the environment where the unit fires or fires maximally is called the place field. Some units fire maximally when the rat is in the place field and performing a specific behavior such as sniffing or receiving a specific stimulus. Some of these units are almost certainly identical to the units that Ranck (1973) called approachconsummate mismatch cells. Signal detection theory implies that two kinds of processes occur when an organism detects a threshold level signal: 1. The signal must be detected in noise at the level of sensory receptors and pathways. 2. A decision to respond must engage the neuronal circuitry of the learned response — the response the animal has been trained to make to indicate that the stimulus has been detected. Consequently, neuronal circuits that are differentially activated on detection and nondetection trials are candidate circuits for the decision–memory system.

4.4.2 EVENT-RELATED POTENTIALS IN CORTEX AND HIPPOCAMPUS IN A P300-LIKE PARADIGM What does it mean when a cat emits a P300 wave following the omitted stimuli? Assume the cat has heard four tones repetitively at constant intervals of 2.5 s. The cat must have performed a type of cognitive process and is probably in an expectancy state. The fifth tone that would have followed the fourth tone will be omitted. In other words, the cat must develop a type of expectancy or selective attention and its brain must be able to react by producing what is called the P300 wave when the fifth tone does not occur (see Bullock’s definition of expectancy above). If we accept this view, we can further admit that the evoked potentials of the cat brain should have various levels of cognitive processing during repetitive stimuli in comparison to the conventional evoked potentials where stimulation is applied randomly. The question is now whether the cat HI will generate different types of evoked potentials with changing amplitudes at various levels of cognitive processing with repetitive stimuli. We again underline our assumption that during such a paradigm, the cats are put in a complex state of expectancy, attention, learning, and short-term memory (STM). This means that all states of the brain (alertness, attentiveness, arousal, sensory memory, and STM) are in play and interaction (see Chapter 9).

4.4.3 SELECTIVELY DISTRIBUTED THETA SYSTEM: INVOLVEMENT FRONTAL, AND PARIETAL AREAS

OF

LIMBIC,

The omitted stimulus paradigm used on freely moving cats was also applied to human subjects. The responses showed considerable enhancement of theta components in the responses to the stimuli preceding the omitted ones (Demiralp and Basar ¸ , 1992; Figure 6.20, this volume). The theta response increases were significant over the frontal and parietal areas in the auditory modality. In visual modality, significant theta increases were more diffuse over the frontal, parietal, and occipital areas and at vertices, and again more pronounced at frontal and parietal sites. In light of these results, we are now searching for similar changes in the responses to stimuli that precede the omitted stimuli in cats with implanted electrodes. Certainly the cognitive contents of the omitted stimulus experiments carried out on passively learning cats and instructed human subjects are not exactly comparable. However, the presence of a P300-like response of the cat brain to the omitted stimulus (see Chapter 3) enables us to assume that the cats may also, to a certain

© 2004 by CRC Press, LLC

TF1663_book.fm Page 80 Friday, May 7, 2004 8:47 AM

extent, develop expectancy and focused attention in response to repetitive stimulation and the regular omission of a stimulus. The amplitude increase in the responses of the hippocampus and motor cortex to the stimulus preceding the omitted one was correlated with a selective amplitude increase in the theta frequency band in amplitude frequency characteristics (AFCs). Statistical tests of HI2, HI3, and HI4 recordings revealed that the amplitude increases in the theta band were significant. Accordingly, we assume that the experiments with cats add a highly significant explanation regarding the location of the theta increase. Theta rhythms as a common feature of the limbic structures are also involved in the mechanism of selective attention. In the light of our findings, we arrive at the following conclusion. The theta response enhancements in HI and frontal cortex areas of the cat brain and theta response enhancement in human frontal and parietal locations probably reflect the general responsiveness of the hippocampal–frontal–parietal system during focused attention and expectancy. The concept of theta resonance was analyzed extensively by Miller (1991) who proposed that Hebbian processes of synaptic strengthening select patterns of loops passing from hippocampus to cortex and back to the hippocampus. Miller’s view was explained in detail by Basar ¸ (1999). The theta response behavior presented in this chapter supports Miller’s interpretation. 4.4.3.1 Integrative Analysis of Increased Theta Response The present chapter demonstrates a selective strong theta component increase in the cat HI in response to stimuli preceding the omitted one. Expectancy and increased attention states may be developed by the cat at the onset time of the repetitive stimuli, similar to theta activity increases in frontal and parietal responses of human subjects (Demiralp and Basar ¸ , 1992). We postulated a selectively distributed theta system in the brain that is involved in the cognitive states of focused attention and expectancy. The parallelism between the theta response increases recorded in human neocortical areas and those recorded intracranially from cat hippocampi show that the HI plays an important role in this postulated theta system. The close anatomical relationships between the HI and neocortical association areas, especially the frontal and parietal association areas, suggest that the interaction of the HI with the neocortical association areas in the theta frequency band might be the basis of the theta response system involved in focused attention and expectancy.

4.4.4 INTERPRETATION

OF

CHANGES

IN

ERPS

What changed in the hippocampus during these types of experiments? What can we learn from experiments involving recording of evoked potentials and the evaluation of amplitude frequency characteristics? The hippocampus is a supramodal structure involved with motivation, emotion, and STM — in other words, all reactions in which humans and animals are tasked with comparing all peripheral and endogenous stimulation of the central nervous system (CNS) with events that happened earlier and were seemingly stored in short- or long-term memory. The primary cortex, for example, the auditory cortex, immediately processes information and reacts to sensations. The HI does not react to sensation without signal detection and comparison. In the experiments cited, the cats and humans were supposed to make judgments: what was the first stimulation and what was the second? Since the evoked potentials (EPs) recorded were not only related to sensation, but also to learning, attention, and reasoning processes, one would expect the HI to react in a completely different way in comparison to cortices and RFs. Indeed, the results of these types of experiments confirm this ability of the HI which shows a type of plasticity.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 81 Friday, May 7, 2004 8:47 AM

4.4.4.1 Comparison with Human Responses The following is a brief summary of our human and cat experiments. As stated earlier, terms such as alertness, emotion, attention, and short-term memory are used often by psychologists and physiologists. They are descriptive and often subjective. Accordingly, we will try to relate the experiments to observed electrophysiological changes. What happens during experiments in human subjects who have been told to note an omitted stimulation — the fourth or fifth after the first stimulation? They all reported they had to increase their expectancy at the third one in order to focus their attention on the fourth (omitted) one. In comparison to conventional EPs (recorded with random stimulation), frequency components of EPs for third tones showed significant increases of theta responses in frontal and parietal locations. Because it is not possible to ask cats about their expectations and feelings during presentation of repetitive and omitted stimuli, we must use objective criteria, for example, noting that the omitted stimulation gave rise to a type of P300 wave (see Chapter 3). The cats may experience a type of attentive stage somewhat comparable to reactions noted in human experiments with the same parameters and design. Although no undisputed technique exists for comparing human and animal experiments, similar experimental design and common significance of electrophysiological responses permit certain interpretations. What changes occur in the cat brain and human brain during repetitive stimulation? In vertex and occipital recordings of humans, few changes occurred; in parietal and frontal recordings, significant increases of theta responses were observed. In the cat brain, no significant changes were noted in the auditory cortex (in the primary sensory area), but important changes were found in the HI and frontal cortex. The significant changes were observed with an increase of the 4-Hz response (40%). Our immediate interpretation of these experiments is that if human subjects or cats (seemingly) pay attention to sensory stimulations, the EPs to the attended stimuli show significant increases in theta frequency responses. This is good accord with results obtained when we applied the third signal paradigm to human subjects. Again we noted significant increases in the theta frequency ranges in the attended channels and the significance was again high in frontal and parietal regions.

4.4.5 WHY COMPARE EP RESULTS

WITH

CONVENTIONAL EXPERIMENTS?

Early in this chapter we described in detail experiments on signal detection of the cat HI after functional firing of displace and place units. The conventional experiments demonstrated that these place units are theta cells that can be considered feature detectors. Hippocampal theta activity is described in most experiments as the dominant rhythmic reaction. Training also induces unit activity in the HI; in the decision memory system, the HI plays a more important role than the colliculus inferior and medial geniculate nucleus. The stimulation of the brain induced significant theta enhancements in the HI as a major response that was highly influenced by cognitive tasks. These results led to exploration of brain function with the help of ERPs. Factors other than movements (cats) and decision making (humans) induce theta rhythms. Target (cognitive) stimulations also give rise to genesis of high-amplitude theta components manifested as evoked potentials. An EEG originates as a brain signal stemming from unknown inputs to the CNS that can be considered internally evoked or induced evoked potentials. This view is strongly supported by the pioneering work of Adey (1966) that showed induced theta rhythms in the hippocampi of learning and exploring cats.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 82 Friday, May 7, 2004 8:47 AM

4.4.6 STRUCTURES INVOLVED

IN

STATES

OF

APLR ALLIANCE

4.4.6.1 Hippocampus as Supramodal Structure According to Mesulam et al. (1977), each primary sensory area projects first to unimodal association areas of the HI that primarily underlay sensory-specific discrimination functions. These projections are essential because different parts of a primary sensory field are usually not interrelated by association fibers. Unimodal association areas in turn project to sites of sensory convergence Mesulam et al. designated polymodal association areas. In the primate, these areas are found, for example, in parts of the prefrontal region and near the superior temporal sulcus. Finally, certain regions in the inferior parietal lobule and in other parts of the cingulate gyrus and prefrontal region appear to receive their major cortical inputs from polymodal association areas. These are called supramodal associations. Functional studies of the inferior parietal lobules in monkeys (Motter and Mountcastle, 1981) suggest that the supramodal cortex may be involved in complex functions associated with visual attention, spatial orientation, perception, and the integration of visual and manual operations. 4.4.6.2 Frontal Cortex Event-related potentials (ERPs) obtained with paradigms inducing focused attention, P300 reactions, and high expectancy states show marked electrophysiological changes in the frontal cortex, parietal cortex, and the limbic system. The frontal areas of the human cortex reacted with enormous theta enhancements to cognitive stimulation requiring states of focused attention and short-term memory (see Chapters 3 and 6). A theta enhancement increase of 50% was recorded while a subject paid attention to an expected target. Similar experiments with cats demonstrated an enhancement of 40% in the CA3 layer of the hippocampus. In P300 experiments, learning tasks again led to a theta increase with a time delay in frontal and parietal recordings (see Chapters 3 and 6). These results clearly demonstrate that tasks requiring attention give rise to marked theta increases in evoked potential components. When comparing the results of experiments with simple light or sound stimulation in which the EPs contained dominant alpha responses, we are inclined to state that cognitive loading increased the weight of theta components in comparison to alpha components. Furthermore, the increase in theta responses mostly occurred in frontal hippocampal or parietal structures. Even the omitted stimuli that gave rise to a P300 response in cat hippocampi had dominant theta components, again with the largest components in the CA3 layers. Based on results described in the previous sections, we propose that association processes are accompanied by theta enhancements (see Chapters 3, 6, and 8). As Fuster (1991) stated, the frontal cortex is highly involved in anticipation. When we applied repeated light stimulation to a subject and asked him to focus his attention to the third signal applied, he certainly anticipated the presentation of the target. Since anticipation is one of the major functions of frontal lobes, and since a significant theta increase is observed in the evoked potentials, we can assign the increase to the frontal cortex in linkage with the hippocampus — a dominant localization of theta processes. Experiments by Basar ¸ et al. (1998) showed clearly that frontal EPs are strongly controlled or influenced by the theta activity of the frontal cortex. Accordingly, it is assumed that theta activity is a major operating rhythm of the frontal cortex. According to Brandt et al. (1991), Rahn and Basar ¸ (1992), and Chapters 3 and 5 (this volume), the major operating rhythm of the occipital area is alpha. It is certain that alpha and theta responses can be detected in various intracranial and cortical structures. The existence of a significant difference in the major operating rhythms in occipital or frontal areas strongly supports the possibility that spontaneous, evoked, and induced theta and alpha rhythms have fundamentally different functional operations. During some functional states, major operating rhythms can change their functional roles; the nature of the experiment (task) can influence these functional components on brain rhythms. © 2004 by CRC Press, LLC

TF1663_book.fm Page 83 Friday, May 7, 2004 8:47 AM

4.4.6.3 Global Function of Reticular Formation The reticular formation (RF) is a complicated polysensory structure of the brain stem. It is a minute neural network in the central part of the brain that processes awareness of the world and our ability to think, learn, and act. The seat of the power to think and perceive, indeed to respond to a stimulus with something more than a reflex reaction, lies in the cortex of the brain, but the cortex cannot perceive or think unless it is awake. The brain stem system and its linkage to the thalamic system were discovered by Moruzzi and Magoun (1949). They mapped the cat brain stem using electrical shocks and showed that high frequency stimulation in the core of the brain stem produces arousal responses in the cortex. The sites of stimulation generally correlated with the RF. Lesions of the RF produced a state of deep sleep. These studies suggested the hypothesis that arousal is mediated by a reticular activating system stimulated by sensory collaterals and activated through nonspecific thalamic nuclei. Is RF a polysensory high-command structure? Although we do not mention all the functions attributed to the RF here, we will describe a relevant working hypothesis formulated by HernandezPeón (1961): 1. The brain stem reticular system is a region where impulses of all sensory modalities converge. It is reached by impulses from the lower segments of the specific afferent paths and by those arising from the cortical receiving areas. 2. The same central region can decrease or increase the excitability of most sensory neurons. Therefore, it is able to inhibit or facilitate sensory transmissions at all levels of the specific afferent paths. The centrifugal control of sensory paths exerted by the reticular system is tonic and selective. According to the hypothesis of Hernandez-Peón, the core of the brain stem may be viewed as a form of high command that constantly receives and controls all information from the external and internal environments and from other parts of the brain. At a given moment, only a limited part of the information reaches this central area and a large number of informing signals are excluded. The exclusion of afferent impulses from sensory receptors takes place just as the impulses enter the CNS. Therefore, it is assumed that the first sensory synapse functions as a valve in which sensory filtering occurs. This may mean that the reticular mechanism of sensory filtering is formed by a feedback loop with an ascending segment from second-order sensory neurons to the RF and descending limb carrying impulses in the opposite direction. Hernandez-Peón states that it is unlikely that both centripetal and centrifugal limbs of the loop contain specific facilitatory and inhibitory neuronal connections. Such an arrangement would prevent over-activation of sensory neurons and the resulting excessive bombardment of the brain by afferent impulses. Thus, the dynamic equilibrium operating at the entrance gates of the CNS would preserve the delicate and selective mechanisms of sensory integration. We refer to the flow charts in Chapter 9 in order to see the alpha and theta enhancements in the RF that react with alpha responses to both auditory and visual modalities. Compare also Figure 6.5. 4.4.6.4 Cognitive Functions of Cerebellum Although all textbooks describe functional correlates of the cerebellum as being related to movements and motor processes in general, this view is now changing because of accumulated new data on cognitive cerebellar functions. Leiner et al. (1993) described new evoked connections between the cerebellum and cerebral cortex and reviewed results of published data concerning cerebellar participation in human neural functions including:

© 2004 by CRC Press, LLC

TF1663_book.fm Page 84 Friday, May 7, 2004 8:47 AM

1. 2. 3. 4. 5.

Cerebellar activation during mental imagery Cerebellar activation during word processing Cognitive deficit in rapidly shifting attention between sensory modalities Cognitive planning and cognitive operation in three-dimensional space Verbal and nonverbal intelligence and verbal associative learning

The review of Leiner et al. (1993) includes data on measurements made by positron emission tomography (PET), single-photon emission computer tomography (SPECT), and computer tomography. Important information emerged from studies with the cat cerebellum (Basar ¸ , 1999). The cerebellar cortex has dynamics similar to those of other structures of the brain. The frequency characteristics contain, for example, maxima at 10 Hz, 40 Hz, and other known frequency bands. Since we used auditory stimulation to analyze evoked potentials, we can tentatively assume that after the stimulation auditory associative reactions are processed also at the cerebellar level. We can definitely state that cerebellar dynamics are perfectly coordinated (correlated) with the integrative dynamics of the entire brain.

4.4.7 SECONDARY ALPHA RESPONSE

AND

ALPHA RESPONSE

WITH

DELAY

Chapter 6 will show that inadequate stimuli cannot generate significant and time-locked cortical alpha enhancements in the first 300 ms after stimulation. The occipital cortex of the cat brain does not respond with enhanced 10-Hz if the inadequate stimulation is auditory. With these types of potentials in the poststimulus interval between 250 and 300 ms, a small 10-Hz enhancement can be recognized. The time locking is weak in comparison to results with adequate stimuli. Therefore, although single sweeps may contain fairly high 10-Hz amplitudes, the responses are delayed and are not perfectly phase-locked. These types of responses are also obtained in the auditory cortex via visual stimulation (inadequate stimulation) and emitted 10-Hz oscillations following the omitted stimuli in the cortex and reticular formation of the cat brain (see Figure 4.4). These secondary or delayed 10-Hz responses were also recorded from human brains during cross-modality experiments with EEG (see Chapter 10) and magnetoencephalography (MEG) recordings. What can cause the delayed oscillatory 10-Hz activities we call secondary alpha enhancements? Thalamic 10-Hz enhancements are only recorded if stimulation is adequate (Basar ¸ et al.,1997). This means the medial geniculate nucleus depicts 10-Hz alpha enhancements only if auditory stimulation has been applied, but does not react to visual stimulation with an immediate (within the first 200 ms) 10-Hz enhancement. In contrast, the HI always reacts with ample 10-H oscillatory behavior upon stimulation. This oscillatory response is around 9 to 10 Hz for auditory stimulation and approximately 12 Hz for visual stimulation. It seems that gates in the HI open to every type of sensory stimulation and that this structure merits Swansson’s designation of “supramodal gate par excellence” (1981). We focus our attention now on the model for generation of resonance loops between hippocampus and cortex as described by Miller (1991) who estimated the time of signal transfer through hippocampal formation, prefrontal cortex, and nonlimbic association cortex by using conduction velocities. According to Miller, the total loop time may be more than 200 ms, which means conveying information one way from the hippocampal formation to the nonlimbic association cortex may take 120 to 200 ms. According to results presented in this book, every sensory stimulation evoked 10-Hz oscillations with durations of 250 to 300 ms in the HI. An auditory sensory stimulation will evoke 10-Hz oscillatory waves; a visual stimulation will then trigger a 12-Hz oscillation immediately after the auditory stimulation. This sensory information arising upon stimulation is probably conveyed to

© 2004 by CRC Press, LLC

TF1663_book.fm Page 85 Friday, May 7, 2004 8:47 AM

ry da it on ircu c Se ha c alp

Cortex Primary sensory

Association cortex

Limbic alpha system (supramodal)

Thalamic alpha system (sensory) Thalamus

Limbic system hippocampus

Brain stem

FIGURE 4.15 Hypothetical diagram of primary and secondary alpha circuits. (Modified from Basar, ¸ E. (1999), Brain Function and Oscillations. II. Integrative Brain Function: Neurophysiology and Cognitive Processes, Springer, Berlin, p. 1.)

the association cortex, to frontal nonlimbic association cortex, and to other parts of the cortex, including primary sensory areas. A current spread or volume conduction may also exist, but most probably the volume conduction is negligible because of the long distances between auditory and visual cortices. To compensate when recording a response with our electrodes in a primary sensory area, we add 20 to 50 ms. In other words, if the HI generates an oscillatory 10-Hz pattern following every stimulation mode, and if the HI transmits this signal to other association areas, the primary sensory areas would receive data about 150 to 200 ms after the application of the stimulation. Accordingly, it can be hypothesized that the secondary 10-Hz oscillations recorded in the primary visual or auditory areas have their pacemaker sources in the HI or in the brain stem. Our data may show that such a transfer of information in the 10-Hz frequency range from HI to cortex is possible. The HI receives signals from brain stem structures such as the RF (see Section 9.2). If an auditory stimulation generates 10-Hz activity in the RF, hypothetically the signal can be transferred to the HI and from there to the prefrontal cortex and nonlimbic association cortex. In this case, it is possible that the signal can reach the primary sensory areas too via the nonlimbic association cortex, as recorded in the cortical potentials (Figure 4.15). Specific afferents from sense organs reach the primary cortical areas and the same sensory information is sent to association areas of the cortex via the limbic system (HI); the sensory input reaches the mesencephalic RF. As stated by Hernandez-Peón (1961), the reticular formation acts as a filter and is another gate structure par excellence that controls information flow to the primary sensory areas, the limbic system, and polymodal association areas of the cortex. The 10-Hz response recorded in the RF also shows marked enhancements similar to those in the HI. The information flow over the mesencephalic RF also produces delayed signals. This transmission over the RF can be considered an additional system acting parallel to the prolongation of the 10-Hz activities. If we follow the same scheme cited for the HI, the delay of signals over the RF would be again in the range of 200 to 300 ms. We must emphasize the experiments of Schürmann et al. (2000) showing the relevance of cortico-hippocampal coherences and the concept of selectively distributed coherence explained in Chapter 6. These new results support the role of the hippocampus in 10-Hz organization illustrated in Figure 4.15.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 86 Friday, May 7, 2004 8:47 AM

4.4.8 COMPARISON

WITH

HUMAN BRAIN RESULTS

Fernandez et al. (1999) performed experiments to identify neural correlates of memory formation by using intracerebral electrodes implanted in the brains of patients with temporal lobe epilepsy. ERPs were recorded directly from the medial temporal lobe as a patient studied single words. ERPs elicited by words subsequently recalled in a memory test were contrasted with ERPs generated by unrecalled words. Memory formation was associated with distinct but interrelated ERP differences within the rhinal cortex and the HI that arose after approximately 300 and 500 ms, respectively. The authors indicate that the rhinal cortex and HI participate in human declarative memory formation. The data provide evidence that this mnemonic operation was dissociated into subprocesses that were executed sequentially and initiated within about 300 ms. This is in line with anatomical data indicating a close interaction between the rhinal cortex and HI. Fernandez’ results (1999) and results from imaging studies imply a modularly structured and serially organized memory system comprising the posterior parahippocampal cortex, rhinal cortex, and HI. Fell et al. (2001) point out the relevance of phase synchronization of EEG oscillations around 40 Hz in the medial temporal lobe. According to data presented throughout this book, particularly those in this chapter, we comment on Fernandez et al.’s results (1999) as follows: 1. The delayed gamma response superimposed with theta response in the CA3 layer of the ˘ and Basar HI was measured by Basar ¸ -Eroglu ¸ (1991) who designated it the P300–40 Hz response (see Figure 4.12). 2. As a consequence, the omitted stimulation paradigm induced a gamma response similar to the reaction to a working memory task. A recent publication by Haig et al. (2000) confirmed the P300–40 Hz response also in healthy human scalp recordings. Accordingly, the correlation of the ERP late window and 40-Hz response is established in recordings from epileptic patients, cat data, and in studies of healthy human subjects. 3. According to the cat data presented in this chapter and in Chapters 3 and 6, activations of theta and gamma components are selectively distributed in the whole brain. ERPs in the cat brain have also been measured in the sensory cortices and RFs of the brain stem. We argue that it is impossible to localize a memory system in a given structure of the brain without parallel recordings from long-distance structures. As the coherence studies also indicate (see Figure 6.25), every simple task produces strong coherences in the alpha and theta frequency ranges. Therefore, the interpretation of the important results of Fernandez et al. (1999) should be limited to confirming the association of limbic gamma responses, working memory tasks, and late ERPs. 4. The statements of Fuster (1995 and 1997), fMRI findings by Courtney el al. (1997), and results of ERP experiments related to the recognition of grandmother pictures emphasize the impossibility of localizing neural activations during memory processes. We must note that the results mentioned in Chapter 6 are different, but converge to the same statement (see Section 8.4.6). 5. The delays and the appearance of the second gamma window could be interpreted as results of the Hebb mechanisms explained in Chapters 1 and 2.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 87 Friday, May 7, 2004 8:47 AM

Factors Controlling 5 Causal Brain Responsiveness and Memory: Prestimulus EEG Activity, Entropy, and Genetics Most functions require the integrated action of neurons located in many regions. Localization of function means that certain areas of the brain are more concerned with one kind of function than with others. The existence of a significant difference in the major operating rhythms in occipital or frontal areas gives strong support to the possibility that theta (spontaneous, evoked, induced) and alpha rhythms (spontaneous, evoked, induced) have fundamentally different functional operations. But during some functional states, major operating rhythms can change their functional roles; the nature of the experiment, i.e., tasks, can influence the weight of these functional components on brain rhythms. Erol Basar ¸ (1999)

5.1 INTRODUCTION The principle of causality has an important role in the prediction of changes in dynamic systems. According to the principle, a strong relationship exists between the cause and the effect; the cause always precedes the effect. In Newtonian dynamics, the relation between cause and effect is strongly linear. Causality assumes a statistical nature in quantum dynamics and gas laws Section 1.5.2 and Section 1.5.3. Causality also has a statistical nature in the brain, as described by Basar ¸ (1980) with several analogies to resonant systems in nature. With brain responsiveness, the precedent state of a recording site, i.e., the state of the electroencephalogram (EEG) oscillations in a given area that precede exogenous or endogenous stimulation is an important causal factor. Based on this fact, the state of the brain must be analyzed before we can understand brain reactions to all types of stimuli and understand the mechanisms of recognition and remembering. This chapter covers two causes that influence brain responsiveness: (1) amplitude of oscillations preceding stimulations and (2) changes in entropy of brain states. Causality related to the age of a human brain will be discussed in Chapter 9. Causality in 40-Hz responses of individual subjects is explained in Section 6.3.3.2 on the basis of recent results of Karakas¸ et al. (2003). Sections 9.2 and 9.3 will reemphasize the importance of searching for causality behind EEG oscillations. The role of prestimulus EEG activity in brain responsiveness and short-term memory was explained in Chapter 3. This chapter further discusses prestimulus EEGs and brain responses will be analyzed in a more analytical manner, thus allowing the interpretation of the prestimulus EEG as an important causal factor for generation of brain responses. This causality — which is strongly related to endogenous brain activity and in turn related to the brain’s cognitive states — is also an important controlling factor for the reciprocal activation of functions of the attention, perception, learning, and remembering (APLR) alliance. Chapter 11 will outline causality and the plasticity of brain states.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 88 Friday, May 7, 2004 8:47 AM

5.2 RELATIONSHIP OF EEG AND ERP Several parameters of the prestimulus EEG have been related to the amplitudes and latencies of components of event-related potentials (ERPs), as noted in Chapter 3. Basar ¸ (1980) stated that an average ERP says little about the brain’s responses to an individual stimulus. The responses to single stimuli may vary widely, largely depending on the nature of the EEG at stimulus onset. Accordingly, background EEG activity is seen to reflect the brain’s momentary state of activity and the background activity in turn determines the response to the stimulus. For alpha and theta frequency bands, an inverse relationship between the root–mean–square (rms) voltage level of the prestimulus EEG and the maximal poststimulus evoked potential (EP) amplitude was established by single-trial experiments in human subjects and cats (Basar ¸ , 1980; Basar ¸ et al., 1989a). A number of studies subsequently examined this hypothesis. Special emphasis was placed on the influence of prestimulus spectral EEG patterns reflecting central nervous system (CNS) levels of activation (Romani et al., 1988; Brandt et al., 1991a) and the phase angle of the spontaneous activity at the time of stimulation (Jansen and Brandt, 1991; Barry et al., 2003). Others studied the effects of the presence or absence of occipital alpha activity (Maras et al., 1990; Brandt et al., 1991b). Arieli et al. (1996) confirmed the inverse relationship at the cellular level. A synchronized pattern EEG reduces the probability of marked time-locked responses after sensory stimulation (Chapter 3). Accordingly, the amplitudes of the background EEGs are considered to be the main factors influencing EP amplitudes; ERPs can be predicted from EEGs.

5.3 ALGORITHM FOR SELECTIVE AVERAGING According to the selective stimulation paradigm of Rahn and Basar ¸ (1993a), stimulation was EEG dependent. The rms values of the alpha band (8 to 14 Hz), the theta band (4 to 8 Hz), and a combination of both (4 to 14 Hz) were computed online. The activity was checked to see whether it exceeded a predetermined rms level in a 1000-ms period. If a stimulus was triggered, the next one could occur after a specified time interval that was identical to the mean interstimulus intervals (ISIs) of the controls. Five or six stimuli frequently occurred in series. All three frequency bands were used to control stimulus application, each with its respective control experiment. The selective stimulation paradigm is shown schematically in Figure 5.1.

5.3.1 DEPENDENCE OF EP AMPLITUDES AND WAVEFORMS ON PRESTIMULUS EEG: VERTEX RECORDINGS 5.3.1.1 Auditory-Evoked Potentials Figure 5.2 allows a direct comparison of selectively averaged auditory-evoked potential (AEP) applications and controls. Superimposed evoked responses of 10 subjects under different experimental conditions are shown. The grand averages depict clear amplitude differences between selective and conventional stimulus conditions. The percent gain in amplitude of the EPs of the selective stimulation experiments was different for the three frequency bands. Evoked potentials with prestimulus amplitude restrictions in a single frequency alpha or theta band showed comparable mean increases in amplitude of 47% (p < 0.01) and 41% (p < 0.05), respectively. Evoked potentials with low prestimulus activity in the broad range of 4 to 14 Hz did not differ significantly (mean of 28%) from the control condition (Wilcoxon–Wilcox test). If the ISI correction is taken into account, alpha- or theta-dependent stimulation affected mean amplitude increases of 30 to 35% (nearly 20% for the third condition). To quantify these observations in single trials with little standard deviation, correlation coefficients between single trials and averaged EPs both filtered in the 0.5- to 30-Hz range were evaluated

© 2004 by CRC Press, LLC

TF1663_book.fm Page 89 Friday, May 7, 2004 8:47 AM

FIGURE 5.1 Concept of selective stimulation. Stimulation is triggered only in the absence of highly synchronized alpha or theta EEG activity. Selective stimulation in a defined frequency band is supposed to result in a reduction of the number of stimuli required to produce a clearly recognizable, measurable EP. (From Rahn, E. and Basar, ¸ E. (1993), Int. J. Neurosci., 7, 123.)

within 300 ms after stimulus onset. Median correlation coefficients were 0.66 for alpha, 0.67 for theta, 0.58 for the alpha-and-theta contingent AEPs, and 0.52 for controls. Significant differences were found between alpha (p < 0.01) and theta (p < 0.05) band contingent experiments and controls (Wilcoxon–Wilcox test). Standardization of prestimulus EEG conditions by quantification of prestimulus frequency band activities seemed to decrease the variability of single trials. 5.3.1.2 Visual-Evoked Potentials The paradigms for auditory modality were used also for visual stimulation to test (1) whether similar amplitude enhancements occurred in visual-evoked potentials and, if so (2) to what extent the enhancements appear on several different recording sites. The results confirmed the findings in the auditory modality. A marked increase (about 35% at the vertex electrode) in bioelectrical activity was recorded upon application of a selective stimulation paradigm. Moreover, this effect was observed around Cz in frontal, temporal, and parietal sites (Rahn and Basar ¸ 1993b). Figure 5.3 shows transient responses under both conditions for all 12 subjects under study, the selective stimulation visual EPs (bottom) and matching controls (top). There was an obvious increase in the maximal peak-to-peak amplitudes in poststimulus range if the stimulus application was contingent upon prestimulus activities. The alpha, theta, and alpha-and-theta band contingent visual EPs depicted significant increases at the vertex. The visual responses to the rms contingent stimuli at the vertex location showed marked increases in the range of 35% in the amplitudes of the N1–P2 complexes compared to the standard visual EPs.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 90 Friday, May 7, 2004 8:47 AM

FIGURE 5.2 Superimposed individual auditory EPs (top) and grand averages (bottom) of all subjects under study (n = 10) obtained during different experimental conditions of conventional or selective stimulation. From Rahn, E. and Basar, ¸ E. (1993a), Int. J. Neurosci., 7, 123.)

FIGURE 5.3 Superimposed averaged visual EPs of 12 subjects recorded at the vertex and filtered at 1 to 45 Hz. Standard visual EPs (top) and corresponding selective stimulation visual EPs (bottom). Control parameters: (A) prestimulus alpha activity; (B) prestimulus theta activity; (C) prestimulus alpha and theta activities. (From Rahn, E. and Basar, ¸ E. (1993b), Int. J. Neurosci., 7, 123.)

5.3.1.3 Topographic Aspects Figure 5.4 is a histogram representation of the medians and 95% confidence intervals of percent amplitude increases of the selective stimulation visual EPs compared to the controls, sorted according to experimental conditions and recording sites. The histogram representation allows a direct © 2004 by CRC Press, LLC

TF1663_book.fm Page 91 Friday, May 7, 2004 8:47 AM

FIGURE 5.4 Percent presentation of the amplitude increase of selective stimulation visual EPs versus standard visual EPs (set to 100%) based on median values of 12 subjects. (From Rahn, E. and Basar, ¸ E. (1993b), Int. J. Neurosci., 7, 123.)

visual comparison of the amplitude increases together with scalp distribution under three different prestimulus conditions. Open bars correspond to alpha as a control parameter; the diagonally striped bars show theta results; the laterally striped bars refer to both frequency bands as control parameter. At the vertex used as the reference point for EEG filtering, rms value computation, and evaluation, the rms contingent stimulation led to an amplitude gain of roughly 35%. The amplitudes of the visual EPs obtained by prestimulus alpha contingent stimulation showed an increase of 37% (p < 0.01); the theta contingent visual EPs, 35% (p < 0.05); and the alpha-andtheta contingent, 38% (p < 0.05). The frontal, temporal, and parietal recording sites showed amplitude enlargements comparable to that of the vertex, whereas little effect appeared in the occipital region.

5.3.2 FRONTAL VISUAL-EVOKED POTENTIALS In addition to checking the validity of the algorithm, frontal-evoked potentials were also studied to gain further insight into the response susceptibility of frontal lobes in theta frequency ranges. The F4 lead was used as the input channel directing selective stimulation. In three series, alpha, theta, and alpha-and-theta components were used as control parameters for stimulus triggering. The alpha and alpha-and-theta contingent selective stimulation conditions resulted in significant amplitude increases (p < 0.05) at the input reference channel F4, partly at ipsilateral temporal and parietal leads, and at Cz. The most significant increase of 35% at F4 (p < 0.01) resulted if stimulation was contingent on prestimulus theta components. It was concluded that the major operating rhythm (MOR) of © 2004 by CRC Press, LLC

TF1663_book.fm Page 92 Friday, May 7, 2004 8:47 AM

F4 F3

+ 10%

T4

Cz

T3

P3

P4

O1

O2

– 10%

Alpha

Theta

Alpha and Theta

FIGURE 5.5 Effect of selective stimulation on amplitudes of wide-band filtered evoked potentials. The changes in the amplitudes of the selective stimulation visual EPs as the percentage of standard visual EP amplitudes (set to 100%), filtered 1-30 Hz. The different bar styles refer to control parameters Alpha; Theta; and Alpha and Theta, respectively. Median values of nine subjects are presented (From Basar, ¸ E. et al. (1998), Electroencephalogr. Clin. Neurophysiol., 108, 101.)

frontal lobes is in the theta frequency band. This means that frontal-evoked potential analysis should consider the theta states of the brain for the interpretation of variability of frontal EPs. The percent changes of the amplitudes of the selective stimulation visual EPs and those of controls were computed for all recording sites. Figure 5.5 is a histogram representing the medians of amplitude increases (in percent) of the selective stimulation visual EPs compared to the controls, sorted according to experimental conditions and recording sites. The black bars correspond to theta as a control parameter. At the F4 location — the input channel for EEG filtering, rms value computation, and evaluation — all three selective stimulation conditions effected amplitude increases in comparison to the respective controls. The most distinct difference can be observed between the theta-dependent visual EPs and the corresponding controls. The theta contingent stimulation led to an amplitude gain of about 35% (p < 0.01). This high significance level was found only for theta condition at F4.

5.3.3 INVERSE RELATIONS

OF

EEGS

AND

VISUAL RESPONSES

Systematic analyses of combined epochs of EEG and EPs in the auditory and visual pathways of the cat brain led to the concept of an inverse relationship between the amplitudes of prestimulus and poststimulus activities (Basar ¸ , 1980). Recent studies (Rahn and Basar ¸ , 1993a and b; Basar ¸ et al., 1998) using visual and auditory stimulation suggest that different states of EEG rhythms contribute directly to differences in amplitudes of the evoked cortical responses. The following results are pertinent:

© 2004 by CRC Press, LLC

TF1663_book.fm Page 93 Friday, May 7, 2004 8:47 AM

1. Vertex EPs elicited by auditory stimulation during periods of low amplitude alpha or theta band activity showed about 40% higher amplitudes than conventional AEPs. 2. Findings in the auditory modality were confirmed and extended in the visual modality. At the vertex, the visual EP amplitudes increased about 35% compared to controls when prestimulus alpha or theta amplitudes did not exceed certain amplitude levels. Visual EP amplitudes increased also at neighboring electrode sites. 3. An inverse relationship between visual EP amplitudes and spontaneous EEGs immediately preceding stimulation was also seen for frontal recordings. Selective stimulation in the frontal area differed in two major aspects from the vertex-contingent stimulation used in previous studies: (1) the amplitude enhancements due to selective stimulation were localized and not consistently found all over the head; (2) alpha activity seemed to be less important for frontal EP generation than theta activity. The most prominent amplitude enhancement was observed for theta-dependent visual EPs. This indicates a preferred response susceptibility of frontal lobes in theta frequency ranges. We suppose that the most prominent effect could be achieved by adapting the frequency range for rms evaluation to the range of maximal responsiveness of the evoked potential (prestimulus adaptive filtering). This approach would extend the adaptive filtering method introduced by Basar ¸ (1980) for EPs to the prestimulus EEG segment. Moreover, length of prestimulus epochs (1 s) is not necessarily optimal; shorter epochs may be more relevant for poststimulus epochs.

5.4 FREQUENCY CONTENT OF EROS FROM DIFFERENT LOCATIONS: MAJOR OPERATING RHYTHMS Basar ¸ et al. (1992b) indicated the impossibility of designing a purely sensory or purely cognitive paradigm in EP research. It is expected that various cognitive processes in addition to sensory processing come into play during a standard EP. Recent studies of Posner and Petersen (1990) emphasized the topographical characteristics of cognitive processing. A neuroanatomical study by Goldman-Rakic (1988) showed parallel-distributed networks in primate association cortex. Interpretations of results described in several chapters of this book suggest a distributed sensory–cognitive parallel processing system in the brain. The primary sensory processes and various associative or cognitive functions may be coactivated in different brain structures during the perception of a physical stimulus. This type of distributed parallel processing could be responsible for the differences of frequency contents of responses obtained in different locations.

5.4.1 MAJOR OPERATING RHYTHM (MOR)

OF

FRONTAL LOBE: THETA?

Rémond and Lesèvre (1957) reported on a predominance of theta rhythms in the frontal central region, whereas Mundy–Castle (1951) described more pronounced theta activity in temporal regions. Westphal et al. (1990) showed that the theta amplitude is highest over the anterior midline (Fz and Cz locations); this agrees with mapping findings (Walter et al., 1984; Mizuki et al., 1983). Miller (1991) noted that theta activity recorded from the hippocampus (HI) was difficult to find in humans because of human central electrophysiology. Evidence that the midline prefrontal region of the cortex can generate 5-Hz theta activity in certain cognitive states was reported by Mizuki et al. (1980). Spectral analysis of frontal EEGs showed that theta frequencies increased during motor or ˘ et al. (1991 a and verbal learning tasks (Lang et al., 1987; Westphal et al., 1990). Basar ¸ -Eroglu b) found that the significant theta response at the CA3 pyramidal layer could not be recorded analogously in the cortex due to volume conduction (see Chapter 4). The significant cognitive theta enhancements (Miller, 1991; Mizuki et al., 1980; Lang et al., 1987) in frontal and parietal recordings may have occurred via some Hebbian cooperation mechanism among the neuronal populations of the frontal cortex, the parietal cortex, and the HI. © 2004 by CRC Press, LLC

TF1663_book.fm Page 94 Friday, May 7, 2004 8:47 AM

The resonant theta response of the HI to auditory and visual stimuli was explained in detail in an earlier report of a component analysis of hippocampal-evoked potentials (Basar ¸ and Ungan, 1973). The concept of theta resonance was analyzed in extenso by Miller (1991), who described the cortico–hippocampal interaction as a basic resonance phenomenon in the theta frequency range. Based on anatomical and physiological evidence, Miller takes the view that theta modulated signals are likely to influence limbic and prefrontal areas, and also — directly or indirectly — other areas of (mainly association) cortex. Tentative interpretations of previous results led us to postulate the existence of selectively distributed alpha and theta response systems in the brain (compare Chapter 6). It was proposed that the theta component of the EPs and/or slower responses may reflect the responsiveness of various brain areas involved with global associative behavior (Schürmann et al., 1997). Theta increases during time prediction tasks were especially evident in the frontal and parietal recording sites (Demiralp and Basar ¸ , 1992). These results suggest an association between the theta frequency components of transient evoked responses, the association areas of the brain, and cognitive performance. Results from experiments with freely moving cats using a passive P300 paradigm led to the assumption that P300-like potentials have multiple cortical and subcortical generator sites including the reticular formation of the brain stem, HI, and auditory cortex. The P300 potential is most significant, stable, and has the largest amplitudes in the CA3 layer of the HI of the intact cat brain. The hippocampal P300 correlates with enhancement of the theta activities of field potentials and/or ˘ et al., 1991). a type of resonance phenomenon in the theta frequency range (Basar ¸ -Eroglu

5.4.2 MORS

OF

OCCIPITAL

AND

CENTRAL REGION (VERTEX)

Brandt and Jansen (1991) and Brandt et al. (1991) studied the relationship between levels of prestimulus alpha amplitude and the N1 to P2 peak-to-peak amplitude of the parieto-occipital visual EP obtained upon brief photo flashes with a method similar to that used by Basar ¸ (1980). Root–mean–square amplitude derived from power spectral measures in the alpha band of the 1-s prestimulus EEG were related to the peak-to-peak amplitudes of the N1 and P2 components of the visual-evoked potentials. Brandt reported a highly significant correlation between prestimulus alpha amplitude and N1 and P2 amplitudes and a general inverse relationship between visual EP enhancement and prestimulus alpha amplitude. Visual stimulation elicited a marked alpha response, i.e., an enhancement of alpha activity in the first 300 ms following stimulation. Brandt and Jansen (1991) also found an inverse relation between the amplitudes of prestimulus and poststimulus alpha activities. Their findings thus agree with our findings that: (1) an inverse relationship between prestimulus and poststimulus oscillations exists; and (2) the major operating rhythm of a brain structure or region controls or dominates the amplitudes of EPs. At the vertex, auditory and visual-evoked potentials can be described as compound alpha and theta responses. At the center of the head where neither alpha nor theta can be denoted as single MOR, both frequencies are present. We found an early equal dependence of auditory EPs and visual EPs on prestimulus alpha and theta components. These findings indicate that the effectiveness of the proposed algorithm depends on the MOR of the brain region investigated.

5.4.3 FUNCTIONAL SIGNIFICANCE

OF

EEG–EP INTERRELATIONS

Romani et al. (1988) have come to the conclusion that vigilance fluctuations (as measured by a vigilance-related index of delta/theta power) strongly affect stimulus processing. Sayers et al. (1974, 1979) stated that effective stimuli act by synchronizing the phases of spectral components of the spontaneous EEG activity already present.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 95 Friday, May 7, 2004 8:47 AM

Studies of the brain’s cognitive responses (Basar ¸ et al., 1989a and b) have shown repeatedly that EPs and ERPs reflect a transition to coherent stages of already existing information channels. No new frequencies appeared in the responses; we noted a kind of tuning of the existing resonance properties. Furthermore, the signals emanating from the brain generators upon sensory stimulation took into account the dynamic changes resulting from the preceding stimuli. The effect of incoming sensory information was modulated by physiological activities endogenous to the nervous system. This is a common aspect of most studies mentioned above. In this framework, one could speculate about the physiological function of the inverse relation between prestimulus and poststimulus EEG oscillations. Information processing may be more effective following stages of poor synchronization in defined frequency bands. The new concept may also lead to a better understanding of cognitive processing. The synchronization of the prestimulus EEG should be considered as an active component in evoked responses (Basar ¸ , 1980). If external stimuli are applied during phases of highly synchronized activity, further enhancement and frequency stabilization may not be elicited. Studies of the development of preparation rhythms (Basar ¸ and Stampfer, 1985; Basar ¸ et al., 1989a and b) have shown that it is possible to measure almost reproducible EEG patterns in subjects expecting defined sensory stimuli (targets). See also Chapter 3. Evidence indicates that pretarget activity interacts with EPs. Pretarget alpha effected a reduction or almost complete absence of the response, namely of the N1 component (Basar ¸ et al., 1989 a and b). Different cortical areas depict different MORs. It is evident that the MOR of the frontal cortex significantly influences the EPs; the theta activity in frontal lobes controls the frontal EPs directly. In other words, the behavioral theta states may play a major role in the genesis of frontal EPs.

5.5 BARRY: PREFERRED STATES IN BRAIN ACTIVITY The theoretical and experimental achievements of Barry et al. (2003) extended the work of Stampfer and Basar ¸ (1985) who suggested that with regular stimuli, phase adjustments occur in prestimulus EEGs. Information in Chapter 4 indicated that in a fixed ISI paradigm with alternating target and nontarget stimuli, the delta activity before target stimuli shows evidence of a stable phase reordering or preferred phase angle at the point of stimulation (see Figure 3.21 and Figure 3.22). The examples show evidence of maximum negativity at the time of stimulation. Alpha activity was also associated with increasing phase alignment at the point of stimulation, thus suggesting that it is also involved in a preferential occurrence of EEG negativity. Such observations led Basar ¸ and Stampfer to conclude that a regular pattern of stimulation can induce a preferred phase angle that appears to facilitate an optimal brain response to sensory input. Barry et al. (2003) started new investigations to analyze prestimulus activity and measure the occurrence of different phases at stimulus onset. Cortical negativity occurred preferentially in both delta and alpha (10 to 11 Hz) frequency bands, with opposite effect apparents in the 4- to 6-Hz theta range. Marked differences in the occurrence of the phases at stimulus onset as a function of frequency indicated substantial phase realignment, presumably associated with the fixed ISI paradigm. These data indicate the importance of moving beyond simplistic notions of the representative nature of the average ERP. Barry et al. (2003) suggest that in a fixed ISI paradigm, the EEG frequency components are dynamically adjusted in order to provide particular brain states at stimulus occurrence to facilitate processing of the stimulus.

5.5.1 CREATION

OF

PREFERRED BRAIN STATES

BY

APLR ALLIANCE

The importance of frontal lobes as a command center for memory organization has been described. The activation of theta activity during memory tasks and stimulations by complex signals has been discussed by several authors (Doppelmayer, 2000; Klimesch, 1999; Sakowitz et al., 2000). The © 2004 by CRC Press, LLC

TF1663_book.fm Page 96 Friday, May 7, 2004 8:47 AM

results cited in this chapter indicate that the major operating rhythm of the frontal lobe is the theta oscillation. However, the existence of a significant difference in major operating oscillations in occipital and frontal areas strongly supports the possibility that spontaneous, evoked, and induced theta and alpha rhythms have fundamentally different functional operations. During some functional states, MORs can change their functional roles. The nature of the experiment (task) can influence the weight of these functional components, often caused by the reciprocal activation of functions of the APLR alliance (see Chapter 3). This behavior of brain oscillations reflects also the plasticity in brain responsiveness.

5.6 CAUSALITY OF BRAIN RESPONSES ACCORDING TO CHANGES IN OSCILLATORY NETWORKS Do brain oscillatory responses change during phases of life? This question was examined by Basar ¸ et al. (1997c). Figure 5.6(a) illustrates instantaneous power spectra in occipital recordings of three subjects from each of the three age groups (3-year-old child, young adult, and middleaged adult). As noted in earlier studies (Eeg-Olofsson, 1971; Petersén and Eeg-Olofsson, 1971; Niedermeyer, 1993), EEGs in 3-year-old children usually do not show spontaneous 10-Hz activity. In fact, no 10-Hz activity was recorded in the EEG of the 3-year-old child in our study. In contrast to the results from children, the young adults had distinct and ample 10- to 12-Hz spontaneous and evoked responses in occipital recordings (Figure 5.6, middle). Results from experiments on middle-aged subjects showed reductions in 10-Hz activities in the occipital areas. Figure 5.6 (bottom) illustrates power spectra of B.R., a 55-year-old subject. The 10-Hz activity of this subject was apparent in comparison with activity in the 3-year-old child, but drastically reduced in comparison with the young adult. Figure 5.6(b) shows the amplitude-frequency characteristics (AFCs) and Figure 5.6(c) displays the averaged evoked potentials upon visual stimulation for the three subjects. Figure 5.6(d) illustrates the filtered average visual EP responses with band limits of 8 to 15 Hz. No alpha responses (defined as oscillatory brain activity in the 8- to 15-Hz frequency range within approximately 200 to 300 ms following external stimulation) were recorded in the visual EPs of children. In young adults, the group mean amplitude of the peak-topeak alpha responses in the averaged visual EPs was 4.5 mV (standard deviation [SD] = 1.9 mV). In the visual EPs of middle-aged adults, the peak-to-peak alpha response was 3.1 mV (SD = 1.01 mV).

5.7 ENTROPY AS CAUSAL FACTOR IN RESPONSES AND MECHANISMS OF SUPER-SYNERGY According to information theory, in dynamic changes of a system a relevant metric is its entropy (Shannon, 1948). Basar ¸ (1980) showed qualitatively that the prestimulus EEG is a causal factor for responsiveness in various frequency windows. Upon application of external or internal signals to the brain, a spontaneous EEG usually proceeds from a disordered state to an ordered state, i.e., upon application of a stimulus, the entropy of oscillations is decreased. This activity has been compared with the behaviors of paramagnetic and ferromagnetic substances and with lasers, and provides the core example for cooperative phenomena in the synergetics approach of Haken (1977). (See also Chapter 1.) Spectral entropy was introduced by Inouye et al. (1991 and 1993) for the analysis of EEG signals. Spectral entropy measures how concentrated or widespread the Fourier power spectrum of a signal can be. Low entropy values correspond to narrow-band (monofrequency) activities characterizing highly ordered (regularized) bioelectrical states. High entropy values reflect wideband (multifrequency) activities (Inouye et al., 1991). However, because of the low time resolution of © 2004 by CRC Press, LLC

TF1663_book.fm Page 97 Friday, May 7, 2004 8:47 AM

FIGURE 5.6 (a) Instantaneous power spectra of consecutive 2-s long EEG epochs. (b) Amplitude–frequency characteristics of visual-evoked potentials (VEPs). (c) Averaged VEPs. (d) VEPs filtered in the range of 8 to 15 Hz. Subjects were a 3-year-old child, young adult, and middle-aged adult. All recordings are from the left occipital site O1. Stimulus onset occured at 0 ms. (From Basar, ¸ E. et al. (1997c), Int. J. Psychophysiol., 26, 1.)

© 2004 by CRC Press, LLC

TF1663_book.fm Page 98 Friday, May 7, 2004 8:47 AM

the conventional Fourier transform, spectral entropy cannot assess fast dynamic changes of EEG states. The new wavelet entropy method was developed to overcome these limitations and quantify more precisely dynamic changes of EEGs from disordered to ordered states. A new method for analysis of wavelet entropy (WE) has been recently developed (Blanco et al., 1998; Quiroga et al., 2001; Rosso et al., 2001). The analysis of wavelet entropy is based on the time–frequency decomposition of the EEG by means of the wavelet transform (WT), providing for optimal time resolution for each frequency (Blanco et al., 1998). Rosso (2001) explored interactions of superimposed event-related oscillations (EROs). Wavelet entropy helped extract the superimposed EROs with optimal time resolution and quantified with fine time resolution the order and disorder states in short-duration signals such as ERPs. Sensory and cognitive stimulation elicit multiple EEG oscillations that may partly or fully temporally overlap the time axes. EEG responses to unimodal (auditory or visual) and bimodal (combined auditory and visual) stimuli in 15 young adults were analyzed by a new wavelet entropy method for short segments of ERPs. For each modality condition, a significant transient decrease of wavelet entropy was observed in poststimulus EEG epochs, thus indicating the transition to a highly ordered state in the ERP. Wavelet entropy minimum was always determined by a prominent dominance of theta (4 to 7 Hz) ERP components in comparison to other frequency bands. The event-related EEG oscillation transition to order was most pronounced and stable at anterior electrodes and following bimodal stimulations. Figure 5.7 shows relative energies of delta, theta, alpha, and beta frequency ranges for minimal entropies along the time axes. Note that for each modality, a strong predominance of the relative theta power is clearly observed at the time point of minimum entropy. In Chapter 7, we will propose that the entropy of the spontaneous activity and degree of order prior to stimulation entropy changes are important for understanding the shaping of percepts in the brain. The degree of order and disorder of oscillations, i.e., entropy in brain oscillatory states, provides one of the most important causalities of brain responsiveness for formation of percepts and thoughts.

5.8 GENETICS AS A CAUSAL FACTOR IN DELTA AND THETA RESPONSES AND BETA RHYTHMS The results from the laboratory of Begleiter and Porjesz´ in New York indicate that brain oscillations are linked with genetic factors. Some of the relevant results are described below.* Porjesz´ et al. (2002) recently showed that human brain oscillations are highly heritable. A common feature of beta oscillations (13 to 28 Hz) is the critical involvement of networks of inhibitory interneurons as pacemakers gated by gamma-aminobutyric acid type A (GABAA) action. Advances in molecular and statistical genetics permit examination of quantitative traits such as the beta frequency of the human EEG in conjunction with DNA markers. A significant linkage and linkage disequilibrium exist between beta frequency and a set of GABAA receptor genes. Uncovering the genes influencing brain oscillations provides a better understanding of neural function involved in information processing. In the study of Kamarajan et al. (in press) the go/no-go paradigm has been used to examine whether alcoholics have poor inhibitory control as compared to control subjects in terms of different oscillatory brain responses. The matching pursuits algorithm was used to decompose the eventrelated EEG into oscillations of different frequencies. It was found that alcoholics (n = 58) showed significant reductions in delta (1.0 to 3.0 Hz) and theta (3.5 to 7.0 Hz) power during no-go trials as compared to controls (n = 29). This reduction was prominent at the frontal region. The decreased delta and theta power associated with no-go processing perhaps suggests a deficient inhibitory ´ Begleiter, and others * After the manuscript of this book was sent to the publisher, a series of important reports by Porjesz, related to genetics and EEG oscillations were published by journals or are in press. They have been included here.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 99 Friday, May 7, 2004 8:47 AM

AEP

F4

F3

T3

C3

VEP

Cz

C4

F3

T4

P4

P3 80

Cz

C4

P3

P4

O1

O2

T4

80

% 40 0

C3

T3

F4

% 40 O1

O2

0

BEP

F3

C3

T3

F4

C3

C4

P3

T4

P4

80 % 40 0

O1

δΘαβ

O2

FIGURE 5.7 Relative energies of delta, theta, alpha, and beta frequency ranges during wavelet entropy minimum. For each modality, a strong predominance of relative theta power is clearly observed during the minimum. (Modified from Yordanova, J. et al. (2002), J. Neurosci. Methods, 117, 99.)

control and information processing mechanism. A neurocognitive model has been provided to explain the findings. The results suggest that the oscillatory correlates during cognitive processing may be endophenotypic markers in alcoholism. A study by Rangaswamy (in press) examines the differences in beta (12 to 28 Hz) band power in offspring of male alcoholics from heavily affected alcoholic families. The authors attempted to investigate whether the increase in beta power is a “state” or “trait” marker for alcoholism. This study also explores gender differences in the expression of this potential risk marker. Absolute beta power in three bands — beta 1 (12 to 16 Hz), beta 2 (16 to 20 Hz), and beta 3 (20 to 28 Hz) — in the closed-eye EEGs of 171 high risk (HR) subjects who were offspring of male alcoholics and 204 low risk (LR) subjects with no family histories of alcoholism were compared for each gender separately using a repeated measures analysis of variance design. Alcoholic and nonalcoholic

© 2004 by CRC Press, LLC

TF1663_book.fm Page 100 Friday, May 7, 2004 8:47 AM

subjects within the high risk group were compared using a repeated measures design as a follow up analysis. The study demonstrated increased beta power in the resting EEGs of the offspring of male alcoholics. Male HR subjects had higher beta 1 (12 to 16 Hz) power and female HR subjects had increased power in beta 2 (16 to 20 Hz) and beta 3 (20 to 28 Hz) as compared with low risk participants. Female HR subjects also showed significantly increased beta 2 and beta 3 power if they had two or more first-degree relatives who were alcoholic, compared with HR females who had only affected fathers. Risk characteristics are expressed differentially in males and females and may be indices of differential vulnerability to alcoholism. The results indicate that increased EEG beta power can be considered a likely marker of risk for developing alcoholism and may be used as a predictive endophenotype. According to Jones et al. (submitted), EROs offer an alternative theoretical and methodological approach to the analysis of event-related EEG responses. The P300 ERP is elicited through the superposition of the delta (1 to 3 Hz) and theta (3 to 7 Hz) band oscillatory responses. The cholinergic neurotransmitter system has a key function in modulating excitatory postsynaptic potentials caused by glutamate, and therefore influences P300 generation and the underlying oscillatory responses. The authors report significant linkage and linkage disequilibrium of target case frontal theta band, visual-evoked brain oscillations, and a single-nucleotide polymorphism (SNP) from the cholinergic muscarinic receptor gene (CHRM2) on Chromosome 7. They also demonstrate significant linkage disequilibrium between CHRM2 SNPs and target case parietal delta band visual-evoked oscillations (LD p<0.001). These findings were not observed for the equivalent nontarget case data, suggesting a role for the CHRM2 gene in higher cognitive processing in humans. The results presented in this section are candidates for opening new and important avenues to describe causal factors that shape human memory. We will discuss this further in Chapter 11.

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 101 Wednesday, May 12, 2004 8:05 PM

of Multiple 6 Correlation Oscillations with Integrative Functions and Memory 6.1 INTRODUCTION 6.1.1 AIM

OF

CHAPTER

Although we already described several experiments on memory-related oscillations in earlier chapters, this chapter aims to bring all the experiments into a common framework integrating physiology and cognition. This chapter will also review various functional states by systematically categorizing them in the frequency windows of the electroencephalogram (EEG), thus preparing readers for new theoretical information in Chapter 7. 6.1.1.1 Emphasis on Multiple Oscillations in Brain Research Freeman (1999) stated that neuroscience was ripe for a change. In the past decade, increasing numbers of reports and reviews on electrical oscillations in the brain appeared in neuroscience literature. Only a few cited the importance of multiple oscillations and their coherences ranging in a large frequency spectrum. To fill the gap, this chapter has the goal of covering oscillations related to sensory and cognitive processing in all frequency channels by humans and animals and reporting more advanced results than those explained in the 1992 volume edited by Basar ¸ and Bullock. We also include a short historical survey on brain oscillations, experiments on animals, evolution, parallel processing, the very high frequency window between 100 and 1000 Hz, and long distance coherent oscillations. As a result, this chapter provides a perspective different from the important review by Klimesch (1999). The functional importance of distributed multiple oscillations in the brain was for the first time emphasized in a series of reports in the 1970s (Basar ¸ , 1992; Basar ¸ et al., 1975a, b, and c; Basar ¸ and Ungan, 1973). Further, the relevance of the superposition principle and long distance synchronisation in the brain was demonstrated in the 1970s (Basar ¸ , 1980; Basar ¸ et al., 1979a and b). The concerted activity of alpha, theta, delta, and beta oscillations was measured in structures such as the reticular formation (RF), hippocampus (HI), thalamus, and sensory cortices. The literature presented the fundamental frameworks for understanding signal processing in the brain. 6.1.1.2 Role of Oscillations in Memory Processing The manifestations of memory processes as oscillatory activities attracted important attention in recent years. Klimesch’s group showed the possibility of differentiating the roles of alpha and theta oscillatory activities during memory tasks. Their results support the hypothesis that event-related potentials (ERPs) can be understood and described in terms of superpositions of several eventrelated oscillations (EROs) recorded in various structures (Klimesch et al., 2000; Doppelmayer et al., 2000). Their experiments involved memory tasks differentiating oscillatory responses of good and bad memory performers and demonstrated functional differences in subalpha and subtheta frequency bands.

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 102 Wednesday, May 12, 2004 8:05 PM

According to Fuster’s (1997) view, memory reflects a distributed property of cortical systems. An important part of higher nervous function (as perception, recognition, language, planning, problem solving, and decision making) is interwoven with memory. Memory is a property of the neurobiological systems it serves and is inseparable from their other functions. By surveying the data presented in this chapter, it can be hypothesized that the selectively distributed oscillatory systems (or networks) may provide a general communication framework and can be useful for functional mapping of the brain (Mesulam 1990 and 1994). New experimental designs (Burgess and Gruzelier, 2000; Egner and Gruzelier, 2001; Haenschel et al., 2000; Klimesch, 1999; Klimesch, Schimke, and Schwaiger, 1994) are extremely important, for they add new understanding to the general framework of function-related oscillations. 6.1.1.3 Steps for New Synthesis and Binding Problem The new trend toward examining brain oscillations implies that single neurons, and neuron assemblies, spikes of single neurons and oscillatory activity of neurons and neural assemblies, and movements and cognitive and memory processes are interwoven in integrative brain functions. These concepts were not included in Sherrington´s description of integrative brain activity, and neuroscience needs a new framework or theories and possibly new rules for analyzing integrative brain function. New propositions may open new experimental avenues as was the case with the EEG brain dynamics framework published over 20 years ago (Basar ¸ , 1980). This chapter discusses a few illustrative experiments chosen to emphasize the large variety of frequency responses ranging from delta to gamma in invertebrate ganglia and human cognitive responses. Publications covering functional relevance of multiple oscillations and the importance of selective distributions, delay, and prolongation of oscillations are rare (Bressler and Kelso, 2001). A fundamental unsolved problem in neuroscience is the manner in which the vast array of parallel processes occurring in the brain at any given time and diverse neural activities are bound together or integrated (Haig et al., 2000; Chapter 7, this volume). For example, a visual image of an object contains a collection of features that must be identified and segregated from features that constitute other objects. This chapter contains three essential discussions: 1. A short chronological survey level related to brain oscillations 2. Comparative studies of various sensory cognitive oscillatory neural populations in human and animal recordings; functional correlates of oscillations and multiple oscillations 3. On a general level, superposition and distribution of selective oscillations and selectively distributed coherences.

6.2 SURVEY OF EEG OSCILLATIONS 6.2.1 ALPHA ACTIVITY 6.2.1.1 Survey by Andersen and Andersson (1968) and Basar ¸ (1999) A particular feature of thalamic relay nuclei is their ability to convert a single afferent volley to a series of rhythmic discharges along thalamo-cortical fibers. Adrian (1941) discovered that a single tactile stimulus elicited a series of waves in the thalamus and he designated the phenomena thalamic after-discharges. Similar rhythmic activity was found by Bremer and Bonnet (1950) in the medial geniculate nucleus in response to a click. All these authors noted that the frequency of the evoked activity was around 10 events per second, similar to the frequency of spontaneous rhythmic cortical waves. Adrian (1941) maintained that the after-discharges consisted of bursts of spikes separated by slow waves. A peripheral stimulus elicited a series of three to seven cycles. By recording from the © 2004 by CRC Press, LLC

TF1663_C06.fm Page 103 Wednesday, May 12, 2004 8:05 PM

white matter below the cortex, Adrian showed that the rhythmic discharge occurred in the thalamocortical fibers, indicating a thalamic origin of the after-discharges. Due to this rhythmic discharge in response to a single afferent volley, a series of waves at a frequency of about 10 per second were initiated in the cortex (Bartley and Bishop, 1933; Bishop, 1933; Bishop et al., 1953; Jarcho, 1949). Adrian (1941) reported that rhythmic activity at a frequency of 10 per second following a single afferent volley could be recorded within or at the dorsal surface of the thalamus even if the appropriate cortical area was removed. In other words, thalamic nuclei contain a mechanism for the transfer of a single volley to a rhythmic 10-per-second sequence in the absence of a cortical area to which the thalamo-cortical fibers can project. Chang (1950) advanced the hypothesis that a cortico-thalamic reverberating circuit should be the basis for the evoked rhythmic activity. The arguments for this explanation were the presence of a similar rhythmic activity in the thalamus and cortex and the difficulty of recording thalamic rhythmic activity after removal of the appropriate cortical projection area. However, this theory is contradicted by the early reports of Adrian (1941), Bremer and Bonnet (1950), Galambos et al. (1952), and also by more recent observations of Adrian (1951) who critically tested the cortico-thalamic reverberating hypothesis. 6.2.1.2 Toward a Renaissance of Alphas The volume edited by Basar ¸ et al. (1997) describes state-of-the-art studies of alpha activity of the brain. The papers included in that volume describe the physiological bases of 10-Hz activities and their functional correlates, with emphasis on sensory, cognitive, and motor states that accompany the 10-Hz oscillations. This endeavor, aimed at establishing a new set of functional implications of alpha activity, was also intended to lead to a “renaissance of alpha studies” 70 years after the first discovery by Hans Berger. The 1997 volume proposed a new nomenclature. Most phenomena in the alpha band were described as alphas or 10-Hz oscillations of the brain (see also Klimesch, 1999). Such terminology draws attention to the multiple phenomena in the alpha band that were regarded formerly as a single rhythm and thus a single phenomenon. In the more recent approach, the mu and tau rhythms are also regarded as components within the associated ensemble of phenomena classes as alphas. The 10-Hz oscillations are, according to the authors, not unitary phenomena either; by their multiform nature, they represent basic functions of the brain.

6.2.2 EARLIER EXPERIMENTS

ON INDUCED OR

EVOKED THETA OSCILLATIONS

Ross Adey`s group (1960) performed pioneering work on theta rhythms of the limbic system of the cat brain during conditioning. These authors were the first to study spectral and coherence functions by performing experiments demonstrating that rhythmic field potentials of the cat brain are related to behavior (Elazar and Adey, 1967; Miller, 1991). The use of the coherence function in comparing EEG activities of various nuclei of the brain was valuable for refuting the view that the EEG was an epiphenomenon (Adey, 1989). The discovery of induced theta rhythm and task-relevant coherence in the limbic system of the cat brain was a milestone in EEG research. Cat hippocampal activity exhibited a transition from irregular activity to coherent induced rhythms. Such results encouraged Basar ¸ and Özesmi (1972) and Basar ¸ and Ungan (1973) to choose the HI as a model for a possible resonance theory. The experiments performed by these and other groups showed that such an explanatory model was feasible (Miller, 1991). Orienting is a coordinated response that appears to indicate alertness, arousal, or readiness to process information. Orienting is, of course, closely linked to attentive states and learning. An extended review on functionally related theta activity was written by O'Keefe and Nadel (1978). Miller (1991) stated that hippocampal theta activity in cats occurred in association with locomotion and other body movements, in a manner similar to movements seen in rats and other small © 2004 by CRC Press, LLC

TF1663_C06.fm Page 104 Wednesday, May 12, 2004 8:05 PM

mammals. However, these correlations include a great many exceptions, for example, in some situations theta activity can occur during immobility or movement can occur without theta activity. Many of these exceptions occur in the course of learning tasks. One suggestion is that theta activity occurs when performance of a learned task is improving most rapidly and declines as a task becomes familiar. Apart from these results, a great deal of evidence indicates that spontaneous and conditioned orienting is accompanied by low-frequency theta activity in cats. Petsche et al. (1962) followed up early evidence from lesion experiments that showed a septal involvement in theta rhythm by recording from units in the septa of curarized rabbits. Theta activity was elicited by sensory or electrical stimulation of the RF. A proportion of septal units showed regular rhythmic bursts of impulses, in phase with hippocampal theta rhythms. These findings have been confirmed many times. Vinogradova and Zolotukhina (1972) found that only a third of neurons in medial and lateral septa showed theta bursts in alert rabbits. Apostol and Creutzfeldt (1974) recorded septal and hippocampal EEGs and spike activities of individual septal units simultaneously during spontaneous rhythmic activity or rhythms evoked by sensory stimuli in curarized rabbits. Brazier (1968) made hippocampus recordings in 30 patients (28 had temporal lobe epilepsy; 2 had no epilepsy) prior to surgery. The hippocampal EEGs showed peak power in the 2- to 4-Hz range. The peak of coherence with EEGs from other sites (e.g., parahippocampal gyrus) was in the 4- to 8-Hz range. Lieb et al. (1974) recorded from the hippocampus, amygdale, and parahippocampal gyrus. A small peak was seen at 8 to 10 Hz,; it was variable among patients and broad peaks were seen at slower frequencies. Halgren et al. (1986) made behavioral observations in a single case study. The hippocampus was synchronized at 5 to 6 Hz during quiet resting, when a patient tensed all his muscles, or made simple alternating movements. Increased synchrony was also seen during testing of patients on a series of word meanings or tapping sequences. Desynchronization occurred during speech or rapid breathing, tying a bow, imitating movements of the experimenter, or providing a verbal description of a word.

6.2.3 GAMMA FREQUENCY RANGE The empirical background of the gamma band dates back to Adrian (1942) who reported that the application of odorous substances to the olfactory mucosa of the hedgehog induced a train of sinusoidal oscillations in the 30- to 60-Hz range. Since 1942, studies of 40-Hz rhythmicity have ˘ and colleagues (1996a and b). The induced passed through four phases according to Basar ¸ -Eroglu character of the gamma band was studied in the first phase. The second phase (1960 through 1980) was characterized by the works of Freeman (1975), Basar ¸ et al. (1972, 1980, and 1979), and Sheer (1976) in which a variety of functions were ascribed to gamma rhythmicity. The third phase started with the work of Galambos, Makeig, and Talmachoff (1981). Their work led to investigations of sensory and cognitive correlates of gamma oscillation, primarily in humans. The fourth phase was initiated by Gray and Singer (1987), whose work led to investigations of the 40-Hz rhythmicity at the cellular level. The current fifth phase is marked by the heterogeneity of approaches and techniques aimed at solving the gamma puzzle (see also Karakas¸ and Basar ¸ , 1998).

6.3 SELECTIVELY DISTRIBUTED OSCILLATORY SYSTEMS: DISTRIBUTED MULTIPLE OSCILLATIONS 6.3.1 CONCEPT, DEFINITIONS,

AND

METHODS

The functional significance of oscillatory neural activity began to emerge from the analysis of responses to well-defined events (EROs phase- or time-locked to a sensory or cognitive event). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 105 Wednesday, May 12, 2004 8:05 PM

Among other approaches, it was possible to investigate such oscillations by frequency domain analysis of ERPs based on the following hypothesis (Basar ¸ , 1980 and 1998). An EEG consists of the activity of an ensemble of generators producing rhythmic activity in several frequency ranges. These generators or oscillators are active usually in a random way. However, by application of sensory stimulation, they are coupled and act coherently. This synchronization and enhancement of EEG activity gives rise to evoked or induced rhythms. Evoked potentials representing ensembles of neural population responses were considered results of a transition from a disordered state to an ordered state. A compound ERP manifests a superposition of evoked oscillations in EEG frequencies ranging from delta to gamma [natural frequencies such as alpha (8 to 13 Hz), theta (3.5 to 7 Hz), delta (0.5 to 3.5 Hz), and gamma (30 to 70 Hz)].

Time-locked responses of a specific frequency after stimulation can be identified by computing the amplitude frequency characteristics (AFCs) of the averaged ERPs (Basar ¸ et al., 1980; Röschke et al., 1995; Yordanova and Kolev, 1997). The AFCs describe the brain system's transfer properties, e.g., excitability and susceptibility, by revealing resonant and salient frequencies. An AFC therefore does not simply represent the spectral power density characterizing the transient signal in the frequency domain. It represents the predicted behavior of the brain if sinusoidally modulated input signals of defined frequencies were applied as stimulation. As reflecting the amplification in a given frequency channel, an AFC is expressed in relative units. Hence, the presence of a peak in the AFC reveals resonant frequencies interpreted as the most preferred oscillations of the system during response to a stimulus (see glossary in Appendix 1 at the end of this book). In order to calculate AFCs, ERPs were first averaged and then transformed to the frequency domain by means of a onesided Fourier transform (Laplace transform (Solodovnikov, 1960; Basar ¸ , 1980), as shown in Figure 6.1). Figure 6.2 illustrates an AFC and demonstrates that auditory stimuli produce prominent resonant responses in the theta, alpha, and gamma (40 Hz) frequency bands in the cat HI (Basar ¸ , 1980). Amplitude frequency characteristics serve also to define filter limits for response-adaptive digital filtering of averaged ERPs. The filtered curves obtained in this way show the time course of oscillatory activity in a certain frequency range (Figure 6.3). Gönder and Basar ¸ (1978) performed a comparative study of power spectral peaks both in spontaneous and single-evoked activities in the cat brain. They used a method based on comparison of single EEG–EP epochs in histogram distribution. The results showed a transition from disordered to ordered states in all EEG frequencies. even covering the gamma and highest frequency window up to 1000 Hz (Figure 6.4). These 25-year-old findings have acquired support by new findings in magnetoencephalography (MEG). More recently, a technique called wavelet analysis has been applied to ERPs. Wavelet analysis confirms results obtained from AFCs and digital filtering. It can also be used for signal retrieval and selection among a large number of sweeps recorded in a given physiological or psychological experiment (Figure 6.5; Demiralp et al., 1999; Basar ¸ , 1999; Basar ¸ et al., 1999). The AFC method has an important advantage over wavelet analysis. It allows a global view of all frequency responses together; in wavelet analysis, the investigator arbitrarily defines windows. Wavelet analysis can produce misinterpretations by orienting the search to special windows (for example, in most studies, only the gamma band is selected). As will become clear below, the combination of these methods yielded results leading to the conclusion that alpha, theta, delta, and gamma responses of the brain are related to psychophysiological functions and constitute “real” signals (Basar ¸ , 1998 and 1999; Basar ¸ et al., 2001). We intend to show that these oscillations have multiple functions and may act as universal operators or codes of brain activity. In addition to frequencies and sites of oscillations, several other parameters are dependent on specific functions, namely enhancement, time locking, phase locking, delay, and duration of oscillations (for a review of methods to assess these parameters, see Kolev and Yordanova, 1997). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 106 Wednesday, May 12, 2004 8:05 PM

FFT

Power spectrum of pre- and poststimulus EEG

n sweeps (n = 100) Digital filtering

Frequency components of single sweeps enhancement factors etc.

Selective averaging

FFT

AFC of averaged EP/ERP

Digital filtering

Frequency components of average EP/ ERP

Average EP

Averaging across subjects (in time domain)

Averaging across subjects (in frequency domain) FFT

AFC of grand average

Grand average

Mean value of AFCs (spectrum average) Digital filtering

Frequency components of grand average

FIGURE 6.1 Stepwise application of system theoretical methods. For the sake of simplicity, evaluation of wavelet decomposition and coherence functions are not included. (Modified from Schürmann, M. et al. (1997a), Int. J. Psychophysiol., 26, 149.)

6.3.2 OSCILLATORY RESPONSES

IN INVERTEBRATE

GANGLIA

Since oscillatory properties were measured in nonmammalian brains and ganglia, we first reviewed some of the findings in the last decade to focus on the invariant character of oscillations throughout evolution (Bullock and Basar ¸ , 1988). As shown in Figure 6.6, an early phase-locked 40-Hz response was recorded in the visceral ganglia of Helix pomatia using electrical stimulation (Schütt, Basar ¸ , and Bullock, 1992; Schütt and Basar ¸ , 1992). Light-induced gamma responses have been observed also in arthropods (Kirschfeld, 1992). Alpha activity is not unique to mammals. Spontaneous and electrically evoked 10-Hz oscillations were measured also in isolated ganglia of Helix pomatia and

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 107 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.2 Averaged EP (A) recorded from the right dorsal hippocampus of the cat after auditory stimulation in the form of a step function (3-s long tone burst at intensity of 2000 Hz), negativity upward. (B) Amplitude frequency characteristics computed from the transient response (A). Double logarithmic presentation along the abscissa–log (frequency) and along the ordinate–relative amplitude in decibels. (Modified from Basar, ¸ E. (1980), in EEG Brain Dynamics: Relation between EEG and Brain-Evoked Potentials, Elsevier, Amsterdam.)

Aplysia as shown in Figure 6.7 (Basar ¸ , 1998b; Bullock and Basar ¸ , 1988; Schütt and Basar ¸ , 1992 and 1999).

6.3.3 GAMMA OSCILLATIONS

IN

SENSORY, COGNITIVE,

AND

MOTOR PROCESSES

Since the gamma band has attracted special attention for the past two decades, we start with discussing this window. Basar ¸ (1998) performed a lengthy general chronological survey related to oscillatory responses at cellular level. Other relevant work in this field will also be discussed in this chapter. The most prominent examples of event-related gamma oscillations are oscillatory responses in the frequency range of 40 to 60 Hz that occur in synchrony within a functional column in the cat visual cortex (Gray and Singer, 1989; Eckhorn et al., 1988). This dynamic has been suggested as a possible mechanism of linking in the visual cortex since it is related to the binding problem. This theory, however, does not fully explain the ubiquity of gamma rhythms (Desmedt and Tomberg, 1994; Schürmann et al., 1997). In this respect, it may be helpful to consider further studies of gamma oscillations subsequent to the work of Adrian (1942). While the interpretations are heterogeneous, the empirical findings may be roughly classified into sensory (or obligatory) versus cognitive gamma responses. Some examples of sensory functions are presented below. Auditory and visual gamma EEG responses are selectively distributed in different cortical and subcortical structures. They are phase-locked stable components of evoked potentials (EPs) in cortices, hippocampi, brain stems, and cerebellums of cats that occur 100 ms after sensory stimulation, with a second window of approximately 300 ms latency (Figure 6.8; Basar ¸ , 1980; Basar ¸ , 1998b; Basar ¸ et al., 1997b). The earliest studies of gamma auditory response in the cat hippocampus were recorded by Basar ¸ and Özesmi (1972) and Basar ¸ and Ungan (1973) and later confirmed by Leung et al. (1992). A review by Keil, Gruber, and Müller (2001) presents evidence © 2004 by CRC Press, LLC

TF1663_C06.fm Page 108 Wednesday, May 12, 2004 8:05 PM

A 0– 8 Hz

B 8 –18 Hz

C 18 –30 Hz

D 30 – 55 Hz

l lll ll lV

Vll

V

Xl E 55 – ∞ Hz

Xlll

Vl IX

XlV

Xll

Vlll X 0

100

200

300

400

500

600

Time (ms) T1

T2

T3

T4

FIGURE 6.3 Filtering of a selectively averaged hippocampal EP with different stop- and pass-bands. Solid lines represent filtered averaged EPs obtained with application of pass-band filters. Dashed lines represent filtered averaged EPs obtained with application of stop-band filters. The band limits (at right of averaged EPs) of the applied filters were chosen according to the TRFC method. The original selectively averaged EP is shown for comparison with filtered averaged EPs. Time sections T1 through T4 are shown at the bottom. Note the ample gamma responses with comparable values to alpha response filters. (Modified from Basar, ¸ E. (1980), in EEG Brain Dynamics: Relation between EEG and Brain-Evoked Potentials, Elsevier, Amsterdam.) © 2004 by CRC Press, LLC

TF1663_C06.fm Page 109 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.4 Histograms showing frequency distribution. (A) prestimulus power spectra. (B) poststimulus instantaneous frequency characteristics of the reticular formation. Bottom: each spectral peak (or amplitude maximum) is represented by an approximate bandwidth (horizontal line segment) and center frequency. The histograms were prepared by plotting the number of center frequencies falling into each set of 20-Hz slots versus frequency. Seventy-one power density-instantaneous frequency characteristic pairs were used to obtain the histograms. (Modified from Gönder, A. and Basar, ¸ E. (1978), Biol. Cybernetics, 31, 193.)

supporting the assumption that gamma oscillations are involved in several aspects of visual processing. Sensory processes — A phase-locked gamma oscillation is also a component of the human auditory and visual response as Figure 6.9 shows. A new strategy by application of six cognitive paradigms showed that the 40-Hz response in the 100 ms after stimulation has a sensory origin, and is independent of cognitive tasks (Figure 6.10; Karakas¸ and Basar ¸ , 1998). Auditory MEG gamma response — This is similar to human EEG response with a close relationship to the middle latency auditory-evoked response (Pantev et al., 1991). Cognitive processes — Several investigations dealt with cognitive processes related to gamma responses, some based on measuring the P300 wave. This positive deflection typically occurs in © 2004 by CRC Press, LLC

TF1663_C06.fm Page 110 Wednesday, May 12, 2004 8:05 PM

Auditory Stimulation GEA

HI

Visual Stimulation OC

µV

0.0 100.0

–12.5 0.0

–25.0 0.0 25.0

–12.5 0.0

–25.0 0.0 25.0

12.5

alpha

–12.5 0.0

–25.0 0.0 25.0

12.5

theta

–25.0 0.0 25.0

–12.5 0.0 12.5

delta

–12.5 0.0

–25.0 0.0 25.0

12.5

–1000

OC

–100.0

12.5

beta

HI

µV

–25.0 0.0 25.0

gamma

GEA

0 1000 –1000 0 1000 –1000 0 1000 ms

–1000

0 1000 –1000 0 1000 –1000 0 1000 ms

FIGURE 6.5 Results of wavelet decomposition in a typical animal Left: auditory stimulation. Right: visual stimulation. Each column refers to an electrode site (GEA = auditory cortex; OC = visual cortex; HI = hippocampus). The top row shows the wide-band filtered curve. The other rows show the frequency components [gamma (32 to 64 Hz), beta (16 to 32 Hz), alpha (8 to 16 Hz), theta (4 to 8 Hz), and delta (0.5 to 4 Hz)]. (Modified from Basar ¸ et al. (1999c), Brain Language, 66, 146–183.)

human ERPs in response to oddball (OB) stimuli or omitted stimuli interspersed as targets into a series of standard stimuli. A P300–40-Hz component has been recorded in the cat HI, RF, and cortex (with omitted auditory stimuli as targets). This response occurred approximately 300 ms after stimulation, being ˘ and Basar superimposed with a slow wave of 4 Hz (Basar ¸ -Eroglu ¸ , 1991). See Figure 4.12 through Figure 4.15. Preliminary data indicate similar P300–40-Hz responses to OB stimuli in humans ˘ et al., 1992). However, a suppression of 40-Hz activity after target stimuli has also (Basar ¸ -Eroglu been reported (Fell, Hinrichs, and Röschke, 1997). In a recent study, gamma band activity in an auditory OB paradigm was analyzed with the wavelet transform. A late oscillatory peaking at 37 Hz with a latency around 360 ms was observed only for target stimuli (Gurtubay, et al. 2001). The superimposed theta–gamma complex pattern was analyzed in modeling studies (Baird, 1999; Kiss et al., 2001). A study of the human limbic system also confirmed the synchronization of gamma rhythms during cognitive tasks (Fell et al., 1997). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 111 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.6 Averaged evoked potentials. The 40-Hz activity increased after dopamine administration. Top: controls. Bottom: dopamine (10-2 M). Left: wide-band filtered from 1 to 250 Hz. Right: pass-band filtered from 30 to 70 Hz. (Modified from Schütt, A. et al. (1992a), Comp. Biochem. Physiol., 102C, 169–176.)

FIGURE 6.7 Helix pomatia. Field potential fluctuations in different ganglia; samples of 2 s each. Top: cerebral ganglion. Bottom: visceral ganglion. Thin line: wide-band component (1 to 50 Hz). Thick line: narrow-band component (8 to 15 Hz). The 8- to 15-Hz fluctuation is not merely a filtered white noise because it is also clearly visible in the wide-band record. (Modified from Schütt, A. et al. (1992a), Comp. Biochem. Physiol., 102C, 169–176.) © 2004 by CRC Press, LLC

TF1663_C06.fm Page 112 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.8 Responses to auditory stimuli in cats filtered in gamma band (30 to 70 Hz) and filtered auditory responses of cats in GEA, RF, HI, and CE. The gamma-band filtered grand averages of responses are shown in the bottom row. Note the large gamma responses also in the cerebellum. (Modified from Basar, ¸ E. et al. (1995), IEEE Eng. Med. Biol., 14, 400.)

During visual perception of reversible or ambiguous figures, a significant increase (almost 50%) ˘ et al., 1996). The in human frontal gamma EEG activity was recorded (Figure 6.11; Basar ¸ -Eroglu spatio-temporal magnetic field pattern of gamma band activity has been interpreted as a coherent rostrocaudal sweep (Llinas and Ribary, 1992). Sakowitz et al. (2001) reported significant gamma response amplitude increases in distributed areas of the brain and a 100% frontal theta enhancement by bimodal stimulation compared to unimodal stimulation. Tallon-Baudry (1998) reported that © 2004 by CRC Press, LLC

TF1663_C06.fm Page 113 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.9 (A) Ten randomly selected single EEG–EP trials (digital filters of 30 to 50 Hz). Asterisk represents average of these trials. Stimulation was applied at 0 ms. (B) Ten single EEG–EP trials (digital filters of 30 to 50 Hz) selected for high enhancement, i.e., high amplitude increase after stimulus (in comparison with prestimulus EEG amplitude). Double asterisk represents average of these trials. (C) Selectively averaged evoked potentials (EPs); averages of 40 single trials each, with different selections of trials (human vertex recordings). (a) EP averaged from randomly selected single trials (conventional averaged EP); (b) same EP filtered at 30 to 50 Hz; (c) EP averaged from single trials selected for specific criteria (marked amplitude enhancement in the 40-Hz range after stimulation); (d) same EP filtered with band limits of 30 to 50 Hz; (e) EP averaged from single trial with low amplitude enhancement; (f) same EP filtered with band limits of 30 to 50 Hz; (g) same as (e); average of low enhancement trials; (h) result of applying a 30- to 50-Hz stop-band filter (which theoretically rejects a 40-Hz response) to the conventional averaged EP of (a). Note the similarity of (h) obtained with stop-band filtering and (g) obtained with low enhancement trials. (Modified from Basar, ¸ E. et al. (1987), Int. J. Neurosci., 33, 103.)

retrieval of visual experience from short-term memory is associated with 40-Hz activity (Pulvermüller et al., 1996). 6.3.3.1 Multiple Functions in Gamma Band The wide spectrum of experimental data presented is in accordance with a hypothetical selectively distributed parallel processing gamma system with multiple functions. Instead of serving as highly specific correlates of a single process, gamma oscillations may be important building blocks of electrical activity of the brain. Because they are related to multiple functions, they may (1) occur in different and distant structures, (2) act in parallel, and (3) show phase locking, time locking, or weak time locking. Simple electrical stimulation of isolated invertebrate ganglia evokes gamma oscillations in the absence of perceptual binding or higher cognitive processes. In conclusion, gamma oscillations may represent a universal code of central nervous system (CNS) communication (Basar ¸ , 1998 and 1999). This view may serve to overcome controversies arising from earlier reports. ˘ See also Basar ¸ et al., 2001 and Basar ¸ -Eroglu, 1996.

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 114 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.10 Topography of grand averages of filtered (28 to 46 Hz) EEG–ERPs from target stimuli of the Oddball–Easy (OB–EZ) paradigm (From Karakas¸, S. and Basar, ¸ E. (1998), Int. J. Psychophysiol., 31, 13.)

6.3.3.2 Important Causality Factor for Human Gamma Response The goal of a recent study by Karakas¸ et al. (2003) was to investigate whether early time-locked sensory gamma band response is correlated with scores derived from neuropsychological tests. Neuroelectric responses were obtained under the active auditory OB paradigm. Of 67 reportedly healthy adults, 35 displayed time-locked early gamma [G(+)] and 24 did not [G(–)]. Of 52 neuropsychological test scores, G(+) and G(–) groups differed on the basis of Wisconsin Card Sorting Test, Wechsler Memory Scale (Revised), and Serial Digit Learning Test scores. Results of logistic regression analysis were statistically reliable (overall success rate of prediction = 93.33%). The results showed that early gamma response can be classified on the basis of neuropsychological test performance and is thus associated with higher cognitive functions, supporting the view that the brain integrates bottom-up with top-down processing.

6.3.4 ALPHA OSCILLATIONS

IN

PERCEPTION

AND

COGNITION: THE ALPHAS

Alpha oscillations in functional EEG gained more importance in the past decade (Basar ¸ et al., 1997). Observations at the cellular level are noteworthy: Evoked oscillations in the 8- to 10-Hz frequency range in visual cortex neurons upon visual stimulation suggest a relation to scalp-recordable alpha © 2004 by CRC Press, LLC

TF1663_C06.fm Page 115 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.11 Single EEG epochs digitally filtered in the gamma band (30 to 50 Hz) at F4. Left: single epochs of control EEG. Right: single EEG epochs during naive observation. The epochs included both reversal and nonreversal states. The subjects were required to observe the Stroscopic Alternative Motion (SAM) pattern without task. All subjects reported late in the experiment that they detected the perceptual switching of the ˘ pattern. (From Basar ¸ -Eroglu, C. et al. (1996b), Int. J. Neurosci., 85, 131.)

responses (Dinse et al., 1997; Silva et al., 1991). Thalamo-cortical networks oscillate in the alpha range (Steriade et al., 1992). Alpha responses from different structures of the cat brain are presented in Figure 6.5. The sum of these observations permits a tentative interpretation of alpha as a functional and communicative signal with multiple functions. This interpretation of 10-Hz oscillations (at the cellular level or in populations) may be comparable to the putative universal role of gamma responses in brain signaling. In the auditory and visual pathways in cats, adequate stimuli elicit alpha responses (damped 10-Hz oscillations of approximately 300 ms) that are visible without filtering (Basar ¸ , 1980, 1998, and 1999). For confirmation by wavelet analysis, see Basar ¸ , 1998a and Basar ¸ et al., 2001. Thalamocortical circuits are not unique in generating alpha responses. Hippocampal and reticular 10-Hz responses are relatively modality–independent, hinting at possible supramodal functions. The interpretation of alpha response as an idling rhythm rests on observations such as blocking of spontaneous occipital alpha oscillations upon opening of the eyes or blocking of central mu rhythm upon movement onset (Kuhlman, 1978), also known as event-related desynchronization. © 2004 by CRC Press, LLC

TF1663_C06.fm Page 116 Wednesday, May 12, 2004 8:05 PM

(Pfurtscheller et al., 1997). A reverse effect (increase of mu rhythm during visual information processing or event-related synchronization) has also been reported (Koshino and Niedermeyer, 1975; Pfurtscheller et al., 1997). However, coexisting with these well-known phenomena and in relationship with Adrian’s evoked alpha concept (1942), several forms of functional alpha have been observed during sensory and cognitive processes (Basar ¸ , 1998 and 1999; Basar ¸ et al., 1997a, b, and c). Examples of such function-related alpha responses are discussed in the next section. 6.3.4.1 Sensory Components Example: alpha response in cross-modality measurements in the cat brain — As an example of the function-related alpha responses mentioned in the previous paragraph, we will deal with topographic differences of frequency components. In particular, we will summarize results of measurements from auditory and visual areas. The conditions were either adequate stimulation (auditory cortex recording of auditory EP; visual cortex recording of visual EP) or inadequate stimulation (visual cortex recording of auditory EP and vice versa). Such experiments are also known as cross-modality measurements (Basar ¸ , 1998 and 1999; Hartline, 1987). Panel A in Figure 6.12 shows single-trial EPs filtered in the 8- to 15-Hz range. The left column shows auditory stimulation with visual cortex recordings; the right column shows visual stimulation with visual cortex recordings, i.e., inadequate versus adequate stimulation. Responses to visual (adequate) stimulation show amplitude increase and time and phase locking. A distinct response is also seen in the filtered averaged EP in shown in panel B. The unfiltered averaged EP also shows an alpha-like waveform. In contrast, responses to auditory (inadequate) stimulation show no amplitude increase or phase locking, nor can we see an alpha response in the filtered average. The type of response in the unfiltered EP in panel C is not an alpha response. Figure 6.13 depicts another example of time locking and amplitude increase in single-trial responses to adequate stimulation. Single sweeps filtered in the 8- to 15-Hz range (visual stimuli, visual cortex recordings) are superimposed. The superimposed single sweeps (top) are similar in waveform to the wide-band filtered curves (bottom). It is thus not only by filtering that the alpha response can be illustrated in these sweeps. Alpha responses are even visible in the broadband filtered sweeps that are alpha-type responses. By contrast, Figure 6.14 refers to a circumstance under which alpha responses cannot be recorded, i.e., a measurement with inadequate stimulation. Note the lack of time locking and amplitude increase. The alpha responses were recorded with adequate stimuli in primary sensory areas. Adequate versus inadequate differences were larger for alpha responses than for theta responses, demonstrating the functional relevance of frequency components. As an aside, in cross-modality recordings from the auditory cortex (gyrus ectosylvianus anterior) of the cat brain, we observed a complementary effect. Large alpha enhancements were present in auditory EP recordings; in visual EP recordings from the auditory cortex, such alpha enhancements were not observed. Examples: alpha responses in human EEG and MEG in cross-modality experiments — It is useful to compare cat data to EEG and MEG recordings in humans. EEG measurements were performed in 11 subjects. Figure 6.15 shows filtered curves computed from grand averages of occipital recordings (O1). The top half of the figure shows theta responses and the lower half shows alpha responses. The alpha response to auditory stimulation (inadequate for the visual cortex, occipitally located) is on the left; the response is of low amplitude. The response to visual stimulation, however, is on the right and shows a distinct alpha response. Note that the adequate–inadequate difference is less for the theta response. The hypothesis cited previously is thus supported. In cats, the alpha response is dependent on whether or not a stimulus is adequate. A correlation between the alpha response and primary sensory processing is thus plausible both for human and cat EEG–EP data. Magnetoencephalography (MEG) measurements were performed with a BTI 7-channel MEG system (Saermark et al., 1992) and with a Philips 19-channel MEG system (Basar ¸ et al., 1992b; © 2004 by CRC Press, LLC

TF1663_C06.fm Page 117 Wednesday, May 12, 2004 8:05 PM

Cat; Visual Cortex Acoustical Stimulation

Cat; Visual Cortex Visual Stimulation

A

100 µV

100 µV

50 µV

50 µV

100 C µV

100 µV

B

–500 –250 0 250 Time (ms)

500

0 250 –500 –250 Time (ms)

500

FIGURE 6.12 EPs recorded from cat brains by using intracranial electrodes (A) single EEG-EP trials filtered at 8 to 15 Hz. (B) averaged EP filtered at 8 to 15 Hz. (C) averaged EP, wide-band filtered. Left: inadequate stimulation (visual cortex recording with auditory stimulation). Right: adequate stimulation (visual cortex recording with visual stimulation). (From Schürmann, M. et al. (1997a), Int. J. Psychophys., 26, 149–170.)

Schürmann et al., 1992). The methods were similar to those used for EEG recordings where possible. We used auditory stimuli (2000 Hz; 80 dB sound pressure level) and selected sensor positions near the auditory cortex and visual cortex. The data shown in Figure 6.16 were obtained with the 7-channel system and the different positions required two experimental sessions. Panel A shows temporal recordings, panel B occipital recordings. Both cases involved auditory stimuli. For the underlying cortical areas (primary auditory cortex and primary visual cortex), auditory stimuli were regarded as adequate in the first case (panel A) and as inadequate in the second case (panel B). High amplitude alpha responses are visible in panel A (adequate stimulation). In contrast, panel B (inadequate stimulation) does not show such alpha responses. Multiple sclerosis patients with optic neuritis showed reduced alpha responses to visual stimuli, consistent with a sensory function to alphas shown in Figure 6.17 (Basar ¸ , 1998; Basar ¸ et al., 1993).

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 118 Wednesday, May 12, 2004 8:05 PM

Visual stimulus

– 100 µV

8 –15 Hz +

– 100 µV

1– 45 Hz +

–400

–200

0 time (ms)

200

400

FIGURE 6.13 Superimposed single trial EEG–EP epochs recorded from cat brains with adequate stimulation (visual cortex recordings with visual stimulus) Top: 8- to 15-Hz filter. Bottom: wide-band 1- to 45-Hz filter. (From Schürmann, M. et al. (1997a), Int. J. Psychophys., 26, 149–170.)

6.3.4.2 Cognitive Components Cognitive targets significantly influenced the alpha responses in P300. Using an OB paradigm, prolonged event-related alpha oscillations up to 400 ms were observed as noted by Stampfer and Basar ¸ and Basar ¸ and Stampfer in 1985 and later with a clear demonstration via single-sweep analysis by Kolev et al. in 1999 (see also Basar ¸ , 1998 and 1999). Memory-related event-related alpha oscillations can be observed in well trained subjects one second before an expected target (Maltseva et al., 2000). New results (Basar ¸ et al., 1997a and b; Klimesch et al., 1994) demonstrate that alpha activity is strongly correlated with working memory and probably with long-term memory engrams. Example: prolongation of oscillations — Figure 6.18 clearly shows that during cognitive tasks (oddball), the alpha oscillatory response is considerably prolonged in comparison to simple auditory EPs. The coexistence of evoked alpha oscillations with alpha blocking and event-related desynchronization (Pfurtscheller et al., 1997) hints at multiple processes reflected in alpha oscillations. Examples of such coexistence are earlier measurements where high amplitude spontaneous alpha activity coincided with alpha blocking while low amplitude alpha preceded EPs of high amplitude (Basar ¸ , 1998; Klimesch et al., 2000). Klimesch showed a plausible superposition of several types of alpha oscillations in a schematic form. Krause et al. (2001) also reported event-related synchronizations and desynchronization together. For more complete descriptions of function-related alpha, readers are referred to Basar ¸ et al., 1997; some examples follow in the next paragraph. © 2004 by CRC Press, LLC

TF1663_C06.fm Page 119 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.14 Superimposed single-trial EEG–EP epochs recorded from cat brains; inadequate stimulation (auditory cortex recordings with visual stimulus). Top: 8- to 15-Hz filter. Bottom: wide-band 1- to 45-Hz filter. (From Schürmann, M. et al. (1997a), Int. J. Psychophys., 26, 149–170.)

6.3.4.3 Resonance in Brain Responses The experiments showed that damped alpha activity is not present in all parts of the brain or elicited by all types of stimuli: Only by the combination of EP frequency analysis, adequate stimuli, and appropriate electrode positions can such activities be demonstrated. The results underline certain properties of the neural tissues under study. In the 10-Hz frequency range (filter limits of 8 to 14 Hz), we recorded large enhancements of single visual EPs in the visual cortex (also reflected in AFCs in the shape of a dominant 12-Hz peak). In the language of systems theory, significant (sharp) peaks in the amplitude characteristics of the transfer function characterize resonant behavior of the studied system. One may also express this behavior as tuning of the device and describe resonant frequency channels as natural frequencies of the system. 6.3.4.4 Multiple Functions in Alpha Frequency Window Similarly to the gamma band, the selectively distributed alpha system in the brain is interwoven with multiple and control functions: 1. The 10-Hz processes may facilitate association mechanisms in the brain. When a sensory or cognitive input elicits 10-Hz wave-trains in several brain structures, it can be expected that this general activity can serve as a resonating signal par excellence (Basar ¸ , 1980). 2. Alpha activity controls EPs. Experiments of several authors indicate that the amplitudes, time courses, and frequency responses of EPs strongly depend on the amplitude of the prestimulus alpha activity (Basar ¸ et al., 1997 and 1998). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 120 Wednesday, May 12, 2004 8:05 PM

Auditory Stimulation 4–7 Hz

Visual Stimulation –

–

1 µV

1 µV +

+

–500

0

500

–500

0

500

time (ms)

time (ms)

8–15 Hz – 1 µV +

– 1 µV +

–500

0 500 time (ms)

–500

500 0 time (ms)

Occipital (O1)

FIGURE 6.15 Frequency components of grand averaged EPs (n = 11). Top: filter limits of 4 to 7 Hz (theta response). Bottom: filter limits of 8 to 15 Hz (alpha response). Left: acoustical stimulation. Right: visual stimulation. (From Schürmann, M. et al. (1997a), Int. J. Psychophys., 26, 149–170.)

Parallel observations at the cellular level are noteworthy. Evoked oscillations in the 8- to 10-Hz frequency range in visual cortex neurons upon visual stimulation suggest a relation to scalprecordable alpha responses (Dinse et al., 1997; Silva et al., 1991). The sum of these observations permits a tentative interpretation of alpha as a functional and communicative signal with multiple functions. This interpretation of 10-Hz oscillations (at the cellular level or in populations) may be comparable to the putative universal role of gamma responses in brain signaling. Makeig et al. (2002) published a relevant study to determine whether the averaged EP is a tiny signal added to otherwise nonstimulus-related EEG oscillations or is a reorganization of ongoing EEG oscillations. Makeig’s data from 15 subjects and approximately 3000 trials per subject substantially broadened earlier experimental evidence of the role of phase reordering in EP generation. However, phase reordering of spontaneous oscillations is only one of the phenomena indicating the dependence of EPs on spontaneous EEG oscillations. The second phenomenon is enhancement” — an increase in amplitude of spontaneous prestimulus oscillations (see Figure 6.12, Figure 6.13, and Figure 6.15). In particular, higher amplitude EPs were demonstrated for trials with lowamplitude spontaneous EEGs than for their counterparts with high amplitude spontaneous EEGs (see Chapter 5). Such findings may be explained in terms of resonance of the EEG, which is a basic property of brain tissue. Responses to sensory-cognitive inputs occur in the same frequencies as spontaneous oscillations. Makeig et al. (2002) identified only cortical 10 Hz (alpha) responses. However, alpha responses have also been demonstrated in noncortical structures such as the HI, thalamus, and RF in the cat, hinting at a selectively distributed alpha system. Furthermore, alpha responses from these sites showed increased poststimulus coherence.

6.3.5 THETA OSCILLATIONS

IN

PERCEPTION

AND

COGNITION

Theta discharges are recorded from hippocampal neurons (Miller 1991), neurons in nucleus. accumbens, and cortical neurons. Hippocampal–cortical (frontal) networks operate in the theta frequency range. Theta cells in the hippocampus with multiple sensory behavioral correlates are © 2004 by CRC Press, LLC

TF1663_C06.fm Page 121 Wednesday, May 12, 2004 8:05 PM

A

80 + fT – –800

0 time (ms)

800

0 time (ms)

800

B

80 + fT – –800

FIGURE 6.16 Human magnetoencephalography responses to auditory stimulation. Averaged evoked fields recorded in a typical subject. Filter limits of 8 to 15 Hz. (A) seven channels with pure temporal location. (B) seven channels with pure occipital location. (From Schürmann, M. et al. (1997a), Int. J. Psychophys., 26, 149–170.)

described by Best and Ranck (1982). Theta EEG responses have been recorded in rats (Miller, 1991), cats, and humans (Basar ¸ , 1999). The following studies include examples related to theta oscillations in perception and cognition: 1. Experimental data suggest that event-related theta oscillations are related to cognitive processing and cortico-hippocampal interaction (Basar ¸ , 1999; Klimesch et al., 1994; Miller, 1991). 2. Theta is the most stable component of the cat P300-like response (Basar ¸ , 1998 and 1999; Sakowitz et al., 2000). 3. Bimodal sensory stimulation induces large increases in frontal theta response, thus demonstrating that complex events require frontal theta processing (Sakowitz et al., 2000; Basar ¸ , 1998 and 1999). 4. Event-related theta oscillations are prolonged and/or have a second time window approximately 300 ms after target stimuli in OB experiments. Prolongation of theta is interpreted as correlated with selective attention (Basar ¸ and Stampfer, 1985; Stampfer and ˘ et al., 1992). Basar ¸ , 1985; Basar ¸ -Eroglu © 2004 by CRC Press, LLC

TF1663_C06.fm Page 122 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.17 (A) single-sweep EEG filtered at 7 to 12 Hz. (B) filtered averaged EP (7 to 12 Hz). (C) unfiltered averaged EP. D: amplitude frequency characteristics computed from averaged EP. Left: control group subject. ˘ Right: multiple sclerosis patient. (From Basar ¸ -Eroglu, C. et al. (1993), Int. J. Neurosci., 73, 235–258.)

5. Event-related theta oscillations are also observed after inadequate stimulation, whereas event-related alpha oscillations do not exist if the stimulation is inadequate. Accordingly, the associative character of event-related theta oscillations is more pronounced than for ˘ et al., 1992). higher frequency event-related oscillations (Basar ¸ -Eroglu 6. Orienting — a coordinated response indicating alertness, arousal, or readiness to process information — is related to theta oscillations and manifested in cat experiments during exploration and searching and motor behavior (Basar ¸ , 1998 and 1999). 7. Time-locked theta responses reflect interindividual differences in human memory performance (Doppelmayer et al., 2000). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 123 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.18 Filtered averaged EEG–EPs (lead: Cz) and corresponding single sweep wave identification (SSWI) histograms from all single sweeps for four subjects in experimental session without (left) and with task (right). Frequency limits are 7 to 13 Hz. Left y axis: amplitudes of EPs in mV (referring to curve). Right y axis: sum of counts (referring to SSWI histogram bars; time interval for histogram plotting: 8 ms. (See Kolev, V. and Schürmann, M. (1992), Int. J. Neurosci., 67, 199.)

8. Miller’s results (1991) related to cortico-hippocampal signal processing support the functional role of theta transmission in all cognitive states related to association. ˘ and Demiralp (2001) indicates that theta response can be 9. The review by Basar ¸ -Eroglu associated to several sensory and cognitive mechanisms; the distributed theta system of the brain is assigned mostly to associative processes. Examples of theta responses — Demiralp and Basar ¸ (1992) measured significant theta responses following expected visual and auditory targets. Their results helped explain the functioning of the selectively distributed theta system of the brain. Evoked potentials and ERPs were © 2004 by CRC Press, LLC

TF1663_C06.fm Page 124 Wednesday, May 12, 2004 8:05 PM

3. ATT. 5 µV

–

– 5 µV

+

0

250

500

+

0

time (ms)

5 µV

500

0

500

250

500

– 5 µV

+

0

250 time (ms)

– 5 µV

+

0

250

+

time (ms)

–

VEP

– 5 µV

250

500

+

0

250

time (ms)

time (ms)

time (ms)

F3

P3

O1

500

FIGURE 6.19 Superimposed standard VEPs and responses of 10 subjects to third attended light stimuli in the visual omitted stimulus paradigm (3.ATT) obtained from frontal (F3), parietal (P3), and occipital (O1) regions. (Modified from Demiralp, T. and Basar, ¸ E. (1992), Int. J. Psychophysiol., 13, 147.)

recorded from 10 healthy subjects in auditory and visual modalities (only visual EPs will be covered in this chapter). For ERP recordings the omitted stimulus paradigm was employed. The subjects were expected to mentally mark the onset time (time prediction task) of the omitted stimulus (target). The bottom section of Figure 6.19 shows the unfiltered, averaged, and superimposed responses of 10 subjects recorded from F3, P3 and O1 leads upon application of the standard visual EP paradigm visual-evoked potential (VEP). The top row illustrates the responses of the same subjects to the third attended visual stimuli (3.ATT) in the visual omitted stimulus paradigm. Theta responses were visible even without filtering. In particularly, the frontal EP response looks like an almost pure theta oscillation without filtering. The averaged responses were filtered in various frequency bands by means of digital filters with zero phase-shift (band limits selected according to maxima in AFCs). Figure 6.20 shows the visual EPs (VEP, bottom) and the responses to the third attended light stimuli (3.ATT, top) in the omitted stimulus paradigm (superimposed) and the grand averages obtained under both conditions filtered in the theta frequency band (3 to 6 Hz). Note the similarity between wide-band filtered curves (Figure 6.19, top) and theta-filtered curves. Theta responses shown in Figure 6.20 are highest for frontal recordings (pure theta responses). The highest statistically significant theta increases during cognitive performance were obtained at frontal and parietal recording sites. In visual modality, the theta response increase at the frontal recording site was slightly higher than that of the parietal recording site (48% versus 45%). Since the cognitive task was based mainly on anticipation to an expected stimulus, it is not surprising that the greatest changes were in frontal regions. According to the review of results, it is clear that event-related theta oscillations can be considered as important building blocks of functional signaling in the brain. Research shows that, similar to the gamma and alpha bands, a selectively distributed theta system in the brain is interwoven with multiple functions.

6.3.6 DELTA OSCILLATIONS

IN

COGNITION

Thalamic neurons may discharge in the slow frequency range (Steriade et al., 1990). Slow potentials have been recorded in cortical neurons. Delta responses have been recorded in cats and humans © 2004 by CRC Press, LLC

TF1663_C06.fm Page 125 Wednesday, May 12, 2004 8:05 PM

–

–

5 µV + 3.ATT.

5 µV

–

2.5 µV

+ –

2.5 µV

+

0

–

5 µV

250

–

2.5 µV

+

0

500

individual experiments

+

250

n = 10 +

0

500

250

500

time (ms) 5 µV

–

5 µV

+

–

5 µV

+

–

individual experiments

+

VEP –

–

–

2.5 µV +

2.5 µV +

2.5 µV +

0

250

500

0

250

500

0

n = 10

250

500

time (ms) F3

P3

O1

FIGURE 6.20 Superimposed standard VEPs and responses of 10 subjects to third attended light stimuli in the visual omitted stimulus paradigm (3.ATT) and their grand averages filtered in theta frequency band (3 to 6 Hz). (Modified from Demiralp, T. and Basar, ¸ E. (1992), Int. J. Psychophysiol., 13, 147.)

(Basar ¸ and Stampfer, 1985; Schürmann, 1995). Experimental data hint at functional correlates roughly similar to those mentioned for theta oscillations, i.e., mainly in cognitive processing: 1. The responses to visual OB targets showed their highest response amplitude in parietal locations, whereas for auditory target stimuli, the highest delta response amplitudes were observed in central and frontal areas (Basar ¸ , 1998 and 1999; Schürmann, 1995). 2. Cognitive functions — The amplitude of the delta response increased considerably during OB experiments. Accordingly, it was concluded that the delta response is related to signal ˘ et al., 1992). detection and decision making (Basar ¸ -Eroglu 3. In response to stimuli at the hearing threshold, delta oscillations were observed in human subjects in consistency with the hypothetical relation to signal detection and decision making (Basar ¸ , 1998 and 1999). 4. A waveform observed in response to deviant stimuli not attended by the subject, the mismatch negativity (Näätänen 1992) is shaped by a delayed delta response superimposed with a significant theta response (Karakas¸, Erzengin and Basar ¸ 2000). Phase-locked delta responses are probably the major processing signals in the sleeping cat and human brain (Basar ¸ 1980, Basar ¸ , 1999, Röschke et al. 1995). The topographic distribution of the results is again consistent with a distributed response system. The delta response obtained during a typical P300 experiment will be described below and provides good support for the cognitive nature of delta responses. Details are given in Basar ¸ (1998 and 1999). Examples of auditory and visual P300 responses — The P300 human response to a special type of auditory stimulus shows that delta responses can be considered “real” brain responses with precise functional correlates. This was demonstrated in a study using an auditory oddball paradigm ˘ et al., 1992). Standard auditory EPs (delta response amplitude set to 100%) were (Basar ¸ -Eroglu compared with responses to oddball stimuli where the normalized delta amplitude was approximately 500% (see Figure 6.21 and Table 6.1). This remarkable increase is an example of a major change in the frequency content of an EP as mentioned early in this chapter. Taking into account the psychophysiological foundation of the P300 paradigm, this hints at cognitive processing as a © 2004 by CRC Press, LLC

TF1663_C06.fm Page 126 Wednesday, May 12, 2004 8:05 PM

TABLE 6.1 Medians of Maximum Amplitudes of Delta, Theta, and Alpha Frequency Components of Auditory-Evoked Potentials Theta (3–6 Hz)

F3

AEP 3.ATT P300

Cz

AEP 3.ATT P300

P3

AEP 3.ATT P300

O1

AEP 3.ATT P300

Delta (1–3 Hz)

Window 1 (0–250 ms)

Window 2 (250–300 ms)

1.8 3.0 (66%) 10.6 (489%)** 5.3 7.3 (37%) 10.9 (106%)** 2.4 3.2 (33%) 9.7 (304%)** 1.6 1.9 (19%) 8.2 (413%)**

4.7 6.7 (43%)** 4.5 (–4%) 8.2 9.5 (16%) 7.2 (–12%) 4.0 4.4 (10%)* 2.7 (–33%) 2.3 2.3 (0%) 2.8 (22%)

2.0 2.0 (0%) 6.5 (225%)** 3.0 2.9 (–3%) 9.7 (223%)** 1.6 1.8 (13%) 5.8 (263%)** 1.3 1.5 (15%) 3.9 (200%)**

Note: Responses to third attended stimuli in omitted stimulus paradigm (3.ATT). Responses to nonfrequent target tones in the P300 oddball paradigm obtained in frontal (F3), vertex (Cz), parietal (P3), and occipital (O1) recording sites. Percent changes of amplitudes in 3.ATT and oddball conditions as percent of the standard AEP amplitudes are shown in parentheses. Statistically significant differences are marked with symbols representing significance levels (* p < 0.05; ** p < 0.01). ˘ Source: Modified from Basar ¸ -Eroglu, C. et al. (1992), Int. J. Psychophysiol., 13, 161.

functional correlate of the delta response. The same conclusion was drawn from a study employing a visual OB paradigm with standard versus target checkerboard stimuli (Schürmann et al., 1995). In another series of experiments, subjects underwent VEP measurements with reversal of a 50’ checkerboard pattern. Two stimuli were applied in pseudorandom order; nontarget (75% occurrence) was checkerboard reversal. Subjects were instructed to pay attention to target (25%) stimuli, i.e., checkerboard reversal with horizontal and vertical displacement by 25’. Figure 6.22 shows AFCs computed from averaged ERPs. Single trials of target responses clearly demonstrate that pure delta responses are visible in such target responses even without filtering. Maxima in the 10-Hz range were common to VEP and target and largest in occipital positions. Prominent maxima in the 0.5- to 3.5-Hz range were only observed after target stimuli. Filtered averaged ERPs (delta) in Figure 6.22C show a prominent positive deflection in target responses at approximately 400 ms (amplitude: up to 244% in comparison to VEP). Amplitude differences of VEPs versus responses to target were significant for delta but not for alpha (8 to 15 Hz) responses. Thus, the delta response is clearly more dependent on the P300 task than the alpha response (Schürmann et al., 1995). The slow positive wave in target response belongs to the family of the P300 waves (Basar ¸ et al., ˘ and Basar 1993; Basar ¸ et al., 1987; Basar ¸ -Eroglu ¸ , 1991) widely accepted as related to the processing of task-relevant surprising events and reflecting a manifold of cognitive processes. © 2004 by CRC Press, LLC

TF1663_C06.fm Page 127 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.21 Time-domain and frequency-domain representations of grand averages of auditory-evoked potentials (AEP). Responses to third attended tones in the omitted stimulus paradigm (3.ATT) and to nonfrequent target tones in the oddball paradigm obtained in frontal, vertex, parietal, and occipital (F3, Cz, P3, and ˘ O1) recording sites. (Modified from Basar ¸ -Eroglu, C. et al. (1992), Int. J. Psychophysiol., 13, 161.)

Polich and Kok (1995) emphasize that, although the explanations of the P300 center around the basic information processing mechanisms of attention allocation and immediate memory, a substantial portion of P300 variation appears to be caused by factors related to alterations of the task structure and to fluctuations in arousal state of the subject. The universal or general modalityindependent character of the P300–delta response is underlined by similar responses in auditory ˘ et al. 1992; Bullock and Basar P300 experiments (Basar ¸ , 1998; Basar ¸ -Eroglu ¸ , 1988; Scheffers and Johnson, 1994). Furthermore, detecting auditory stimuli close to the hearing threshold produced slow induced delta rhythms, possibly correlates of signal detection and decision making (Basar ¸ , 1999). Section 6.3.3

6.3.7 KLIMESCH: MULTIPLE OSCILLATORY ACTIVITIES

IN

ALPHA BAND

Klimesch et al. (2000a) measured evoked and induced alpha responses by administering memory recognition tasks to humans. Their findings revealed that evoked alpha is due to transient phase locking (100 to 200 ms poststimulus) of three alpha sub-bands that can be obtained only at parietal sites. In contrast, induced alpha shows a widespread pattern of desynchronization at most recording sites. The crucial fact is that opposite alpha responses occur within similar time windows. Figure 6.23 shows this behavior.

6.3.8 OSCILLATIONS

IN

HIGHEST FREQUENCY WINDOW

The highest frequency window (frequencies above 100 Hz) gains importance due to new measurements with the help of MEG (Curio et al., 1994; Sannita, 2000) and recordings at the cellular level (Gray and McCormick, 1996). A detailed account of results in this high frequency window and their interpretations should be the subject of further review. Event-related very high frequency oscillations related to animal studies were described in sufficient detail by our group (Figure 6.4; Basar ¸ , 1980; Bullock and Basar ¸ , 1989; Gönder and Basar ¸ , 1978; Röschke and Basar ¸ , 1988).

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 128 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.22 Responses to visual-evoked potential (VEP) stimuli (left) and responses to target stimuli (right). (A) time domain. Wide-band filtered averaged event-related potentials (ERPs) in subject J.A. (B) frequency domain. Amplitude frequency characteristics computed from ERPs shown in A. Along x axis, frequency in logarithmic scale; along y axis, amplitude in relative units (dB). Amplitudes are normalized so that the amplitude at 1 Hz = 0 dB. (C) time domain. the 0.5- to 3.5-Hz filtered ERPs of J.A., a typical subject. (From Schürmann, M. et al. (1995), Neurosci. Lett., 197, 167.)

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 129 Wednesday, May 12, 2004 8:05 PM

Inset A

Inset B

Pz

P01 O1

–6.00

–4.00

µV

–2.00

0.00

2.00 ERP 4.00

6.00

–1000 –875 –750 –625 –500 –375 –250 –125 0 125 250 375 500 625 750

–1.00

µV

–0.50

0.00

0.50 Alpha ERPs 1.00

FIGURE 6.23 Top: standard ERPs (bold line) and ERPs devoid of alpha (thin line) at O1. Insets A and (B) respective results for PO1 and Pz. Bold vertical bars indicate peak-to-peak amplitudes of the P1 and N2 components. Bottom: alpha ERPs reflecting evoked alpha activity for three sub-bands (bold: lower-1; gray: lower-2; thin line: upper alpha). Note that phase locking among the three alphas contributes to the generation of the P1–N2 complex. (From Klimesch W. et al. (2000a), Neurosci. Lett., 284, 97. With permission.)

High frequency or short-latency components in scalp recordings (Cz) of human VEPs were reported in a group of early studies (Karakas¸ and Basar ¸ , 1983; Karakas¸, Basar ¸ , and Ungan, 1980). The stimulation was performed via optical step functions so as to avoid on-and off-effects that are unavoidable with light flashes (Karakas¸, Ungan, and Basar ¸ , 1980). A period of 1024 ms was analyzed in order to avoid masking effects of lower frequency components. The spectral composition of the neuroelectric responses was obtained through application of the transient response-frequency © 2004 by CRC Press, LLC

TF1663_C06.fm Page 130 Wednesday, May 12, 2004 8:05 PM

characteristics (TRFC) method described in the Appendix. The mean value AFCs showed three frequency ranges: 100 to 600 Hz, 600 to 1600 Hz, and 1600 to 3500 Hz with consistent maxima at 200 Hz, 700 Hz, 1200 Hz and 2200 Hz frequency positions. Such frequencies were much higher than those cited in the literature until publication of the Karakas¸ and Basar ¸ study (1983) showing that the 200-Hz and 2200-Hz maxima are of purely visual origin. Sannita (2000) published an excellent review of recent findings related to high frequency oscillatory systems.

6.4 SUPERPOSITION PRINCIPLE AND SUPERIMPOSED MULTIPLE OSCILLATIONS IN THETA AND DELTA FREQUENCY WINDOWS IN COGNITIVE PROCESSES: EXAMPLES In previous sections, we cited different frequency windows in order to describe functional correlates of oscillations. In this section, we emphasize the role of multiple oscillations during cognitive processes. In recent reviews (Basar ¸ et al., 1999a and b), it was argued that during functional states, EROs of various frequencies interact to give way to ERP waveforms and that this process is realized through communication networks of large populations of neurons. According to the principle of superposition, the presence of different peaks does not require different functional structures; similarly, the disappearance of peaks does not necessarily show that functional groups have ceased their activities (Basar ¸ 1980; Basar ¸ and Ungan, 1973). The application of the superposition principle is shown in the example of the auditory evoked potential of the cat HI (Figure 6.3). This section describes examples from recordings made during cognitive processes. Karakas¸ et al. (2000a and b) investigated the contributions of delta and theta responses to N200 and P300 ERP components recorded from two topographical sites (Fz and Pz) and two experimental paradigms [mismatched negativity (MMN) and oddball (OB)] that triggered different cognitive processes for the respective task performances. The results showed that the interplay of the theta and delta oscillations produced the morphology and the amplitude of the P300 and N200 components. Frequency characteristics analysis showed dominant responses in the delta and theta bands. Consistent with the AFC findings, delta and theta oscillatory responses existed in filtered waveforms in both paradigms (Figure 6.24). Theta response amplitudes of the OB and MMN were comparable within each recording site: However, as with the theta response at the Fz versus the Pz recording site, the delta response at the same recording site was twice as large. According to the literature, the MMN and OB responses seemingly represent diverse paradigms. The cognitive and behavioral states of MMN are sensory processing, sensory memory, and change detection (Näätänen, 1992). In OB, recognition of some aspect of the environment is required; the cognitive and behavioral state of OB thus involves (in addition to the processes in MMN) attention allocation, short-term memory, memory updating for stimulus recognition, and decision for a response (Donchin and Coles, 1988; Karakas¸ and Basar ¸ , 1998). Recent results (Karakas¸, Erzengin, and Basar ¸ , 2000a and b) demonstrate that the delta and theta oscillations exist as time-locked activities in both the MMN and OB paradigms and also at the N200 and P300 latencies. These findings are consistent with another statement of the theory of oscillatory neural assemblies according to which cognitive functions are represented by multiple oscillations (Basar ¸ , 1999; Basar ¸ et al., 2000). Within this multiple set, each oscillation responds to variations in task conditions by variations in latency, amplitude, duration, and strength of stimulus locking. Karakas¸, Erzengin, and Basar ¸ (2000a) demonstrated that the theta response contributes to the amplitude at the N200 latency to a greater degree and it varies congruently, not in amplitude but in duration, as attentional demands prerequisite to a stimulus encoding increase — thus the prolonged theta response under the OB paradigm.

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 131 Wednesday, May 12, 2004 8:05 PM

dB 20

dB 20

Fz

10

10

0

0

–10

Pz

–10 1

10

100 Hz

1

10

100 Hz

ERP

Delta response

µv –10 0 10

Theta response

–1000 –500

0

500 1000 ms –1000 –500

0

500 1000 ms

OB-EZ/D MMN/D

FIGURE 6.24 Amplitude frequency characteristics (abscissa: frequency in logarithmic scale; ordinate: potential amplitude |G(jw)| in decibels). Grand average (n = 42) EEG–ERPs, filtered EEG–ERPs (abscissa: time in ms; ordinate: amplitude in mV). Stimulation applied at 0 ms. Curves for Fz (left) and Pz (right) are superimposed for MMN (dotted line) and OB-EZ (continuous line) paradigms. (From Karakas¸, S. et al. (2000a), Neurosci. Lett., 285, 45.)

Theta response represents a complex set of cognitive processes whereby selective attention ˘ et al., becomes focused on a task-relevant template maintained in short-term memory (Basar ¸ -Eroglu 1992; Demiralp and Basar ¸ , 1992; Karakas¸, 1997; Klimesch et al., 1999). However, theta activity is also obtained in response to inadequate stimuli and after bimodal sensory stimulation ˘ et al., 1992; Klimesch, 1999). This theta response usually occurs later (Basar ¸ , 1999; Basar ¸ -Eroglu and represents associative processes that involve encoding of new information and cognitive oper˘ et al., 1992; Klimesch, 1999; Kocsis, Viana ations in tasks involving uncertainty (Basar ¸ -Eroglu Di Prisco, and Vertes, 2001; Yordanova et al., 1997). Findings of a study by Karakas¸ et al. (2000) further showed that the delta response contributes to the amplitude at the P300 latency and congruently that it varies in amplitude with task-relevant responding that necessitates conscious stimulus evaluation and memory updating. Delta response thus represents cognitive efforts that involve stimulus matching and decision with respect to the ˘ et al., 1992); it is thus obtained also in response response to be made (Basar ¸ 1999; Basar ¸ -Eroglu to stimuli at hearing threshold (Basar ¸ , 1999; Basar ¸ et al., 1992). Because it is recorded from various locations of the scalp, the delta response is physiologically a product of the distributed response ˘ et al., 1992). systems of the brain (Basar ¸ , 1999; Basar ¸ -Eroglu Work of the Klimesch, Gruzelier, and Chen groups confirms and elegantly extends functional relevance of function-related multiple oscillations (Chen and Hermann, 2001; Gruzelier, 1996; Klimesch et al., 2000).

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 132 Wednesday, May 12, 2004 8:05 PM

A EP

Spontaneous EEG

Spontaneous EEG

GEA-MG

1.0

0.5

0.0

0.0 GEA-IC

1.0

0.5

0.0

0.0 GEA-RF

IC-RF

1.0

0.5

0.5

0.0

0.0 GEA-HI

1.0

IC-HI

1.0

0.5

0.5 0.0 MG-IC

1.0 0.5 0.0

0 10 20 30 40 50 60 0 10 20 30 40 50 60 Frequency (Hz)

Coherence

0.0 Coherence

MG-HI

1.0

0.5

1.0

MG-RF

1.0

0.5

EP

RF-HI

1.0 0.5 0.0

0 10 20 30 40 50 60 0 10 20 30 40 50 60 Frequency (Hz)

FIGURE 6.25A Typical set of coherence functions computed from the spontaneous activities and EPs of all possible pairings of the studied brain structures during waking (scale indicated at bottom). Along the abscissa is the frequency from 0 to 60 Hz; along the ordinate is the coherence between 0 and 1. The horizontal broken lines indicate significance level (0.2 for all plots). The area under the coherence function is darkened only if the curve surpasses this level. To facilitate a comparison of coherence values computed from spontaneous and evoked parts of the EEG–EP studies, the respective coherence functions are presented as couples for all pairings of recording electrodes. (From Basar, ¸ E. (1980), in EEG Brain Dynamics: Relation between EEG and Brain-Evoked Potentials, Elsevier, Amsterdam.).

6.5 SELECTIVELY DISTRIBUTED AND SELECTIVELY COHERENT OSCILLATORY NETWORKS The degree of interaction between two signals can be measured by coherence (von Stein and Sarnthein, 2000). Coherence is a statistical measure; the value of coherence depends on the number of repeated correlations between events in the frequency domain. The phase relationship between the two signals is less relevant, but must be stable. Since the signal at each electrode site mostly reflects the network activity under the electrode, coherence between two electrodes should measure interactions between two neural populations. The statistical nature of coherence helps to unravel the interactions from noise if they repeat consistently (von Stein and Sarnthein, 2000). If two brain locations are coherent, one location drives the other or they cooperate reciprocally or they can be activated coherently by a common driver (Bullock and McClune, 1989). Findings of our research groups (Basar ¸ , 1980; Basar ¸ et al., 1979a and b) demonstrated long distance coherences in alpha, beta, theta, and delta frequency ranges in structures such as sensory cortices, hippocampi, and brain stems in waking and freely moving or sleeping cats. Coherence strength depends on stimulation modality and recording sites. During the waking stage, the coupling and/or synchronization of resonant responses from various nuclei in the alpha and beta frequency © 2004 by CRC Press, LLC

TF1663_C06.fm Page 133 Wednesday, May 12, 2004 8:05 PM

B Spontaneous EEG

EP

Spontaneous EEG

OC-LG

1.0

1.0

0.0

0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 OC-SC

1.0

0.0

0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 OC-RF

1.0

0.0

0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 OC-HI

1.0

0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 LG-SC

0.5 0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Frequency (Hz) Frequency (Hz)

Coherence

0.0 Coherence

0 10 20 30 40 50 60 0 10 20 30 40 50 60 SC-HI

0.5

0.5

1.0

0 10 20 30 40 50 60 0 10 20 30 40 50 60 SC-RF

0.5

0.5

1.0

0 10 20 30 40 50 60 0 10 20 30 40 50 60 LG-HI

0.5

0.5

1.0

LG-RF

0.5

0.5

1.0

EP

1.0

0 10 20 30 40 50 60 0 10 20 30 40 50 60 RF-HI

0.5 0.0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Frequency (Hz) Frequency (Hz)

RE 6.25B Typical set of coherence functions computed from spontaneous activities and visual-evoked potentials of all possible pairings of studied brain structures (scale indicated at bottom). Along the abscissa is the frequency from 0 to 60 Hz; along the ordinate is the coherency between 0 and 1. The horizontal broken lines indicate significance level (0.2 for all plots). The area under the coherence function is darkened only if the curve surpasses this level. To facilitate a comparison of coherence values computed from spontaneous and evoked parts of the EEG–EP studies, the respective coherence functions are presented as couples for all pairings of recording electrodes. (Modified from Basar, ¸ E. et al. (1979a), Biological Cybernetics, 34, 21–30.)

ranges can also be demonstrated by using the coherence functions between all possible pairings of spontaneous and evoked activities in structures such as the gyrus ectosylvian anterior (GEA), medical geniculate nucleus (MG), inferior colliculus (IC), RF, and HI. In Figure 6.25a, the coherences in a frequency range of 3 to 60 Hz for spontaneous activities and EPs of all possible pairings of the studied brain structures are illustrated. Due to the reasonably large number of single sweeps included in the computation of averages and the spectral window used, the auto- and cross-spectral amplitudes were smoothed adequately and a significance value of 0.2 was attained for all the curves. Therefore, the area under the coherence function is darkened only if the curve surpasses this value in order to emphasize those parts of the curves above the significance level. One recognizes immediately that in the alpha (8 to 14 Hz) and beta (14 to 25 Hz) frequency ranges, coherence usually has high values between 0.5 and 0.9 for evoked responses. However, the coherence between spontaneous activities of the same pairings of nuclei produced lower values. In the alpha and beta frequency ranges, the coherence of the spontaneous activity barely reached 0.3 in a few cases. In other words, an important coherence increase occurs upon stimulation. During slow sleep stages, the evoked coherences of all brain structures shifted to slower delta frequency ranges (Basar ¸ , 1980; Basar ¸ et al., 1979b). Figure 6.25b presents coherences in the cat brain in response to visual stimuli. In search of experimental support for the hypothesized distributed alpha response system (Basar ¸ , 1999), we measured EEG responses to visual stimuli in the cat brain (with intracranial electrodes in cortical, thalamic, and hippocampal sites). Alpha responses (10-Hz oscillations of 200- to 300-ms duration) were observed in the occipital cortex, thalamus, and HI. Remarkably, © 2004 by CRC Press, LLC

TF1663_C06.fm Page 134 Wednesday, May 12, 2004 8:05 PM

FIGURE 6.26 Coherence values (dimensionless) for electrode pairs. Thalamus–visual cortex (TH–OC, top) and hippocampus–visual cortex (HI–OC, bottom). Each pair of dots represents coherence for corresponding EEG (left) and EP (right) segments in an individual cat. Note the marked stimulus-induced increase in coherence for HI–OC. (From Schürmann et al. (2000), Neurosci. Lett., 197, 167.)

hippocampal alpha amplitudes were higher than thalamic ones, and the coherence of hippocampal cortex was higher than the evoked coherence of thalamus and cortex (Figure 6.26; Schürmann et al., 2000). Coherences increased significantly upon visual stimulation in delta, alpha, beta, and gamma frequency ranges. Hippocampo-cortical coherences were significantly higher than thalamo-cortical coherences for alpha (HI visual cortex (OC) = 0.28 versus lateral geniculate nucleus (LG) occipital cortex (OC) = 0.15), beta, and gamma frequency ranges. The effect of the interaction of channel and experiment factors (difference in coherence changes upon stimulation between the two electrode pairs) was only significant in alpha and beta frequency ranges. Upon visual stimulation, the alpha coherence HI OC increased to 0.41, whereas the LG OC increased only to 0.18 (Figure 6.26). These findings demonstrated also that during sensory stimulation, recordings of various structures showed varied degrees of coherence, thus indicating the interaction between two structures (Kocsis, Viana Di Prisco, and Vertes, 2001; Schürmann et al., 2000). The finding of varied degrees of coherence led to the concept of selectively coherent oscillatory networks. The important role of functional integration and frequency response of long-range interactions in the alpha and theta ranges in processing of the mental context confirms our long-standing view (Basar ¸ , 1980 and 1999; von Stein and Sarnthein, 2000). In the gamma frequency range, coherences between different spatial locations of the brain vary as these areas are activated by different classes of stimuli (haptic and visual) in an associative learning task (Miltner et al., 1999).

6.6 INTERIM CONCLUSIONS We have chosen examples of brain oscillatory activity to produce the following interim summary: 1. Event-related oscillations are real brain responses based on stimulation modality and cognitive states and showing clear differences in frequency, duration, and topography depending on brain state or function in study (Basar ¸ et al., 2001). © 2004 by CRC Press, LLC

TF1663_C06.fm Page 135 Wednesday, May 12, 2004 8:05 PM

2. A given function is manifested by multiple oscillations that are selectively distributed in the brain (alpha, gamma, theta, and delta systems). 3. The reverse is also true. Oscillatory responses are distributed and do not represent a single function; they are involved in several functions. 4. The application of the principle of superposition will probably give more insight into the interpretation of changes in neuroelectricity during sensory cognitive processing. 5. Depending on the function, oscillatory responses show (1) time locking, (2) phase locking, (3) enhancement, (4) delay, and (5) prolongation (or several time windows). 6. Both the mammalian brain and networks of isolated invertebrate ganglia show similar EEG-like oscillatory behaviors, indicating that oscillatory networks tuned to EEG frequencies are possible invariant building blocks in the CNS. Cortical networks and Aplysia ganglia have completely different neural structures, but these networks show similar oscillatory behavior. Such examples also appear in the mammalian brain. Cerebral and cerebellar cortex have completely different neural structures, but the frequency tuning is almost the same. This construct is may be useful for communications among distant parts of the brain as the coherence of oscillations increases in distant parts of the brain upon sensory stimulation. 7. Very high frequency oscillations in the window between 100 and 2000 Hz will probably merit further consideration. With the help of magnetoencephalography, the explorations are promising because the skull is transparent to magnetic signals and does not act as an attenuating filter in frequencies higher than 20 Hz.

6.7 DISTRIBUTED OSCILLATORY SYSTEMS AND DISTRIBUTED MEMORY 6.7.1 EVENT PROCESSING

IN

DISTRIBUTED SYSTEMS

The synchrony of selectivities described earlier by our group may have a conceptual parallel in selectively distributed processing in neurocognitive networks (Mesulam, 1990 and 1994). In Mesulam’s neurological model of cognition, the unimodal areas of the cortex provide the most veridical building blocks of experience. Transmodal nodes bind information in a way that introduces temporal and contextual coherence. The formation of specific templates belonging to objects and memories occurs in distributed form and with considerable specialization. This arrangement leads to a highly flexible and powerful computational system that underlies selectively distributed processing. In our earlier work, we often referred to distributed oscillatory systems and their resonance as selective activities. According to Mesulam, functional selectivities exist in distributed functions that are based on anatomy. The electrophysiological activities of selectively distributed systems also must exert selective behavior. Accordingly, oscillatory response susceptibility of the sensory cortices, HI, thalamus, and cerebellum should also be differentiated, depicting selective behavior to stimulation from the milieu interieur or exterieur.

6.7.2 MULTIPLE FUNCTIONS OF EROS CONVERGENCE OF CONCEPTS

AND

MULTIPLE FUNCTIONS

OF

MEMORY:

Seeing an object, even the simplest light signal, is already a memory process related to a fundamental inborn retrieval process. A baby perceives light and shows reflex responses to it before going through learning processes. The perception is probably a basic decoding process. Figure 6.27 shows responses to target and nontarget stimulations (checkerboard stimulation) in alpha and delta frequencies. The occipital 10-Hz response is large in posterior areas (related to vision). The delta response, however, is distributed and is most marked in posterior areas upon © 2004 by CRC Press, LLC

TF1663_C06.fm Page 136 Wednesday, May 12, 2004 8:05 PM

Grand average (N = 9) filter: 1–5 Hz

A

Grand average (N = 9) filter: 1–5 Hz

VEP 50' F3

F4

Grand average (N = 9) filter: 1–5 Hz target stimuli 50'

nontarget stimuli 50' F3

Cz

F4

F3

Cz

–1000 –500 0 500 1000

F4

Cz

–1000 –500 0 500 1000

–1000 –500 0 500 1000

P3

P4

P3

P4

P3

P4

O1

O2

O1

O2

O1

O2

10 µV

–

10 µV

+

Grand average (N = 9) filter: 8–15 Hz

B

F4

O1

F4

target stimuli 50' F3

Cz

–1000 –500 0 500 1000 P4

O2

2 µV

– +

– +

Grand average (N = 9) filter: 8–15 Hz

nontarget stimuli 50' F3

Cz

P3

10 µV

Grand average (N = 9) filter: 8–15 Hz

VEP 50' F3

– +

F4

Cz

–1000 –500 0 500 1000

–1000 –500 0 500 1000

P3

P4

P3

P4

O1

O2

O1

O2

2 µV

– +

2 µV

– +

FIGURE 6.27 Grand average ERPs (50’ checkerboards; n = 9 subjects), filtered at 1 to 5 Hz (a) and 8 to 15 Hz (b), respectively. Left: visual-evoked potentials (VEPs). Middle: responses to nontarget stimuli. Right: responses to target stimuli. (Modified from Basar, ¸ E. et al. (2000), Int. J. Psychophysiol., 35, 95.)

target stimuli. As explained earlier, the target signal also requires working memory processes. The fact that delta responses are most marked in posterior areas hints at selective distribution of memory oscillations. Ample occipital 10-Hz responses are not recorded in frontal locations. This indicates that the frontal lobes are not involved in primary visual processing. The response is a sign of perceptual memory: When no visual perception occurs, no 10-Hz responses occur. As to delta response in the auditory P300 paradigm, a distributed and highly enhanced response ˘ et al., 1992); the maxima were in frontal and was observed in the whole cortex (Basar ¸ -Eroglu parietal areas. The 10-Hz response to the auditory signal was missing. This finding also indicated a functional selective distribution of 10-Hz responses in primary sensory areas. The findings regarding theta responses are more complicated to interpret. In the auditory P300 paradigm, only target signals showed prolonged theta oscillations (second window) in parietal and frontal recordings, indicating a correlation to working memory. In experiments where subjects paid attention to third applied signals in evoked potential experiments, the third attended signals showed ample theta increases, especially in frontal locations. It is not our aim in this chapter to differentiate © 2004 by CRC Press, LLC

TF1663_C06.fm Page 137 Wednesday, May 12, 2004 8:05 PM

the functional roles of theta and delta responses in working memory processes. We only indicate that these oscillatory responses are selectively distributed, depending on the memory load required for experiments. The P300–40-Hz responses shown in Figure 6.11 illustrate the superposition of theta and gamma responses following omitted stimuli as targets. The combination of gamma and delta response oscillations is again involved with working memory. These findings show remarkable parallels to functional magnetic resonance imaging (fMRI) experiments showing topographical functional selectivity. Moreover, it is possible to define working and perceptual memory components in a time window already 100 ms upon a memory load. At least two types of memory responses with time hierarchies are observed: (1) sensory (perceptual) memory oscillatory responses and (2) late window responses related to working memory. Time hierarchies occurring 1 s after stimulus are not reflected in fMRI studies or found in conventional ERPs that, in the best cases, allow analysis of two relevant peaks.

6.7.3 HUMAN MEMORY PERFORMANCE

AND

TIME-LOCKED THETA RESPONSES

Doppelmayer et al. (2000) measured theta responses in relation to memory performance. Their findings and suggested interpretations appear to be in line with the publications of our group (Basar ¸ , 1999; Basar ¸ et al., 1984; Basar ¸ and Schürmann, 1994; Basar ¸ and Stampfer,1985; ˘ et al., 1992; Karakas¸, Erzengin, and Basar Basar ¸ -Eroglu ¸ , 2000a and b). Their most relevant finding with respect to memory performance indicates that the increase in theta band power is significantly larger for successful as compared to unsuccessful encoding and retrieval attempts. Theta response oscillations appeared in preferred time windows after presentation of targets only for good memory performers (Figure 6.28). Doppelmayer’s report is important for showing the influence of memory tasks on oscillating brain responses.

6.8 EEGS AND EROS AS INFORMATION CODES 6.8.1 FREQUENCY CODING

AT

DIFFERENT LEVELS

The description of neural coding by Perkel and Bullock (1968) merits important consideration. The internal modes of communication of the nervous system are primarily electrical and chemical. Neural coding refers to methods by which information is represented and transformed within the nervous system. As neural signals pass through successive synaptic junctions, the messages are dispersed and combined in new ways, and at each stage are transformed and recorded. Perkel and Bullock state that the formal properties of neural codes may be characterized by several independent aspects: 1. The referent of a code is the signal or information represented. It may be an (a) external physical quantity as in the case of a primary receptor or (b) a previously encoded chemical or electrical neural signal. 2. The transformation is the coding process in which the afferent signal is transduced and transformed, usually combined with other signals. The transformation may be characterized in terms of the carrier of the signal, the representation scheme, the mechanism of the representation, and the reliability of the representation. 3. The transmission of the encoded signal includes the spatial and temporal aspects of conduction from its source to its targets. 4. The interpretation of the signal by its target neurons or effector cells is the last stage of the coding scheme. 5. Neural coding may be described on many levels. The individual neuron, small circuit, and larger system codes involving firing rates of action potentials are very common, especially in the sensory and motor systems. © 2004 by CRC Press, LLC

TF1663_C06.fm Page 138 Wednesday, May 12, 2004 8:05 PM

A

–6

–3

µV

0

3

6 –187.5 –62.5 62.5 187.5 312.5 437.5 562.5 687.5 812.5 937.5 B

–2 M–

M+

–1

µV

0

1

2 –187.5 –62.5 62.5 187.5 312.5 437.5 562.5 687.5 812.5 937.5 C –6

–3

µV

0

3 no theta ERP 6

msec. –187.5 –62.5 62.5 187.5 312.5 437.5 562.5 687.5 812.5 937.5

FIGURE 6.28 Standard ERP (dashed line) and ERP from which theta ERP is subtracted (ERP – q, bold line) for good and bad performers [(a) and (c)]; theta ERP is (b). Note the large evoked theta for M+ in (b) and the corresponding large difference between the standard ERP and ERP – q in (a). (From Doppelmayer, M. et al. (2000), Neurosci. Lett., 278, 141.).

6. In Perkel and Bullock’s classification of neural coding, the descriptions at the higher level are codes involving nerve impulses: labeled line, time of occurrence, phase locking to a stimulus event, short- or long-term firing frequency, degree of variance of successive intervals between impulses, temporal patterns of impulses and number of impulses or duration of a burst. In larger ensembles, candidate codes include topographic distributions in a population of fibers, poststimulus, firing probabilities, and larger population phenomena measured in EEGs and ERPs.

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 139 Wednesday, May 12, 2004 8:05 PM

7. The most important remark of Perkel and Bullock concerning neural coding is perhaps the following: The problems of neural coding are not separable from questions of neural functioning, at both cellular and higher levels. Coding underlies all neural functioning to the extent that the nervous system manipulates information. Form and function on the one hand, and representation and transfer of information, on the other hand, are complementary aspects of nervous systems and must be investigated hand in hand. 8. Bullock used an extended agenda to describe brain codes and noted that the table he published demonstrated candidate neural codes or forms of representation in the nervous system. He also stated that he listed modes or representations for which there was some physiological evidence; he did not generate a list of theoretically possible codes. However, EEG-related integrative neuroscience is based on describing and interpreting codes based on the remark of Perkel and Bullock (item 7 above). We will, however, use common codings for EEGs and EPs, namely the EEG frequencies (2 Hz, 4 Hz, 10 Hz, and 40 Hz). 9. In addition to coding proposed by Purkel and Bullock, we must mention cognitive coding. In the integrative framework of information processing, existing work in cognitive psychology (Baddeley, 1997; Ellis and Hunt, 1993; Klatzky, 1980) was extensively reviewed (Karakas¸, 1997). Their papers indicated that human information processing uses different coding systems at different stages of operations. The sensory register decodes the sensory codes into neural representations of receptor activities. According to Baddeley (1997), Ellis and Hunt (1993), and Klatzky (1980), this representation is veridical. The shortterm memory (STM) processes recode veridical representations into what are basically acoustic or verbal codes. Meanwhile, the long-term memory (LTM) processes encode STM codes into propositional or analogical representations. The model of Karakas¸ and colleagues showed that at several levels of the system, the energy initiated by the stimulus is transduced into different forms. 10. Does multiple coding exist in such a way that EPs or EEGs can be considered compound potentials evoked internally or externally? In other words, do the frequency components of ERPs qualify as candidate codes? Experiments support affirmative answers.

6.8.2 MOST GENERAL TRANSFER FUNCTIONS

AND

MULTIPLE OSCILLATIONS

The transfer function is the ability of a network to increase or impede transmission of signals in given frequency channels. The transfer function, represented mathematically by frequency charac˘ et al., 1992), constitutes the main framework for teristics or wavelets (Basar ¸ , 1980; Basar ¸ -Eroglu signal processing and communication. General transfer functions could then be interpreted as networks with similar frequency characteristics that facilitate or optimize signal transmission in resonant frequency channels (Basar ¸ , 1998 and 1999). In an electrical system, optimal transmission of signals is reached when subsystems are tuned to the same frequency range. Does the brain have such subsystems tuned to similar frequency ranges or do common frequency modes exist in the brain? The empirical results reviewed here provide a satisfactory framework to Fessard’s question cited in Chapter 1. Frequency selectivities in all brain tissues containing selectively distributed oscillatory networks (delta, theta, alpha, beta, and gamma) constitute and govern mathematically the general transfer functions of the brain. To fulfill Fessard’s prediction, all brain tissues of both mammalian and invertebrate organisms would have to react to sensory and cognitive inputs with oscillatory activities or similar transfer functions. The degrees of synchronies, amplitudes, locations, durations, and phase lags vary continuously, but similar oscillations are most often present in activated brain tissues (Basar ¸ , 1999).

© 2004 by CRC Press, LLC

TF1663_C06.fm Page 140 Wednesday, May 12, 2004 8:05 PM

As to the process of coding explained in the previous section, the general transfer functions of the brain manifested in oscillations strongly indicate that frequency coding is one of the major components of brain functioning. Chapter 3 and material in this section led to development of the neurons–brain theory, supersynergy, and super-binding concepts discussed in Chapter 1. Chapter 7 will further describe these concepts that constitute leitmotifs of this book. The results cited in our books provide a real, definitive approach to this question. The described frequency characteristics in all brain tissues that embrace the resonant oscillatory processes or selectively distributed oscillatory systems of the brain (delta, theta, alpha, beta, and gamma) constitute and govern mathematically the general transfer functions of the brain (see definition of frequency characteristics in the glossary in Appendix 1). All animal and human brain tissues including isolated ganglia of invertebrates react to sensory and cognitive inputs with oscillatory activities within almost invariant and general governing frequency channels. Experimental results show that synchronies, amplitudes, durations, and phase lags vary continuously but similar oscillations are always present in activated brain tissues.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 141 Friday, May 7, 2004 8:47 AM

Integrative Brain Functions 7 Are Shaped by Superbinding and Selectively Distributed Oscillations? 7.1 RATIONALE AND USEFULNESS OF THIS CHAPTER As early as the 1940s, Hebb (1949) was advocating a distinction between long-term memory (LTM) that he argued was based on the strengthening of links between assemblies of cells within the brain, and short-term-memory (STM), which he regarded as being based on the temporary electrical activation of the relevant neurons. In Chapter 1 we briefly explained the neurons–brain theory that emerged from analysis of experiments. The concept of super-synergy was developed as a corollary of the neurons–brain theory. Based on these theoretical considerations, we decided to produce a Gedanken model to demonstrate how the grandmother percept could be activated in the cortex. The measured manifestation of the grandmother percept is described in detail in Chapter 8. The presentation of the superbinding concept in this chapter may interest readers in the discussion of the grandmother experiments in Chapter 8. The neurons–brain theory and its super-synergy corollary are covered by the expression whole-brain work and memory (Chapter 11). This chapter is intended to serve as an intermediary one. It informs readers about rapid experimental and conceptual developments over the past 5 years. Various research efforts are discussed in somewhat chronological order. The essential idea is to train experimenters in neuroscience to develop new frameworks to treat the electrophysiology of cognitive functions appropriately. Neuroscience seeks new theoretical frameworks and principles that allow clear understanding of electroencephalogram (EEG) recordings from the brain.

7.2 BINDING PROBLEM IN MEMORY PROCESSING AND GESTALT A fundamental unsolved problem in neuroscience concerns the manner in which the vast array of parallel processing and diverse neural activities occurring in the brain at any given time are bound together or integrated (Haig, 2000). For example, a visual image of an object contains a collection of features that must be identified and segregated from those comprising other objects. It is assumed generally that different features of the image are processed by different areas of the brain. How is the spatially distributed processing related to one percept integrated? This process is known as the binding problem (Singer and Gray, 1995). The results of the broad range of papers surveyed in this book suggest that the binding problem cannot be solved solely with oscillatory dynamics in the narrow gamma band. We suggest that the super-synergy of at least six submechanisms in the whole brain may present a better approach for integration (or binding). The problem of binding (or linking) was posed for experimental study; the finding was that neural activities evoked by complex features of perceptual targets are processed in separate

© 2004 by CRC Press, LLC

TF1663_book.fm Page 142 Friday, May 7, 2004 8:47 AM

neocortical areas. This implies the recombination of a neural image of the stimulus pattern and the linking of the synchronized activity to the stimuli they evoke (Engel et al., 2001). It is also hypothesized that a temporal binding mechanism may establish relationships between neuronal responses over long distances, solving the integration problem imposed by the anatomical segregation of specialized processing areas (Engel et al., 2001). Our 30 years of research demonstrating selectively distributed and parallel processing of multiple coherent oscillations clearly shows that the proposed simple binding mechanism (gamma binding) leads to a reduced framework for explanation of the formation of percepts and memory. The existence of additional empirically founded processes led us to the conclusion that the manifestations of percepts are interwoven with more extended mechanisms than simple gamma binding. New cognitive strategies showed that the delta and theta oscillations underlie both the N200 and P300 components. Complex and integrative brain functions are obtained upon superposition of several oscillations (Basar ¸ , 1980 and 1999; Chen and Herrmann, 2001; Doppelmayr et al., 2000; Haenschel et al., 2000). This constitutes the superposition principle. Selectively distributed delta, theta, alpha, and gamma oscillatory systems act as resonant communication networks through large populations of neurons, with functional relations to memory and sensory–cognitive functions. Grossberg (1999) suggests that all conscious states in the brain are resonant states triggering learning of sensory and cognitive representations in accordance with the general resonance theory described in Basar ¸ (1980 and 1999). With cognitive and sensory events, both primary sensory areas and frontal areas are activated, as functional magnetic resonance imaging (fMRI) and electrophysiological data demonstrate (Basar ¸ et al., 1999a; Courtney, 1997). Complex signals, associations, and remembering are accompanied by occipital theta or gamma activities and a large increase of frontal theta and delta frequency selectivities distributed in the brain. Relevant results collected in a volume on alpha activity clearly demonstrate that functional oscillatory brain activity is not limited to the gamma window (Basar ¸ et al., 1997). With the concept of brain oscillations, the role of temporal and spatial coherence also gained considerable importance (Grossberg, 1999; Mountcastle, 1998). Integrative activity is a function of the coherences between spatial locations of the brain; these coherences vary according to the type of sensory and/or cognitive event and possibly the consciousness state of the species. According to Bullock et al. (1989), if two brain locations are coherent, one location drives the other or they reciprocally cooperate. A common driver can also coherently activate them. Early findings of our research groups (Basar ¸ , 1980) demonstrated coherences over long distances in alpha, beta, theta, and delta frequency ranges in structures such as sensory cortices, hippocampi, and brain stems of freely moving awake and sleeping cats. Following sensory stimulation, varying degrees of coherences, and thus interactions, were obtained between recordings of the various structures (Basar ¸ , 1980; Kocsis, et al. 2001; Schürmann et al., 2000). Results pertaining to varying degrees of coherence led to the concept of selectively coherent neural populations. Thus, distributed neural populations cooperate or drive each other or they are driven by general command mechanisms in the EEG frequency channels (Basar ¸ , 1980; Haenschel et al., 2000; Kocsis et al., 2001; Schürmann et al., 2000). For short EEG segments, calculation of wavelet entropy — a newly emerging method — demonstrates that, upon application of sensory cognitive signals, an EEG goes from a disordered state to an ordered state. The distribution of entropy changes is also a facet of differentiated parallel processing; however, preliminary results show that entropy distribution does not necessarily coincide with the distribution of brain oscillations (Quiroga et al., 1999). The binding of percepts cannot be explained completely by 40-Hz oscillations. Even the simplest percept is represented by multiple oscillations occurring in a large number of selectively distributed oscillatory networks (Basar ¸ , 1999, Basar ¸ et al., 1999a and 2000; Chen and Herrmann, 2001; Haenschel et al., 2000; Karakas¸ et al., 2000b; Kiss et al., 2001; Klimesch et al., 2000; Sakowitz et al., 2001). © 2004 by CRC Press, LLC

TF1663_book.fm Page 143 Friday, May 7, 2004 8:47 AM

The event-related coherence that increases over long distances in multiple frequency-distributed networks is also selective (Basar ¸ , 1980; Schürmann et al., 2000). Accordingly, the special complex percepts or gestalts arising as consequences of 40-Hz binding hypothesized by Pulvermüller et al. (1996) and Tallon-Baudry et al. (1998) led to a high reductionism; the reports neglect the importance of various frequency levels (Basar ¸ , 1999; Chen and Herrmann, 2001; Haenschel et al., 2000; Karakas¸ et al., 2000b; Klimesch et al., 2000; Sakowitz et al., 2001). Over-emphasized interpretations relating gamma band activity only to complex processing are not tenable because isolated ganglia of invertebrates also show significant gamma responsiveness even with very simple stimuli (Schütt et al., 1992; Chapter 6, this volume). The consciousness theory of Crick and Koch (1998) also considers only the gamma band and not the multiple oscillatory behavior of the brain.

7.3 NEURONS–BRAIN THEORY AND OSCILLATORY CODES This section presents certain reviewed and experimentally derived principles in a table in order to show that oscillatory behavior increases in complexity with increasing complexity of functions and memory. The basic measured results are summarized as follows: 1. Basic neuroelectricity manifests oscillatory behavior, both at the cellular and neural population levels. The oscillations ranging from delta to gamma are natural frequencies of the brain and occur spontaneously without external stimulation. They probably arise from internally induced oscillations due to hidden sources. A. Event-related oscillations (alpha, beta, gamma, delta, and theta) are obtained as responses to sensory and cognitive stimulations and represent real responses of the brain (Basar ¸ et al., (2001). B. Oscillations from delta to gamma are recorded from isolated brain ganglia and accordingly can be considered as invariant coding occurring during the evolution. 2. In addition to their frequency values, brain oscillations vary according to such parameters as frequency locking, amplitude enhancement, time or phase locking, prolongation, delay, blocking, topography and also depend on stimulus modalities. Numerous combinations of these parameters exist. Thus the oscillatory response of the brain, with all possible variants, has the potential of representing multiple sensory–cognitive and motor functions. 3. The oscillatory responses in different frequency ranges are arranged as parallel systems that are selectively distributed in the brain. Research has shown that oscillatory responses in different frequency ranges are recorded from various distant brain structures and these responses cannot be ascribed to volume conduction. 4. The selectively distributed parallel oscillatory pathways are integrated in function. Eventrelated potentials (ERPs) are composites of oscillations in different frequency ranges and parameters. The weights of all these parameters and the combinations of the various oscillations give way to the various information processing operations such as those pertaining to the sensory register and short- and long-term memory. Column 4 of Table 7.1 lists activated memory ranging from phyletic to semantic and episodic memories. 5. It is proposed that a “percept of grandmother” is not a copy of itself or an oscillation; it is represented by multiple oscillations in selectively distributed and selectively coherent networks.

© 2004 by CRC Press, LLC

Principles of Neuroelectricity (Measured) 1. Brain shows oscillatory neural activity a. Oscillations are almost invariant in evolution b. Oscillatory responses are real brain responses

Examples (Measured)

Very simple functions or Functional building blocks

• a-response • g-response (components of auditory and visual responses) • d-response at visual and hearing threshold

• Phyletic memory • Simple sensory memory • Decision making

Basar, ¸ 1992 Basar ¸ et al., 2000 Basar ¸ and Schürmann, 1994 Demiralp et al., 1999 Gruzelier et al., 1996 Karakas¸ and Basar, ¸ 1998

• Frontal q, occipital a

{ 1. Encoding, association

Basar, ¸ 1999 Basar ¸ et al., 1987 Basar ¸ and Stampfer, 1985 ˘ et al., 1992 Basar ¸ -Eroglu Karakas¸, et al., 2000 Klimesch, 1999 Maltseva et al., 2000

2.a. Each oscillatory activity represents multiple functions vice versa b. Each function is represented by multiple oscillations Superposition principle

• Oddball Integrative neurophysiology and Cognition

• P300 superposition of dqag responses • a-expectation

Increasing number of neural populations

© 2004 by CRC Press, LLC

{

2. Working memory 3. Decision making

4. Upgrading memory

{ 5. Dynamic memory Memory is selectively distributed

3. Multiple oscillations are selectively distributed in the train as parallel processing

4. Selectively distributed parallel oscillatory systems are integrated in function. Various topology-dependent enhancements, phase locking, delays, and prolongations in parallel processing

Activated Memory (Tentative Hierarchy)

Functional Level

Very complex functions

PROPOSAL, e.g., perception of grandmother GESTALT

Episodic

Semantic memory

Increasing complexity at functional level

Increasing memory complexity

Selected References

Basar ¸ et al., 1975; 2000 Courtney et al., 1997 Fuster, 1995; 1997 Goldman-Rakic, 1997 Gruzelier, 1996 Basar ¸ et al., 2000 Burgess, 2002 Doppelmayr, 2000 Klimesch, 1999; 2000

TF1663_book.fm Page 144 Friday, May 7, 2004 8:47 AM

TABLE 7.1 Principles Derived from Measurements and Related to Electrical Activities of Neural Assemblies

TF1663_book.fm Page 145 Friday, May 7, 2004 8:47 AM

7.4 DESCRIPTION OF FUNCTION–MEMORY TABLE The first column of Table 7.1 outlines from the top down the principles ranging from very simple functions to very complex functions possible with great numbers of neural populations. The second column lists functional levels corresponding to the levels of oscillatory activities; the levels increase from top to bottom. Selected examples from Chapter 3 and Chapter 6 are shown in the third column. Simpler functions appear in the top row. One example from the domain of integrative neurophysiology and cognition is the P300 response manifested by superposition of delta, theta, alpha, and gamma responses. According to the results cited and electrophysiological results indicating multiple distributed oscillations, we complemented this table with a memory column, once more manifesting that perception and memory processes are inseparable. One tentative issue not yet possible to demonstrate is increasing the number of oscillations from sensory memory to episodic memory. The fourth column describes the corresponding tentative hierarchical levels of reactivated memories (Damasio and Damasio, 1994) as inseparable components of perception (see Figure 9.7) based on Fuster's and Hayek’s propositions. Complex memory functions (eposidic and semantic memories) are shown adjacent to the entries related to the perception of grandmother gestalt. This arrangement is in accordance with the references indicating increasing numbers of involved populations and frequency windows with increased complexity at the functional level. Table 7.1 Note: The term theory may be used to signify any hypothesis whether confirmed or not, or may be restricted to hypotheses that have been strongly confirmed as to become part of the accepted doctrine of a particular science. In its best use, it signifies a systematic account of some field of study, derived from a set of general propositions. These propositions may be taken as postulates, as in pure mathematics, or they may be principles more or less strongly confirmed by experiences, as in natural science. (From Encyclopedia Britannica.) The neurons–brain theory describes a systematical account of measurements on electrical activity of neural assemblies, and it also incorporates proposals derived immediately from measurements. Accordingly, in Table 7.1 we underline some principles on electrical activity of neural assemblies that are derived from measurements. 1) 2) 3) 4)

The Brain has natural frequencies or oscillations measurable at the cellular and population levels. Brain oscillatory responses are real responses related to function (Basar ¸ et al. 2001). Oscillations are selectively distributed and selectively coherent. Oscillations are related to multiple functions and a given function is often manifested by means of multiple oscillations (Klimesch, 1999, 2000). Principle of superposition is confirmed by several publications (Karakas¸, 2000 a,b).

A natural consequence of the Neurons–Brain Theory (or Brain Assemblies Theory) is the super-synergy in oscillatory dynamics. The proposition or a model for the electrical manifestation of the grandmother percept is based on evident facts: Research shows that the increasing complexity of percepts is accompanied with increased number of multiple oscillations in parallel with increased number of activated neural populations. Although the proposal related to the perception of the grandmother is a fictive one, it is anchored on tenable arguments: Could the grandmother percept involve multiple and distributed oscillations instead of the firing of a unique cell (Stryker, 1989)? A semi-empirical approach to the explanation of the grandmother percept would be through the concept of super-synergy. According to results described in the present previous reviews selectively distributed and selectively coherent oscillatory networks in the delta, theta, alpha, beta and gamma bands play a major role in brain functioning. Sensory and cognitive events evoke superimposed multiple oscillations that are transferred to spatially distributed tissues almost in parallel with various degrees of amplitude, latency, duration, synchronization and coherence. Sakowitz et al. (2001) reported significant gamma response amplitude increases in distributed areas of the brain and a 100% frontal theta enhancement by bimodal stimulation compared to unimodal one. It was (Basar ¸ et al., 1993) reported that retrieval of visual experience from shortterm memory is associated with 40 Hz activity. The coherences between different spatial locations of the brain vary as these areas are activated with different classes of stimuli (haptic and visual) in an associative learning task (Basar, ¸ 1988). Such findings experimentally substantiate the submechanisms that have been outlined above. Accordingly, we suggest that complex percepts (such as the visual image of one’s grandmother) are formed and/or manifested by means of the ensemble of oscillatory superbinding dynamics. The complex memory function (episodic and semantic memory) is placed in the column adjacent to the proposal to Grandmother or to Perception of Gestalt. This arrangement of the table is in accordance with the references indicating the increasing number of involved populations and frequency windows with the increased complexity at the functional level.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 146 Friday, May 7, 2004 8:47 AM

From a hierarchical view, memory types listed at the top relate to elementary sensations; those at the bottom refer to abstract concepts that, although originally acquired by sensory experience, have become independent from it in cognitive operations (Fuster, 1997). The experiments of Cohen et al. (1997) and Courtney et al. (1997) indicate distributed memory processes in cognitive tasks clearly showing that the selective distribution is demonstrated depending on the type of sensation. These descriptions represent a first draft that will be developed further after more investigation. For example, the idea that the perception of one’s grandmother is due to superposition of oscillations in parallel-distributed systems should be experimentally demonstrated. Although they may prove difficult, our laboratories are planning such experiments. The boundaries of columns and rows of this table presentation and the exact correspondence between function and letters or words are still diffuse. Why is a new theory to extend Sherrington useful? A new proposition is formed by numerous measured observations and may lead to further approaches aiming to establish a brain theory incorporating neural populations and feature detectors, as Sokolov (2001) mentioned. Moreover, in order to establish experimental strategies to explain cognitive processes and integrative brain function, the neurons–brain theory and the notion of superbinding instead of the notion of cardinal pontine cells may play major roles.

7.5 SUPER-SYNERGY: A SPATIO-TEMPORAL AND FUNCTIONAL ORGANIZATION OF MULTIPLE AND DISTRIBUTED OSCILLATIONS One important question in current neuroscience literature is how the activities within and among the separate processing areas are unified in momentary association to form a coherent representational state — the neural basis of a perception. The delineation of binding is assumed to occur by recombination of a neural image of the stimulus as a whole. Synchronized activity is linked to the stimuli that evoke it; only those activities evoked by stimuli with common stimulus features are bound together in a dynamically formed ensemble. The current literature in favor of the so-called binding theory considers binding as the unique functional correlate of gamma activity only, based on experiments with multiple microelectrodes in small portions of cortical areas (Eckhorn et al., 1988; Singer and Gray, 1995*). Although this view is interesting, the results of many laboratory studies presented in the present review clearly show that even the simplest percept is manifested by multiple oscillations in a large number of * Some issues of the temporal correlation hypothesis of Singer and Gray (1995) are: 1. Neurons in different nodes recorded simultaneously should be synchronized on a time scale of milliseconds, including cells within a single column, in separate columns in the same cortical area, between different areas in the same distributed system, and between homologous areas in the two hemispheres. 2. Cell ensembles can be formed and dissolved in milliseconds. Transitions from one ensemble to another may match the serial orders of selective attention to and perception of successive events. Individual neurons should be free to change rapidly from one synchronized ensemble to another. 3. Neurons of an ensemble within a distributed system should show synchronous episodes during stimulus-evoked responses, but should show no such relations with neurons simultaneously active in other ensembles within the same distributed system. Thus more than one (but not many) ensembles may be active simultaneously within the same system. 4. The node-to-node connections over which synchronizations are effected should be specific, and their synapses modifiable by experience. Note also that: 1. The modification of synapses was predicted by Hebb (1949). 2. This hypothesis does not include long-distance coherences, and as a consequence, long-distance information processing, for example, interaction and parallel processing between occipital and frontal lobes. 3. Multiple oscillations are not included in this hypothesis.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 147 Friday, May 7, 2004 8:47 AM

selectively distributed oscillatory networks and long distance coherences (Basar ¸ , 1980; Basar ¸ et al., 1979; Kocsis et al., 2001; Schürmann et al., 2000). The binding of percepts cannot be solely explained by the binding of 40-Hz oscillations evoked by the complex features of perceptual targets processed in separate neocortical areas. The neurons–brain theory is a consequence of measurements related to the oscillatory activity of neural populations in both human and animal experiments. Our research group introduced in the 1970s the notion of distributed and parallel processing and multiple coherent oscillations (Basar ¸ , 1980, 1983, and 1992; Basar ¸ et al., 1975a, b, and c; 1999a and b; 2000 and 2001). The consequence is that sensory–cognitive information processing manifests an extremely large number of combinations of oscillatory parameters. The crucial implication of all results cited in this chapter is that the explanation of complex percepts with 40-Hz binding leads to reductionism because long-distance event-related coherence increases in alpha, beta, theta, delta, and gamma in distributed networks are selective (Basar ¸ , 1980; Basar ¸ et al., 1979; Kocsis et al., 2001; Schürmann et al., 2000). Accordingly, the consideration of numerous selectively distributed oscillatory networks with varied degrees of event-related enhancements and coherences as manifestations of complex percepts is a much more tenable proposition. At every cognitive and sensory event, both primary sensory areas and frontal areas are activated as fMRI and electrophysiological data clearly demonstrate (Basar ¸ et al., 2000; Creutzfeld, 1983). Complex signals, associations, and remembering are accompanied by large increases of frontal theta and delta — not only in occipital theta or gamma. Instead, the huge alpha increase in the occipital regions is not registered in frontal area. Complex events in animals are accompanied by increases in frontal and hippocampal theta oscillations in addition to gamma and alpha responses. In analyses of distributed and parallel processing, we assumed that EEGs proceeded from a disordered chaotic state with large dimensions to an ordered state upon application of sensory–cognitive signals (Basar ¸ , 1980). For short segments, calculation of an emerging method known as wavelet entropy demonstrated that EEGs went from disordered to ordered states upon application of sensory–cognitive signals. The distribution of entropy changes was also part of differentiated parallel processing. However, preliminary results indicate that entropy distribution does not necessarily coincide with the distribution of brain oscillations and that distribution of theta entropy is also selective upon complex stimuli. The frontal areas show the largest entropy decreases after application of complex stimuli (Yordanova et al., 2002; Chapter 5, this volume). Taking together all these results on parallel processing, we tentatively hypothesize that the complex percepts are manifested by a super-synergy consisting of an ensemble of (at least) six submechanisms acting in parallel or in coincidence upon sensory–cognitive input. It is proposed that the coexistence and cooperative action of these interwoven and interacting submechanisms partly explain the mechanisms of integrative brain functions. Super-synergy is a natural consequence of the stated rules of the neurons–brain theory (explained earlier). The long-distance event-related coherence increases in alpha, beta, theta, delta, and gamma in distributed networks are also selective. Thus, selective distribution applies to both amplitudes of oscillations or their frequency components and also to their coherences (Basar ¸ , 1980, 1983, 1988, and 1999; Basar ¸ et al., 1975 and 1979a and b; Kocsis et al., 2001; Schürmann et al., 2000). Consequently, sensory and/or cognitive information processing is manifested in a complex signal formed from the combination of an extremely large number of oscillatory parameters that could be called oscillatory superbinding dynamics. Super-synergy involves at least six mechanisms acting in concert upon sensory–cognitive input. It is proposed that the coexistence and cooperative action of these interwoven and interacting submechanisms is a manifestation of integrative brain functions as follows: 1. The superposition principle that includes the alpha, beta, gamma, theta, and delta bands (Basar ¸ et al., 1999; Chen and Hermann, 2001; Karakas¸, 1997; Klimesch et al., 2000). © 2004 by CRC Press, LLC

TF1663_book.fm Page 148 Friday, May 7, 2004 8:47 AM

2. Activation of more than one type of selectively distributed oscillations in gamma, alpha, theta, and delta bands upon exogenous or endogenous input. These activities are manifested with a multifold occurrence of parameters such as enhancement, delay, blocking (desynchronization), and prolongation (Basar ¸ , 1980 and 1999; Basar ¸ et al., 1999, 2000, 2001). 3. Temporal and spatial changes of entropy in the brain (Quiroga et al., 1999; Rosso et al., 2000 and 2002). 4. Temporal coherences among cells in cortical columns for simple binding (Eckhorn et al., 1988; Gray and Singer, 1989). 5. Varying degrees of spatial coherence that occur over long distances as parallel processing (Basar ¸ , 1980 and 1983; Basar ¸ et al., 1999, Kocsis et al., 2001; Miltner, 1999; Schürmann et al., 2000). 6. The relationship of EEGs and event-related oscillations (EROs), with prestimulus EEGs serving as control parameters for brain responsiveness (see Chapter 5, this volume). Integrative brain function can be described as a spatio-temporal organization. The principle of superposition describes integration over the temporal axis consisting of a relationship between the amplitude and phases of oscillations in various frequency bands. Furthermore, selectively distributed and selectively coherent oscillatory activities in neural populations describe integration over the spatial axis. Consequently, integrative activity is a function of the coherences among spatial locations of the brain; these coherences vary according to the type of sensory and/or cognitive event and possibly the state of consciousness of the species. The empirically founded submechanisms are accompanied by the formation of a percept or apply to all cognitive events through combinations of the response parameters, among which are the varying strengths in the associative links among brain structures. The number of possible combinations of these variables can form an enormous number of neuroelectric configurations. Spatio-temporal and functional organizations of multiple and distributed oscillations in the brain inevitably lead to superbinding as a way to explain the integrative activity of the brain during the formation of percepts and other cognitive events. We tentatively denote superbinding as a combination of superposition, activation of selectively distributed oscillatory systems, and selectively distributed long distance coherence because superbinding reflects binding over long distances along with superposition of oscillations in the whole brain. This means oscillations in multiple frequency windows in widely distributed brain areas act in concert. They are coupled and superimposed. The brain exhibits enormous synergy in space and time. Hebb (1949) first outlined cooperation among cells. Hebb’s rule implies that information processing requires functional cooperation of distributed neurons; more exactly, groups of synapses converging on a single neuron and having a tendency to fire together are strengthened as a group. This is Hebb’s principle of cooperativity. The distributed nature of activations in cognitive tasks described in this chapter may explain why Karl Lashley (1929) thought the brain operated as a whole. The cooperation among distributed structures of the brain can be weak or strong since the coherences are selectively distributed. Parallel analysis of oscillations in several neural populations and in various frequency windows led to refinement of the descriptions of the whole brain and cooperativity. The whole brain is activated in all perceptual and memory-related mechanisms. The intensity of electrical oscillatory responses is selective in neural populations. The links or cooperativity measured by means of coherences and phase differences also show varied degrees of intensities. Accordingly, we may talk about new interpretations of Lashley’s and Hebb’s statements by using these new tools to analyze the electrical activity of the brain during sensory–cognitive processes. Barlow’s single neuron doctrine suggests that activation of individual neurons can serve as a code for complex and highly integrated functions, whereas Hebb’s rule implies that information © 2004 by CRC Press, LLC

TF1663_book.fm Page 149 Friday, May 7, 2004 8:47 AM

processing requires functional cooperation of distributed neurons. It postulates that groups of synapses converging on a single neuron and having a tendency to fire together will become strengthened as a group. This is the principle of cooperativity (Barlow, 1972; Hebb, 1949). Abeles and Prut (1996) pointed out the existence of a spatio-temporal organization in single neurons in the frontal cortex of the monkey. The superbinding is interwoven with cooperative phenomena, but relies on cooperativity of neural populations over long distances at multiple frequency levels. Superbinding is manifested by the metrics of coherence function and entropy and encompasses various degrees of cooperativity in the brain. Thus, spatio-temporal and functional organization of multiple and distributed oscillations leads to superbinding as a way to explain the integrative activity of the brain during the formation of percepts and other cognitive events. The principle of superposition describes integration over the time axis, consisting of temporal relationships in the amplitudes and phases of oscillations in various frequency bands. Selectively distributed and selectively coherent oscillatory neural populations describe integration over the spatial axis. For the formation of percepts, the entropy state of the prestimulus EEG is as important as the coherence or tuning of distributed structures. As shown in Chapter 5, entropy (measure of order and disorder) is an important causal factor for shaping electrical brain responsiveness. It is possible that hearing and/or vision reduction can occur at low entropies and high amplitudes in prestimulus EEGs. Varela et al. (2001) reviewed the mechanisms of large-scale integration that counterbalance the distributed anatomical and functional organization of brain activity to enable the emergence of the coherent behavior of cognition. They argued that the most plausible candidate is the formation of dynamic links mediated by synchrony over multiple frequency bands. This view is similar to the principle of superposition of multiple oscillations. Chapter 11 relates the neurons–brain theory and super-synergy concept in order to propose a new construct we will designate whole-brain-work and memory.

7.6 GEDANKEN MODEL: INVOLVEMENT OF SELECTIVELY DISTRIBUTED AND COHERENT ACTIVITIES OF NEURAL POPULATIONS IN GRANDMOTHER PERCEPT Does a grandmother neuron really exist or may the grandmother percept involve multiple and distributed oscillations instead of the firing of a unique cell as discussed in Basar ¸ (1999) and Stryker (1989)? A semiempirical approach to the explanation of the grandmother percept may be provided by the superbinding concept. According to previous reviews, selectively distributed oscillatory networks in the delta, theta, alpha, beta, and gamma bands play major roles in brain functioning (Basar ¸ , 1980 and 1999; Basar ¸ et al., 1997, 1999a, and 2000). Sensory and cognitive events evoke superimposed multiple oscillations that are transferred to spatially distributed tissues, almost in parallel with various degrees of amplitude, latency, duration, synchronization, and coherence. Sakowitz et al. (2001) reported a significant gamma response amplitude increase in distributed areas of the brain and an enhancement of 100% frontal theta by bimodal compared to unimodal stimulation. Retrieval of visual experience from short-term memory is also associated with gamma activity (Tallon-Baudry et al., 1998). The coherences of different spatial locations of the brain vary as they are activated by different classes of stimuli (haptic and visual) in an associative learning task (Miltner et al., 1999). Such findings substantiate the submechanisms outlined above. Accordingly, we suggest that complex percepts (such as the visual image of a grandmother) are formed and/or manifested by means of an ensemble of oscillatory superbinding dynamics. Alpha oscillations with large amplitudes were obtained in the occipital and parietal lobes upon visual stimulation, while cognitive and complex signals elicited large theta responses (Basar ¸ , 1999; © 2004 by CRC Press, LLC

TF1663_book.fm Page 150 Friday, May 7, 2004 8:47 AM

qg

Frontal lobe

a

g

qa

Temporal lobe

Parietal lobe

qg

a

Occipital lobe

g

qa

g

a

q

FIGURE 7.1 A fictive representation of oscillatory activity evoked by the grandmother percept.

Basar ¸ et al., 1999a and 2000). Recent publications demonstrate that the number of activated brain areas and the amplitudes of oscillatory activities increase with the complexity of sensory or cognitive inputs. For example, bisensory stimulation (visual plus auditory) elicited enhanced alpha response in occipital and temporal areas, whereas simple visual and auditory stimuli evoked alpha responses only in adequate areas. Memory tasks greatly increased frontal theta response (Sakowitz et al., 2001). Thus recognition in general and of particular faces is manifested by multiple frequency activations in distant locations in the brain. This alone should exclude the possibility of finding a single grandmother cell. In accordance to all these facts, Figure 7.1 hypothetically illustrates oscillations in relation to each other that were obtained in response to a picture of the grandmother emanating from the brain. For the sake of simplicity, only alpha, theta, and gamma oscillatory responses are shown. In line with the studies cited, the input of the grandmother face should elicit large theta and gamma responses and smaller alpha responses in the frontal lobes. In occipital areas, large alpha responses accompanied by smaller theta and gamma responses should be observed. Furthermore, strong theta oscillations with varying latencies should be observed in all association areas. This schematic summarizes experimental findings concerning sensory and cognitive oscillatory responses of the brain although the varying degrees of evoked or task-induced coherences have not been incorporated. With respect to the grandmother percept, one would expect occipital–frontal coherence increases in the alpha, theta, and gamma bands due to the visual grandmother image. It was proposed that the semi-empirical fictive illustration may be developed further based on appropriately planned experiments. Chapter 8 includes analysis of brain responses to pictures of subjects’ grandmothers and demonstrates how the Gedanken model (Figure 7.1) has a good prediction value after comparison with empirical measurements.

7.7 NEURAL POPULATIONS AND “FEATURE” CELLS Barlow (1995) presented an update of the neuron doctrine for perception and supports the view that the activities of single neurons or small groups of neurons can provide a sufficient basis for perceptions. Mouncastle (1998) comments as follows: It seems unlikely, however, that there is a sufficient number of detectors to account for the virtually infinite number of sensory stimuli we readily perceive, nor for solving the binding and relational problems in perceiving complex scenes. The evidence that the update account of the neuron doctrine cannot provide a basis for perception comes from results of experiments related to perception of a picture of a subject’s grandmother (see Chapter 8). In order to establish experimental strategies to reveal cognitive processes and © 2004 by CRC Press, LLC

TF1663_book.fm Page 151 Friday, May 7, 2004 8:47 AM

integrative brain function, the consideration of the neurons–brain theory and superbinding instead of the notion of the cardinal pontine cell may play a major role by extending the functional properties of single neurons to neural populations. An important proposition arises from numerous measured observations and may lead to additional approaches aimed at establishing a brain theory incorporating both neural populations and feature detectors. Sokolov’s view (2001) is relevant because he presents a constructive proposal to associate the functional roles of feature detectors with those of frequency codes of neural populations as presented in the following section.

7.7.1 SOKOLOV: FEATURE DETECTORS* E. Sokolov (2001) wrote a constructive and critical review of our previous work: E. Basar ¸ on the basis of systematic studies of brain oscillations using EEG and event-related potentials suggests, “frequency encoding.” Particular frequencies occurring at particular time intervals constitute “letters” of the code. Combinations of different frequencies build up “words.” The distribution of such a “frequency code” in different brain areas is regarded as a basis of cognitive processes. Such an approach is supported by coincidence of spectra of background EEG and following event-related potentials. Distributed frequency encoding opposes “detector theory” and “gnostic unit concept.” There is a bulk of evidences concerning specificity of cortical neurons in different animals. In humans such a specification refers to verbal encoding. At the same time neurons generate “bursts of spikes” and regular pacemaker potentials. I suggest a compromise between “frequency code” and “feature detector code.” The compromise is based on unification of these aspects of neuronal activity. Specific feature detectors and gnostic units are tuned according to the vector encoding applied to presynaptic inputs and synaptic contacts. From this standpoint maximal response of a feature detector is achieved when presynaptic excitation vector and synaptic excitation weight vector of a detector coincides [sic] in orientation, so that scalar (inner) product of these vectors reaches maximum. Frequency code makes it possible to extend specificity in the time domain producing “frequency and phase selective tuning” of feature detectors. The EEG is a result of dominating neuronal oscillations under particular conditions. This means that frequencies are tools for more precise neuronal tuning. An important role in such a tuning of endogenous pacemaker oscillations results in frequency and phase-tuning of the feature detector. Thus the feature detector becomes state-dependently tuned. The suggested compromise is a working hypothesis that has to be tested by intracellular recordings to evaluate the relationship between stimulus specificity of neurons and frequency aspects of presynaptic spikes and postsynaptic pacemaker oscillations. Parallel-recorded focal potentials will show intracellular contribution to extracellular phenomena evident as EEG oscillations.

A possible metaphor for the efficiency of large neural populations and for powerful feature detectors in human societies is the following: In a population of humans there are genies or leading individuals with high level and multiple (numerous) abilities. However for the performance of an efficient action (instrument building, wars, football, any type of powerful performance, orchestral concerts) collective and mostly coherent behavior of a great number of individuals is necessary.

* Feature detectors are explained in the glossary. For information about the positions of feature detectors in the hierarchy of memory types, see Section 9.3.3.4.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 153 Friday, May 7, 2004 8:47 AM

Experiments 8 Grandmother in Perception of Memory: Recognition of Gestalts Erol Basar ¸ and Murat Özgören 8.1 INTRODUCTORY REMARKS The experiments described in this chapter were performed during the course of writing this book. References to the experiments have been included in chapters throughout the book since the resulting data were found to be relevant to understanding perception and memory and also the transition between memory states. In Chapter 7, we presented a model predicting how a picture of a grandmother is perceived and how the percept could be manifested by selectively distributed and enhanced multiple oscillations. This semi-empirical Gedanken (thought) model was published and explained at several conferences. We started a series of experiments at our laboratory in Bremen to test the validity of this model in February 2002. The results partially supported the fictive model presented in Section 7.6. We also discovered a number of novelties and more precise findings not predicted by the model. The results were highly encouraging: On one hand, we had developed a theory that was highly speculative although it was anchored on a large number of measurements and published interpretations of the last 30 years. Conversely, the theory appeared testable despite the difficulty and complexity of the grandmother experiments. The experiments with 12 subjects were performed over a 6-week period in 2002. The computations were performed in Bremen and in Izmir. The outcome of the grandmother experiments favored the superbinding approach and clearly showed the immense reductionism of the simple binding theory based solely on gamma oscillations. The experiments described in the next sections represent only the beginning of a new perspective. We hope that repetitions with more subjects and extensions of the new strategy in collaboration with colleagues in other laboratories may validate our approach and open new horizons in electrophysiology-oriented memory research. The rationale for writing these preliminary remarks is the completion of promising experiments that enabled us to realize how well the results fit the hypothetical and empirical approaches presented in Chapter 7. The Gedanken model became reality in a very short time. For preliminary publications of the results, see Basar ¸ et al. (2003a,b; 2004a,b). Postscript — Experiments with 29 subjects were completed in 2003. The results accorded well with the preliminary results and showed greater efficiency. Furthermore, the addition of temporal recordings and the evaluation of differences present in both brain hemispheres confirmed improved differentiation of the grandmother’s face and the anonymous face. Details of these extended results cannot be described in this book. The temporal recordings and the transition from semantic memory to episodic memory will be discussed briefly.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 154 Friday, May 7, 2004 8:47 AM

8.2 KLIMESCH: ROLE OF THETA AND ALPHA OSCILLATIONS IN MEMORY AND ATTENTION FUNCTIONS Reports analyzing memory functions via event-related theta and alpha oscillations are rare. Klimesch and coworkers pioneered a series of experiments related to alpha and theta oscillations, working memory, and semantic and episodic memories. More recently, personal communications from Wolfgang Klimesch (2002) describe several important issues in memory research. A series of studies provided evidence that different frequency bands in the theta and alpha range are associated with different types of cognitive processes. While event-related changes in the theta band appear to be related to encoding and retrieval processes of a complex working memory system (WMS), the upper alpha frequency range responds selectively to sensory–semantic memory processes of a complex long-term memory system (LTMS) and the lower alpha band to attention processes. Most importantly, the functional specificity of these frequency bands can be observed only if frequency boundaries are adjusted to individual alpha frequency (IAF) and rather narrow bands are used (Klimesch et al., 1996; Doppelmayr et al., 1998). In an early study, upper alpha desynchronization in semantic and episodic memory tasks was investigated (Klimesch et al., 1994). Subjects first performed a semantic task that required them to judge whether sequentially presented concept–feature pairs (such as eagle–claws or pea–huge) were congruent. Then, without prior warning, they were asked to perform an episodic recognition task. They then had to indicate whether a particular concept–feature pair was presented during the semantic task. Because pairs of items were presented, the episodic and semantic tasks could be performed only after the second item of a pair (i.e., the feature) was presented. Thus, the critical issue was to compare the extent of band power changes in the theta and lower and upper alpha frequency ranges during the presentation of the concept and feature word in the episodic and semantic tasks. During semantic processing, a significant decrease in upper alpha power was observed, whereas during episodic retrieval, a significant increase in theta power developed. The conclusion from this study was that dissociation occurred between theta synchronization (maximal during the processing of new or episodic information) and upper alpha desynchronization (maximal during retrieval and processing of semantic information). These findings were replicated in more recent experiments (Klimesch et al., 1997a and b; 2000a and b; 2001a and b; cf. Klimesch, 1999 for a review). Because episodic memory is closely related to WMS processes, typical working memory tasks were used to analyze theta band power changes with respect to successful or unsuccessful episodic encoding. During encoding, words that could be remembered in a later free recall task produced significantly larger phasic increases in theta power than words that could not be remembered later (Klimesch et al., 1996; 1997a). Similarly, during successful retrieval in a word recognition task, correctly recognized words showed significantly larger phasic theta responses than correctly identified distractors and false alarms. Other groups reported similar results (Gevins et al., 1997; Kahana et al., 1999; Tesche and Karhu, 2000). In a recent study (Sauseng et al., 2002), designed to test possible interplay between working and LTMSs, theta and alpha responses were analyzed upon presentation of a cue in a recall paradigm. Subjects had first to learn a verbal label (cue) for each of a set of eight pictures (up to a 100% learning criterion) and then to retrieve each picture after the respective label was presented. The findings indicate that in response to the presentation of the label, theta waves revealed an interesting topographical phase relationship. When subjects attempted to retrieve (up to about 800 ms after the cue was presented), evoked theta oscillations spread from frontal to occipital recording sites. After about 800 ms poststimulus, the direction reversed and theta spread in the opposite direction to frontal sites. Most interestingly, when the latency of the phase reversal was determined for each subject and correlated with retrieval success, a significant negative correlation was observed. This indicated that good performers showed shorter latencies for the reversal of theta at occipital sites.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 155 Friday, May 7, 2004 8:47 AM

Thus, theta is associated with the actual retrieval of a picture from a posterior storage network. It is important to note that a change in phase direction is meaningful only if a concomitant and transient change in frequency can be observed. During that time, a significant but transient increase in theta frequency could indeed be observed.

8.3 GRANDMOTHER PARADIGM AND GESTALT EXPERIMENTS According to our hypothesis of superbinding in integrative brain function, the grandmother picture would activate selectively distributed and coherent oscillations in the whole brain; frontal theta response, occipital alpha response, and distributed gamma were the major components (Basar ¸ , 2001). This hypothesis was recently tested by recording event-related oscillations (EROs) of 12 subjects upon a random sequence of presentations involving three stimulations cited below.

8.3.1 EXPERIMENTAL STRATEGY We designed a strategy consisting of the application of three experimental data recording sets and three different types of stimulations. The stimulations consisted of (1) a simple light signal whose luminance was around the same level as the luminances of the other two stimulations; (2) a stimulation signal showing the face of an anonymous elderly woman; and (3) the face of the subject’s grandmother. The subjects were known, healthy, people ranging in age from 15 to 32 years. The pictures of the faces of the subjects’ grandmothers were prepared before measurements were made. 8.3.1.1 Electrophysiological Recording Electroencephalograms (EEGs) were recorded from silver/silver chloride electrodes at F3 , F4 , Cz , C3 , C4 , P3 , P4 , O1, and O2 locations according to the 10-to-20 system (Jasper, 1958). In later experiments, T5 and T6 locations were also used. The electrodes were affixed with Grass electrode paste. Linked earlobe electrodes served as references. Electrode impedance did not exceed 5 kV. Electrooculogram (EOG) readings from the medial upper and lateral orbital rims of right eyes were also registered. The EEGs were amplified via a Nihon Kohden 4421G EEG apparatus with band limits of 0.1 to 70 Hz and 24 dB/octave. The EEG was digitized online with a sampling rate of 250/s and stored on the hard drive of a computer for offline analysis. An additional notch filter (36 dB octave) was also applied to remove the main interference. For EOG recording, a time constant of 0.3 s with a low-pass filter at 70 Hz was used. All channels were displayed on paper and online by monitor scope in order to observe single trials and averaged signals. Electromyelogram (EMG) recordings were also used to estimate the contributions of the motor potentials to the main recordings. 8.3.1.2 First Data Recording (Random) Set The three stimulations described above were applied in random sequences with intervals varying between 3.5 and 4.5 seconds. Seventy-five stimulation signals were applied with approximately the same distribution of each type. After stimulation and recording, the grandmother, face, and light (controls) responses were divided into subsets. 8.3.1.3 Second Data Recording (Regular) Set In the first subset of recordings 30 light signals were applied; in the second and third subsets, the anonymous face and grandmother’s face, respectively, were applied. The subjects in all experiments reported that they clearly recognized and differentiated the faces of their own grandmothers. The results were analyzed by digital filtering of data first in five frequency windows of the EEG © 2004 by CRC Press, LLC

TF1663_book.fm Page 156 Friday, May 7, 2004 8:47 AM

FIGURE 8.1 Photographs of grandmother and unknown person presented to one of the subjects.

convention. The unfiltered conventional EPs showed important topological differences for all types of stimulations. For the sake of simplicity, we will describe in this chapter only the filtered evoked potentials (EPs) that showed more differentiated results. This is important for demonstrating the distributed nature of the manifestations of the grandmother percept. Figure 8.1 shows the anonymous and grandmother faces presented to one of the subjects.

8.3.2 EVENT-RELATED OSCILLATIONS ARISING FROM LIGHT, ANONYMOUS FACE, AND GRANDMOTHER FACE STIMULATIONS We analyzed grand average results from 20 experiments with 12 subjects. For the sake of simplicity, we only describe the responses from five locations and illustrate only EROs in frontal and occipital areas. All individual experiments showed results to the grand averages. These preliminary results manifest stable and clear differentiations in response to various stimuli. Figure 8.2 presents the grand average of the event-related responses to all types of stimuli from 20 experiments. It is interesting that the grand average of unfiltered conventional event-related potentials (ERPs) shows almost no difference between the picture of an anonymous face (dotted line) and the subject’s own grandmother’s picture (solid black line). Figure 8.3a illustrates only frontal and occipital alpha, theta, and delta responses upon application of the three different types of stimulations. We did no include all frequencies and all recording sites because it is difficult to completely evaluate the large amount of information obtained at a preliminary stage. We also limited statistical evaluations to the results illustrated in Figure 8.3. The analysis of oscillatory responses showed evident differences upon presentation of the two faces. In the following sections, we will explain these differences in detail. 8.3.2.1 Topologies of Delta Responses Delta responses progressively increased from exposure to the light, the anonymous face, and the subject’s grandmother. A phase change probably occurred between the light stimulation and presentation of the picture of the subject’s grandmother. Significant increases were observed in posterior delta responses during perception of the anonymous face and the grandmother’s face. In comparison to frontal delta responses, posterior (P3 , P4 , O1, and O2) delta responses showed increases of 200%. The occipital delta responses to simple light did not show this significant increase. In the light condition, delta demonstrated almost the same amplitude in occipital and frontal responses. Accordingly, (1) the phase delay and (2) the increase of delta in posterior areas in face and grandmother in comparison to light stimulation constitute important findings of these experiments. © 2004 by CRC Press, LLC

TF1663_book.fm Page 157 Friday, May 7, 2004 8:47 AM

F3

C3

F4

Cz

C4

P3

P4

O1

O2

Anonymous person Face of subject’s own grandmother Simple light

FIGURE 8.2 (See color insert following page 200.) Grand average of event-related responses to 3 types of stimuli from 20 experiments.

As seen in Chapter 3 and Chapter 7, the amplitude of the delta response increased considerably during oddball experiments. We concluded that delta response is related to signal detection and decision making. Further, delta responses to visual oddball (OB) targets produced the greatest response amplitudes in parietal locations. For auditory target stimuli, the greatest delta response amplitudes were observed in the central and frontal areas (Stampfer and Basar ¸ , 1985; Basar ¸ ˘ et al., 1992; Schürmann et al.,1995; Demiralp et al., 1999a and b). Eroglu 8.3.2.2 Topologies of Theta Responses Frontal theta responses also showed important differences after exposure to three types of stimuli. Frontal theta activity was greatest in the grandmother response (approximately 8 mV). The second largest reaction was the frontal theta response to light. Theta response decreased from the frontal to posterior areas. In the frontal areas, the grandmother response was 15% larger in comparison to simple light and 70% larger than the face response. However we must note that the frontal face theta response was more prolonged in comparison to other stimulations. Theta responses in the central, parietal, and occipital areas also had more prolonged character for the face and grandmother stimuli in comparison to the light response. A second time window peaked around 300 ms in the face and grandmother responses. The occipital light response was significantly higher in comparison to responses to both faces (60%). The most important finding in this frequency range was the significant dominance of the frontal response. Note that the frontal theta response was the greatest; the second greatest was the Cz response. The © 2004 by CRC Press, LLC

TF1663_book.fm Page 158 Friday, May 7, 2004 8:47 AM

0 2

µV

–2

µV

–2

µV

–2

0 2

–1000 –800–600 –400 –200

0 200 400 600 800 1000 msec

–1000 –800–600 –400 –200

F4 Delta

0 200 400 600 800 1000 msec

F4 Theta

F4 Alpha

µV

0 2

0 2

–1000 –800–600 –400 –200

0

200 400 600 800 1000

0 200 400 600 800 1000 msec

–2

µV

0 2

c

–1000 –800–600 –400 –200

–2

µV

–2

0 2

–1000 –800–600 –400 –200

0

200 400 600 800 1000

–1000 –800–600 –400 –200

0

200 400 600 800 1000

msec

msec

msec

O2 Delta

O2 Theta

O2 Alpha

FIGURE 8.3A Frontal and occipital alpha, theta, and delta responses upon application of three types of stimulations. Gray line = light; dotted line = anonymous face; solid black line = subject’s grandmother.

T6

uV

–2

T5

Light

–1 0

Anonymous Face

Subject’s Grandmother

–1000

0 ms

1000

–1000

0 ms

1000

FIGURE 8.3B Theta as grand average random set, T6 recording (left column), and T5 recording (right column). Upper lines show responses to simple light. Middle lines show responses to the unknown face. Lower lines show responses to grandmother’s face.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 159 Friday, May 7, 2004 8:47 AM

major operating rhythm (MOR) of the frontal lobe was the theta oscillation.* As noted in Chapter 6, theta oscillations are mostly involved in association processes. In recent experiments, we included recordings of temporal lobes (T5 and T6). Although we have data from only nine subjects, the experiments indicate a stable trend as shown in Figure 8.3b. The theta responses from the right hemisphere show the largest responses to the grandmother and anonymous face pictures in comparison to all recordings. The differences between the right and left hemisphere are crucial. The amplitude of the posterior right temporal lobe (T6) is 600% higher than the amplitude of the left posterior temporal lobe (T5). Accordingly, the comparison of left and right hemispheres adds valuable information to the electrophysiological mapping of complex signal processing in the brain, provided that analysis using the oscillatory approach with multiple components is performed. Grüsser’s group published a series of relevant pioneering studies for the recognition of known and unknown faces in 1984. According to their results, the perception of a face by a human observer presumably evokes a complex set of data processes in the observer’s brain that differs from processes that occur when he perceives other complex visual objects. The authors described the relevance of the temporo-occipital areas to face recognition based on animal experiments (see Bötzel et al., 1989). However, the conclusion drawn from these experiments is the fact that the temporal regions, central cortical areas, and limbic structures participate in the process of face recognition. Unfortunately, this group did not comment on frontal activities and could not apply the oscillatory approach.** The theta responses to the anonymous face picture varied during the course of the experiments. The measurement of three subsets of the experiments was time-consuming. Toward the end of the experiment, the frontal theta response usually manifested a dynamic change: The mean values of theta responses at the beginnings and near the ends of subsets of experiments were modified. The frontal theta responses to the anonymous face near the end of the experiment were less prolonged and enhancements were higher than they were at the beginning of the experiment. In the experiments cited in Section 3.4.1, the amplitudes of theta responses were enhanced, and the prolongation of responses diminished as the subjects became accustomed to the repetitive targets. We could speculate that the anonymous figure was more familiar to the subjects later in the experiments and tended to elicit oscillatory responses similar to those manifested by the perception of the pictures of their grandmothers. The results indicate plasticity of oscillatory behavior during the transition from semantic memory to the episodic memory state. These preliminary results of transition between memory states emphasize the fact that the reciprocal activation of the attention, perception, learning, remembering (APLR) alliance is essential for transition between the two memory states. The observation of EEG oscillations is a good strategy to measure these changes. * Multiple functions of major operating oscillations (rhythms) in the theta band. Event-related potentials obtained with paradigms inducing focused attention, P300, and stimuli giving rise to high expectancy states showed marked electrophysiological changes in the frontal cortex, parietal cortex, and limbic system. Published studies (Basar, ¸ 1990; Demiralp and Basar, ¸ 1992 and 1994) and Section 6.3.5 indicate that the frontal areas of the human cortex reacted with enormous theta enhancements to cognitive stimulation requiring focused attention and short-term memory. In the human frontal cortex, a theta increase of 50% was recorded while the subject paid attention to an expected (100% probability) target. Similar experiments with cats demonstrated a theta increase of 40% in the CA3 layer of the hippocampus. In P300 experiments, learning tasks led to a theta increase with a time delay in frontal and parietal recordings. These results clearly show that cognitive tasks lead to marked theta increases in evoked potential components. When comparing the results of experiments with simple light or sound stimulation in which the evoked potentials contain dominant alpha responses, we are inclined to state that cognitive loading increases the weights of theta components in comparison to alpha components. Furthermore, the increase in theta responses mostly takes place in frontal, hippocampal, or parietal structures. Even the omitted stimulus that gives rise to a P300 response in the cat hippocampus has a dominant theta component, again with the largest component in the CA3 layer (Chapter 4, this volume). ˘ proposed to provide the necessary programs for an oscillatory approach to ** In the 1990s E. Basar ¸ and C. Basar ¸ -Eroglu relevant recordings of O.J. Grüsser, who visited Lübeck several times. Unfortunately, this project was not realized.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 160 Friday, May 7, 2004 8:47 AM

We ask readers to compare the models of memory levels and hierarchies and the possible transition loops shown in Figure 9.7 and Figure 9.12. More advanced analysis involving 28 subjects and related to change of theta responses will be described in Section 8.5.4. It should be noted that the experiments using different types of P300 paradigms discussed in Chapter 3 and the grandmother experiments may be extended to include measurements of responses of large groups of subjects and applications of combinations of all these paradigms in order to observe and compare learning effects and plasticity in brain responses. Chapter 11 discusses plasticity and the theory of whole-brain work. We describe changes in expectancy alpha oscillations in Chapter 3. 8.3.2.3 Topologies of Alpha Responses Frontal alpha responses to light stimulation are not phase-locked to stimuli; they are delayed and have low amplitudes. They show longer durations in frontal areas in response to the light. However, phase-locked and large alpha responses are present in responses to the pictures of the grandmother and the anonymous person. F3 and Cz alpha responses to pictures of the anonymous person and the subject’s grandmother picture were approximately 250% larger. Occipital alpha responses to the grandmother pictures were very high in amplitude and phase-locked at the signal application (0 ms). The results of light stimulation confirm our earlier data with significant enhancements (see Section 8.2). After application of stimulation with the anonymous face, occipital alpha responses had lower amplitudes and were not phase-locked in contrast to results from the light experiments. However, oscillation was prolonged (possibly in the second window). Less distinguished alpha responses were observed in F4, C4, P4, and O2. Occipital alpha responses to the grandmother picture showed approximately 9 mV pp (peakto-peak) at the application of stimulation. They were phase-locked and almost congruent with occipital light alpha response. Note the similarity of alpha response to stimulation with the light and the grandmother’s picture in contract to stimulation with the photo of the anonymous face. Occipital alpha response to the grandmother picture was about 300% larger in comparison to response to the anonymous face. Occipital large alpha response reflects the processes of iconic memory (phyletic and inborn, persistent memory; see Figure 9.7). The occipital low and prolonged alpha response reflects the activation of semantic memory — a type of memory more abstract and in the case of an unknown face more difficult to identify and recognize. The low alpha amplitude response may indicate that the numbers of neural populations activated are smaller and have less synchrony in comparison to the grandmother activations. The prolongation of the alpha response may mean that the occipital area is engaged or works for a longer period to identify the yet-unknown face. The same interpretations are also valid for the frontal theta response described above. The major operating rhythm of the occipital lobe is alpha, as described in Chapter 5 and the alpha response of the occipital lobe is correlated with visual processes (Section 6.3.4.1). 8.3.2.4 Distributed Beta and Gamma Responses The preliminary findings in the beta and gamma frequency ranges also manifested selective topological distribution similar to those in lower frequency ranges. However, the results showed larger individual fluctuations. We will briefly discuss important features without illustrating them. The frontal beta responses as a grand average (F3 and F4) produced the highest amplitudes after stimulation with the unknown face (approximately 1.2 mV peak to peak). The value was lower in occipital recordings (1 mV peak to peak) and showed clear enhancement. Conversely, beta responses to the picture of the grandmother indicated lower frontal responses with amplitudes of approximately 1 mV. Responses were clearly more prolonged. The gamma frequency range results showed approximately 3 mV pp in the grand average of occipital recordings from 28 subjects. The gamma responses were selectively distributed and showed © 2004 by CRC Press, LLC

TF1663_book.fm Page 161 Friday, May 7, 2004 8:47 AM

prolonged oscillations for both pictures. However, the gamma responses to the anonymous face and the grandmother were less clear than the beta response. Without appropriate statistical analysis, it is difficult to clearly describe the gamma responses.

8.3.3 DIFFERENTIATION OF RESPONSES IN DELTA, THETA, AND LOWER AND UPPER ALPHA FREQUENCY BANDS: PRELIMINARY STATISTICS Figure 8.4 and Table 8.1 present the first statistical evaluations related to the experiments. The first results indicate good congruence with the oscillatory responses shown in Figure 8.3a. The occipital alpha responses seem insignificant at first glance. However, by separating the alpha frequency window into lower and higher alpha bands, differentiated responses with large enhancements and desynchronizations were observed. For the grandmother picture (episodic presentation), large enhancements both in the lower (7 to 10 Hz) and higher alpha bands (10 to 13 Hz) were observed (see Figure 8.5). For the anonymous face (semantic presentation), desynchronization in the upper alpha band and only a negligible enhancement in comparison to the grandmother picture were observed. These results may open significant future research possibilities for investigating face recognition, gestalt psychology, and differentiation of semantic and episodic memories.

8.4 RECOGNITION MEMORY AND GAMMA OSCILLATIONS Burgess and Ali (2002) recently examined gamma oscillations during two subjectively distinct memory states elicited through recognition memory. During a verbal recognition memory test, the subjective experience of recollection induced higher amplitude gamma oscillations than the subjective experience of familiarity 300 to 500 ms after stimulus presentation. The results indicate that memory conditions differing only in the richness of the subject’s subjective experience show distinct patterns of functional connectivity based on covariance analysis in the gamma frequency ranges. The richer memory experience was associated with greater gamma amplitude and greater functional connectivity between frontal and parietal sites. The authors note that the hypothesis that gamma is associated with feature binding is supported by their results. The results of Burgess and Ali (2002) are of great importance because they emphasize the greater amplitudes of gamma oscillations that occur in the late 300- to 500-ms window. Section 3.4.5 describes increased gamma amplitudes during difficult working memory tasks that also occurred in the late time window. The cat hippocampus (HI) reacted to omitted stimuli with higher gamma amplitudes in the late window, around 300 ms after endogenous cognitive input (Chapter 4). Based on the grandmother experiments, it is clear that feature binding cannot be solely explained by simple binding theory and gamma oscillations.

8.5 WHAT DOES THE GRANDMOTHER PARADIGM MEAN? ARE OSCILLATIONS DISTRIBUTED TEMPLATES IN MEMORY ACTIVATION? The preliminary and tenable consequences of the so-called grandmother experiments cannot be fully described yet. At this point, the most important and relevant consequence is that the whole brain and all oscillations are activated during recognition or remembering of one’s own grandmother and an anonymous face that was unknown at the beginning of the experiment. The ensemble of responses behaved like a three-dimensional construct consisting of temporal, spatial, and frequency spaces. The responses to both faces were not represented solely by one location and unique frequency or at the same position along the temporal axis, as differentiated delay and prolongation of multiple oscillations selectively distributed throughout the whole cortex show. © 2004 by CRC Press, LLC

TF1663_book.fm Page 162 Friday, May 7, 2004 8:47 AM

Maximum Amplitude

Maximum Amplitude 8

Maximum Amplitude Maximum Amplitude θ

4

µV

2

0

F3

F3

F3

Location B

F3 Location

C

FIGURE 8.4 First statistical evaluation: a = alpha responses; b = delta responses; c = theta responses.

© 2004 by CRC Press, LLC

Results of Repeated Measures Main and Interaction Effects Maximum Amplitude Delta band Theta band Alpha band

Main Effects

Interaction Effect

Condition

Location

*** ** —

*** — **

*** — ***

** p < 0.01, *** p < 0.001 Results of detailed analysis: one-way ANOVA (+Posthoc-Tests, Bonferroni and Scheffe) Alpha Location (Scheffe Test): condition light significant difference between maximum amplitudes of

Condition (Bonferroni-correction): position f3 significant difference between maximum amplitude of f4 significant difference between maximum amplitude of Delta Location (Scheffe Test): condition grandma significant difference face Condition (Bonferroni): position o1 significant difference between max. Amplitude of o2 Theta No Location effect Condition (Bonferroni): at position f3 significant difference between max. Amplitude of

© 2004 by CRC Press, LLC

f3 and o1: p < 0.005 f3 and o2: p < 0.0001 f4 and o1: p < 0.0001 f4 and o2: p < 0.0001 face and light: p < 0.05 face and light: p < 0.01 grandmother and light: p < 0.01 f3 and o2: p < 0.005 f4 and o2: p < 0.005 f3 and o2: p < 0.005 f4 and o2: p < 0.005 grandmother and light: p < 0.01 grandmother and unknown face: p < 0.01 grandmother and light: p < 0.0001 face and light: p < 0.001

grandmother and face: p < 0.05

TF1663_book.fm Page 163 Friday, May 7, 2004 8:47 AM

TABLE 8.1 First Statistical Evaluation Related to Experiments

TF1663_book.fm Page 164 Friday, May 7, 2004 8:47 AM

O2

Light

uV

–2

Anonymous Face

0 Subject's Grandmother

2

–600

–400

–200

0 msec

200

400

600

Lower Alpha 7–10 Hz O2

Light

uV

–2

Anonymous Face

0 Subject's Grandmother

2

–600

–400

–200

0 msec

200

400

600

Upper Alpha10–13 Hz

FIGURE 8.5 Occipital alpha responses. The alpha frequency window is divided into lower and higher alpha bands.

O

q a

d O

q da O

O

O

O

Subject’s Grandmother

d q

a O

O

O

O

d

O

q a

O

O

d O

q

Anonymous Face

O

d

O

q

a

O

a O

Light

FIGURE 8.6 Differences among three types of stimuli in a global qualitative view.

8.5.1 SELECTIVELY DISTRIBUTED ENHANCEMENTS

IN

WHOLE CORTEX

The pictures of the grandmother and the anonymous face elicit varied degrees of enhancements in various frequency windows and recording sites of the whole cortex. Accordingly, the distributed alpha, beta, delta, theta, and gamma systems were selectively activated, again demonstrating one of the important facets of supersynergy and the neurons–brain theory (Chapters 1, 7, and 11). Selectively distributed enhancements of amplitudes in response to the grandmother picture originated from a great number of populations of activated neurons. This involvement of neural © 2004 by CRC Press, LLC

TF1663_book.fm Page 165 Friday, May 7, 2004 8:47 AM

populations in all frequencies probably required more ample and distributed memory activation. We again refer to Fuster (1997, p. 457) regarding the joint activation of posterior and frontal lobes during memory processes. Figure 8.6 illustrates the global qualitative differences among three types of stimuli in order to show that the prediction of Figure 7.1, with the large occipital alpha and large frontal theta, is a good prediction. The increase of alpha responses in frontal areas for face stimulations was not predicted in Figure 7.1. The important message to be taken from Figure 7.1 is the expectation of multiple and topological differentiated oscillatory responses to the grandmother picture in comparison to the light stimulation. Accordingly, the message of the fictive model has been confirmed as a global reality, thus also validating the supersynergy proposal discussed in Chapter 7.

8.5.2 EFFICIENCY OF GRANDMOTHER PARADIGM FOR DIFFERENTIATION OF MEMORY COMPONENTS OR STATES This paradigm can serve as a good strategy to analyze mental disorders involving reduction or complete loss of memory. The refinement of the evaluation of results and possible interpretations cannot be achieved by simple application of ERPs. A complete strategy including different recording areas in all frequency windows, consideration of enhancement, delays, and prolongations is the most useful approach, as required by the neurons–brain theory. All these factors must be analyzed to achieve a deeper understanding of memory activation and reactivated memory components. The components of memory work from a given template employing the entire brain. The components are major activation areas that employ major operating rhythms described in Chapter 5. Frontal theta is the major operating rhythm; frontal theta response plays a major role in differentiating semantic and episodic memories. Alpha activity is the major operating occipital rhythm; it plays a major role differentiating the episodic memory related to the subject’s grandmother and the semantic memory activated by an anonymous face.

8.5.3 DOES ACTIVATION OF LARGER NEURAL POPULATIONS INDICATE REACTIVATION OF EPISODIC MEMORY? As noted earlier, selectively distributed enhancements of amplitudes in response to the grandmother picture originated from activated neural populations. The involvement of neural populations at all frequencies may require greater and higher level selectivity in distribution of memory activation. Fuster (1997, p. 457) presents the idea that the cortical dynamics of evoking episodic memory is identical to that of evoking a familiar stimulus, such as the cue in a delay task. Although the cue is represented in the posterior cortex, the prefrontal cortex is essential for its retention and to propel it toward prospective action. Thus, the prefrontal cortex is important for the sequencing of behavior, thinking, and speech. All three activities require working memory. The grandmother picture portraying a well-known face activates delta and alpha oscillations in posterior areas and parallel theta and alpha oscillations in frontal areas. These phenomena in a way show the joint activation of posterior and frontal areas by processing of episodes. However, our data show that other cortical areas are activated. Moreover, in humans, selected neural populations in the whole brain are also activated (see Chapter 4).

8.5.4 TRANSITION FROM SEMANTIC TO EPISODIC MEMORY: DISTINCTIONS BETWEEN SEMANTIC AND EPISODIC MEMORIES The distinction between episodic and semantic memories is appealing and suggestive of two separate memory systems, but the empirical evidence for corresponding neural systems is weak. Fuster (1995) states that the reason may be that the two implied memory systems are neurally inseparable. The phenomenology is somewhat obvious: The two kinds of declarative memories derive from and © 2004 by CRC Press, LLC

TF1663_book.fm Page 166 Friday, May 7, 2004 8:47 AM

F 4 Theta Anonymous Face

Subject’s Grandmother

µV

–2 0 2

–1000 –800 –600 –400 –200

0 ms

200 400

600 800 1000

0 ms

200 400

600 800 1000

F 4 Theta Anonymous Face

µV

–2

Subject’s Grandmother

0 2

–1000 –800 –600 –400 –200

FIGURE 8.7 Dotted lines show responses to the unknown face; solid lines show responses to the subject’s grandmother’s face. A: early part of the experiment; B: later part of the experiment.

blend into each other without a sharp transition. Even if they could be separated as distinct entities, they maintain dynamic interactions throughout a lifetime. Fuster emphasizes the neural level: “Nonetheless, it seems unwarranted to claim a separate cortical system or process for each.” As stated in Section 8.3.1, measurements with the random set were performed at the beginnings of the experimental sessions and with the regular set at the ends of the sessions for all subjects. Figure 8.7A illustrates the frontal theta responses for the random set (beginning of the session) and Figure 8.7B shows regular set (end of the session) responses as the grand average curves of 28 subjects.* At the start (lower curves), the peak-to-peak amplitude of the frontal theta response to the grandmother picture is approximately 40% higher than the frontal theta response to presentation of the unknown face. At the end of the session (regular set), both responses were almost the same. This is a relevant finding in that it indicates the transition from semantic to episodic memory because the picture of the unknown face was no more anonymous after about 50 repetitions. The test subjects probably learned the characteristics of the unknown face and treated the presentation as an episodic event rather than a semantic one. This finding also supports Fuster’s work (1995) by enhancing the view that semantic and episodic memories may be based on the same brain network. In our model related to the hierarchy of memory states (Section 9.3), we hypothesize, based on observations made during the APLR * During February 2003, additional experiments were performed. We reached a total of 28 subjects. Although no statistical evaluation has been performed yet, the results of almost all subjects fitted to the grand averages. Accordingly, we include this very recent relevant information here.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 167 Friday, May 7, 2004 8:47 AM

alliance, that memory states lack exact boundaries. After learning occurs, newly acquired information induces changes in theta responses. (compare also Sections 3.4.1 and 3.4.4).

8.5.5 IMPORTANCE OF FRONTAL LOBES PROCESSING AND PERCEPTION

AND

OTHER BRAIN AREAS

FOR

MEMORY

Fuster made an important statement in 1995 and 1997 related to the distributed nature of memory processing: From those of general assumptions, especially from distributed nature of cortical networks, it follows that we cannot rightfully consider the cognitive functions of the prefrontal cortex in isolation from those of the rest of the frontal cortex or, for that matter, from the totality of the neocortex and the subjacent anatomical stages of the executive hierarchy. Pursuing methodological neatness, we have often been misled to the localization of cognitive functions that are not localizable. In my opinion, this is true for the so-called working memory, for the so-called ‘central executive’ for spatial memory and for various forms or aspects of attention. All these are indeed cognitive functions within the physiological purview of the frontal lobe, but none of them is localized there [emphasis added].

We also refer to Fuster’s experiments with delay tasks performed with monkeys (1995 and 1997). When temporal integration demands retention of old, reactivated perceptual memory across a time gap, the retention is a joint function of the posterior and prefrontal cortex. Fuster hypothesized that the underlying mechanism was the reverberation of activity through recurrent circuits. This view was clearly supported by the results of the grandmother photo experiments. Occipital theta prolongation, occipital alpha enhancement, and increased posterior delta responses showed that the whole cortex was strongly involved and that the reactions depended on the nature of percepts. 8.5.5.1 fMRI Experiments Related to Distributed Memory in Cortex Section 1.3.1.1 (Chapter 1) explained important functional magnetic resonance imaging (fMRI) experiments performed by Courtney et al. (1997) and Cohen et al. (1997). Courtney at al. (1997) presented human subjects with pictures of human faces, and asked them to recall whether the picture shown was the same or different from one presented 8 s earlier. The authors found that activations in the prefrontal areas correlated most strongly with delay periods, compared with activations in the visual areas that correlated more strongly with sensory stimulation. Based on the fMRI results of Courtney et al. (1997), early extrastriate visual areas demonstrated transient, relatively nonselective responses to complex visual stimuli and later demonstrated transient, selective responses to faces, indicating a more specialized role in the processing of meaningful images. Both extrastriate visual and prefrontal cortical areas demonstrated sustained activity during memory delays, indicating a role in maintaining an active representation of the face in working memory. The results of grandmother photo experiments support the global fMRI results with greater time resolution of EEG oscillations in the range of milliseconds. The distinct roles of frontal and occipital lobes in the recognition of faces emerged also from EEG measurements. Moreover, the multiple frequency and distributed nature of the oscillatory responses clearly showed the involvement of the entire cortex. It is possible that electrophysiological analysis will reveal more about topology and time–frequency responses involved in the recognition of faces in general. 8.5.5.2 Critique of Experiments of Fernandez and Fell We again mention the experiments by the groups of Fernandez et al (2001), Fell et al. (2001), and ˘ and Basar Basar ¸ -Eroglu ¸ (1991). The limbic system is important for the functions of the APLR alliance, but such functions are not localizable because they are distributed throughout the brain. © 2004 by CRC Press, LLC

TF1663_book.fm Page 168 Friday, May 7, 2004 8:47 AM

The results of the experiments with faces clearly show that the occipital area and the 10-Hz responses are involved in remembering processes. The experiments indicate that the interpretations of Fernandez et al. (2001) and Fell et al. (2001) have serious shortcomings. 8.5.5.3 Major Activation Areas of Semantic and Episodic Memories As Figure 8.3a and Section 8.3 show, the oscillatory components of the activated memory are distributed with varied degrees of amplitude, prolongation, and delays in all brain areas under study. Comparing results obtained from different electrode locations provides rich information. Every brain site develops various templates with the superposition of oscillatory responses. In the oscillatory responses to the photos of the grandmother and the unknown person, major activation areas similar to those mentioned in discussions of major operating rhythms (Chapter 5) were involved. Frontal theta appears to be the major operating rhythm. Frontal theta response plays a major role in differentiating semantic and episodic memories. Alpha activity is the major operating occipital rhythm; it plays a major role in differentiating the memory related to the grandmother photo from the semantic memories activated by other faces (Figure 8.3a and b and Figure 8.5). 8.5.5.4 Superbinding and Stryker’s Question about Oscillations The data from the experiments with photos of the subjects’ grandmothers did not favor the idea of simple binding as discussed in the literature and in Chapter 7. The hypothesis of supersynergy and superbinding predicted multiple frequency components distributed selectively in the brain. Our data clearly fit the framework of supersynergy and superbinding in the brain, and the shortcomings of the simple binding theory for explaining complex percepts with only gamma responses have been demonstrated clearly. Stryker (1988; see also Basar ¸ et al., 1997a and 1999a,b,c) described cellular gamma activity by commenting that neurons in the visual cortex activated by the same object tended to discharge rhythmically and in unison. He raised the question, “Is grandmother an oscillation?” According to the results described above, observation of a subject’s grandmother’s picture activated oscillations both in the visual cortices and in other parts of the brain (including frontal, central parietal and occipital areas showing multiple oscillatory responses in delta, alpha, beta, theta, and gamma bands). Every simple input triggers diverse oscillations in selectively distributed areas of the brain. Accordingly, distributed neural groups of all frequencies are involved in the processing of the complex grandmother percept manifested by multiple and selectively distributed oscillations.

8.5.6 DO GRANDMOTHER EXPERIMENTS FAVOR HEBB’S HYPOTHESIS? As early as the 1940s, Hebb (1949) advocated a distinction between memory states: Long-term memory was based on strengthening of links between assemblies of cells within the brain. The short-term memory was regarded as a result of temporary electrical activation of relevant neurons. Do the periods of APLR alliance reflect a phenomenon of temporary learning as proposed by Hebb? During times of APLR alliance, the brain reaches a coherent state in temporal space and achieves greater spatial coherency. The last phenomenon indicates that neural populations are linked to and/or influence each other. Although two important points favor Hebb: (1) increased electrical activity during learning and memory performance and (2) cooperation of distant groups, what happens at the single neural synaptic level cannot be answered with the help of our findings (see Chapters 1, 2, 7, and 11).

© 2004 by CRC Press, LLC

TF1663_book.fm Page 169 Friday, May 7, 2004 8:47 AM

8.6 ARE THE DESCRIPTIONS OF GESTALTS AND EMOTIONS RELATED TO MORE COMPLEX PERCEPTS POSSIBLE? The strategy of the grandmother experiments should be applied in the future to analyses involving more general states of memory, recognition problems, and emotional states. Do differences in oscillatory responses arise after presentation of photographs of well-known faces and less known faces? How will the oscillatory responses of a young man differ after presentation of pictures of his girlfriend, teacher, or chief at the office? Do his oscillatory responses differ when he is shown pictures of a good-looking anonymous girl and his own fiancée? These paradigms can be and should be applied in the future to achieve more refinement in analyzing responses resulting from use of our percept–memory system. Moreover, in experiments intended to describe and analyze thoughts, all properties including superposition, coherence, and entropy should be applied in parallel, as was proposed in Chapter 7 covering supersynergy of electrical events in the brain.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 171 Friday, May 7, 2004 8:47 AM

Part III Memory Function: Models and Theories

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 173 Wednesday, May 12, 2004 8:17 PM

Models 9 EEG-Related of Memory States and Hierarchies 9.1 INTRODUCTION OF A NEW CONSTRUCT ON MEMORY CATEGORIZATION The major aim of this chapter is the introduction of a draft of a memory model based on electroencephalogram (EEG) oscillations. This construct is new; no one has tried to link existing memory descriptions with new results that emerged from brain dynamics tests. Therefore, this new model or fresh construct may need critics, expansions, or possibly reductions — relevant alterations —in the coming years because EEG oscillations are attracting considerable attention in neuroscience. The dynamics of the brain’s electrical signals is oscillatory in nature. Accordingly, the phrase oscillatory brain dynamics is often used in publications. Although the expression is redundant, it has a certain didactical character and therefore it is used in this book. Books written by our research group were always considered workshops for trying new ideas in the field of oscillatory brain dynamics research. This chapter has also a workshop character and it is intended to: 1. Introduce the concept of physiological memory. 2. Emphasize evolving memory incorporating reciprocal actions or reverberations in the attention, perception, learning, remembering (APLR) alliance during working memory processes. 3. Introduce the hierarchy of memory as a continuum. 4. Use longer-acting memory instead of long-term memory (LTM) as a descriptive phrase.

9.2 PHYSIOLOGY OF SELECTIVELY DISTRIBUTED OSCILLATORY PROCESSES A precise description of temporal–spatial functioning of selectively distributed alpha, theta, delta, and gamma systems (see Chapter 6) is not possible for the following reasons: 1. These oscillatory networks have no precise space-boundaries. 2. No exact specification can be stated with regard to the time domain because these systems do not always appear at the same points in time due to delays, prolongations, phases, and time-locking mechanisms (Basar ¸ , 1999; Section 9.3.1). 3. Cognitive functions are not localizable (Fuster, 1995).

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 174 Wednesday, May 12, 2004 8:17 PM

In order to proceed to describe physiologic memory and perceptual memory, we need to know the EEG correlates of sensations, perception, learning, and remembering and we need the experimental background presented in Chapters 3, 4, and 6. Accordingly, at the beginning of this chapter we discuss activation of alpha, theta, and delta systems and illustrate the concepts schematically in order to facilitate the reading of the remainder of this chapter, particularly the sections devoted to hierarchy of memories and whole brain matching. The experimental results related to mutual activation of selectively oscillatory systems play an essential role. Following the descriptive illustrations of selectively distributed oscillatory systems, we will again define various types of memories and concepts introduced in earlier chapters. Reading this material should not be considered a redundant task. We will try to bring together the conventional definitions and the general scheme of the hierarchy of memories. It is not our intention to deny the relevance of conventional memory concepts and definitions. However, the new psychophysiology incorporating oscillatory brain dynamics has become an important area of behavioral neuroscience. For this reason, we will try to link conventional memory models with newly emerged concepts and results.

9.2.1 CONNECTIONS

OF

SENSORY–COGNITIVE SYSTEMS

The connections in the brain are far more complicated than originally thought, and the chain of ideas and proposals presented in this chapter provides useful new insights for memory research. Flohr (1991) described the physiological and anatomical connections in the brain in a simplified and transparent manner (Figure 9.1). 1. Specific afferents from sense organs reach specific thalamic nuclei before going to primary cortical areas. For instance, auditory information is transmitted through the medial geniculate nucleus to the primary auditory area; visual afferents are transmitted through the lateral geniculate nucleus to area 17 of the occipital cortex. 2. Nonspecific afferents reach the cortex from the mesencephalic formation. It has now been established that reticular formation is connected to different nuclei with specific afferent connections. There is a second site where the reticular formation influences the processing of primary afferents: the thalamic relay nuclei. The nucleus reticularis thalami, a thin sheet of neurons, surrounds the dorsal thalamus and inhibits the thalamic relay nuclei. Its control function is, in turn, affected by collaterals of thalamo-cortical pathways, by collaterals from cortico-thalamic projections, and by inhibitory afferents from the mesencephalic reticular formation. Important connections within the cerebral cortex involve the association areas. Primary auditory, somatosensory, and visual fields each project to adjacent unimodal association areas, which, in turn, project to secondary unimodal association fields. The unimodal association areas project to a number of polymodal sensory areas lying in the cingulate gyrus, parietal, temporal, and frontal lobes. The functions of these areas are vaguely described as cross-modal association and synthesis (the cross-modality experiments and the concept of cross-modal association are described extensively in Chapter 6). The polymodal association areas project to the inferior parietal lobe that has been designated a supramodal area. Polymodal and supramodal regions have connections to the limbic system; these connections provide an anatomical substrate by which motivational states influence cortical processing of sensory stimuli. This simplified illustration is tremendously important for understanding additive cognitive information processing of cortical areas. After sensory and event-related stimuli, every sensation in our brain also induces cognitive loading, at least for matching processes. Furthermore, all presented cognitive targets evoke sensations; the respective neural processes are interwoven and © 2004 by CRC Press, LLC

TF1663_C09.fm Page 175 Wednesday, May 12, 2004 8:17 PM

Association cortex

Primary sensory cortex

VIS

Inferior parietal lobe

Polymodal association cortex

AUD

VIS

SOM

AUD SOM

Limbic system

Nucleus reticularis thalami

LG

Thalamic

MG

Relay

VPL

Nuclei

Mesencephalic reticular formation SOM AUD VIS Sensory input

FIGURE 9.1 Flow of information in auditory, somatosensory, and visual pathways, reticular formation, limbic system, and association areas of the cortex. (Modified from Flohr, H., Theory and Psychology, 1, 245, 1991.)

require for final processing at least three neural loops next to purely sensory connections in the simple sensory systems. These loops are (1) secondary connections to the cortex over reticular formation; (2) secondary connections over the limbic system; and (3) connections between association areas within the cortex.

9.2.2 ACTIVATION

OF

ALPHA SYSTEM

WITH

LIGHT

When a human or feline subject has been stimulated visually, large alpha enhancements are recorded in the mesencephalic reticular formation, lateral geniculate nucleus and in parallel, in the hippocampus (HI). Large alpha enhancements also appear in the visual and association cortices. The information flows to the polymodal association cortex and limbic system. The results have also shown large theta enhancements recorded in the thalamus, hippocampus, primary cortex, and association cortices including the frontal lobes (Chapter 4 and Chapter 6). Figure 9.2 shows the hypothetical information flow in the human cortex during the first second following stimulation. Large and dominant alpha responses occur in the occipital cortex. The alpha information then reaches the parietal cortex, and via the known association areas, the frontal cortex, but not the auditory areas. The neurophysiological processes in subcortical structures are explained in Figure 9.3. Large alpha and theta enhancements were observed in the limbic system, reticular formation, and lateral © 2004 by CRC Press, LLC

TF1663_C09.fm Page 176 Wednesday, May 12, 2004 8:17 PM

α

α Human

FIGURE 9.2 Activation of alpha response modules upon visual stimuli. Frontal and parietal responses are not parallel; they are secondary, delayed, and smaller.

geniculate nucleus (see Figure 6.4 and Figure 6.5). No alpha enhancements were recorded in nuclei of the auditory pathways. Although the theta enhancements are major response components in the thalamus and cortex, alpha responses were not recorded in the auditory cortex or medial geniculate nucleus. Electrophysical (EP) testing recorded large enhancements in the occipital and parietal cortices, but generally only small and delayed alpha responses. The hypothetical schemes in Figure 9.2 and Figure 9.3 are also supported by neuro-anatomical findings of connections that travel long distances from the occipital cortex to the frontal cortex and from there to the temporal cortex. (Basar ¸ , 1999).

9.2.3 ACTIVATION

OF

ALPHA SYSTEM

WITH

AUDITORY STIMULATION

Figure. 9.4 and Figure 9.5 explain similar hypothetical flows of information after auditory stimulation. Alpha enhancements dominated the structure in the auditory pathways, i.e., reticular formation and hippocampus. The neurophysiological processes in subcortical structures depicted large alpha response components in the reticular formation, medial geniculate nucleus, and auditory cortex. It is noteworthy that the reticular formation and hippocampus showed large alpha responses, whereas the lateral geniculate nucleus and visual cortex showed only theta responses. Theta enhancements were present in all structures, whether a stimulus was inadequate or adequate. The thinking related to auditory stimulation can be applied to visual stimulation also, but the largest alpha enhancements appeared in temporal, parietal, and occipital areas.

9.2.4 ACTIVATION OF THETA AND DELTA SYSTEMS FOLLOWING COGNITIVE INPUTS We again use Flohr’s illustrations to explain the electrical responses occurring in the region of 300 ms upon presentation of a cognitive target. For the sake of simplicity, we do not consider the delayed and prolonged 10-Hz responses and take into consideration only the dominance of delta and theta responses. Experiments described in Chapter 3 and Chapter 6 confirmed large delta and theta enhancements around 300 ms following stimuli. Figure 9.6 illustrates the process approximately 1 s following a cognitive (i.e., target) stimulus. When an applied stimulation contains an event-related target, the hypothetical flow of information would show large theta and delta enhancements in all areas of the cortex, and probably also in the substructures. Large theta and delta responses and delayed alpha responses would be observed initially 300 ms after stimulation (see Chapter 3, Chapter 4, and Chapter 6). We could also present similar scenarios for experiments in which auditory stimulation was used and the subjects had to pay attention to the third signal (see Figure 6.19 and Figure 6.20). With © 2004 by CRC Press, LLC

TF1663_C09.fm Page 177 Wednesday, May 12, 2004 8:17 PM

θ α

θ

α

Association cortex

Primary sensory cortex

α

θ VIS Polymodal association cortex

Inferior parietal lobe

AUD

VIS

SOM

AUD SOM

Limbic system

θ

Nucleus reticularis thalami

α

LG

Thalamic

MG

Relay

VPL

Nuclei

Mesencephalic reticular formation SOM AUD VIS Sensory input

θ

α

FIGURE 9.3 Flow of oscillatory information in alpha and theta frequency channels in visual pathway, reticular formation, limbic system, and association areas of the cortex. The alpha and theta indicate strong enhancements.

regard to experiments with humans, the third stimulation elicited large enhancements in the theta frequency range, especially in the frontal and parietal recordings. However, these large theta enhancements occurred approximately 300 ms after the stimulation. Although we have no information on changes in theta activity in the human hippocampal recordings for this paradigm, we hypothesize the existence of a theta response circuit including hippocampus, frontal, and parietal areas of the cortex. In all experiments described in this book, tasks involving focused attention produced theta enhancements. See also the experiments on cat brains cited in Chapter 4 and recent publications by Fell et al. (2001). © 2004 by CRC Press, LLC

TF1663_C09.fm Page 178 Wednesday, May 12, 2004 8:17 PM

FIGURE 9.4 Activation of alpha response modules upon auditory stimuli. Frontal responses are not parallel; they are secondary, delayed, and smaller.

9.2.5 NONSPECIFIC INTERACTIONS Nonspecific interactions in the brain merit important consideration in this chapter because of the cognitive processes in which the central nervous system (CNS) is involved. Many animals have quiescent or inattentive states. In vertebrates, a neural system associated with arousal to the attentive state (part of the reticular activating system) receives inputs from many senses. The arousal system influences neurons in most brain nuclei that deal with specific sensory information. Thus midbrain, thalamic, and cortical neurons primarily responsive to visual stimuli may be activated or modulated by auditory or somato-sensory stimuli via nonspecific arousal mechanisms (see Basar ¸ , 1999, Chapter 2). Cross-modality response experiments demonstrate that the differences between responses to adequate stimulation and to inadequate stimulation were marked in the alpha frequency range and less marked in the theta range (see Figure 9.3 and Figure 9.5). This hints at a possible special role of the alpha response in primary sensory processing. This means that alpha responses are also correlated with cognitive processes, but the auditory alpha response is not ample in the occipital cortex; accordingly, the APLR-related alpha response of the occipital lobe is activated during visual processing. The reverse applies in the primary auditory areas (temporal lobe). The APLR-related alpha response is also activated during visual processing. In contrast, theta responses appear to be less dependent on whether a stimulus is adequate. This is compatible with a predominantly associative–cognitive role of the theta response as hypothesized ˘ et al., 1992; above and published earlier (Basar ¸ et al., 1991; Demiralp and Basar ¸ , 1992; Basar ¸ -Eroglu Chapter 6, this volume). Hartline’s statement (1987) about the visual cortex, which is considered to be involved exclusively with vision, noted that about a third of the neurons in areas 17 through 19 (in cats) appeared to be responsive to sound and visual input. Briefly, it was hypothesized that adequate stimulation of a cortical area (e.g., visual stimulation/visual cortex) is followed by a response with a high amplitude alpha component and a theta response of small amplitude. Inadequate stimulation elicits responses with less difference in the theta and alpha responses, i.e., in the case of inadequate stimulation, the theta response is less reduced than the alpha response. In other words, the alpha response appears to be more dependent on the adequacy of the stimulus than the theta response. Hartline (1987) dealt with multisensory representation of space in the CNS and stated, “The visual cortex is thought to participate in processes that serve to the recognition of objects or patterns. Though this part of the brain is considered to be involved exclusively with vision, in cats about one third of the neurons in area 17 (striate cortex), 18, and 19 are reported to be responsive to sound as well as to visual input.” On the other hand, the sensory stimulation of second order usually © 2004 by CRC Press, LLC

TF1663_C09.fm Page 179 Wednesday, May 12, 2004 8:17 PM

θ α

θ

α

Association cortex

Primary sensory cortex

α

θ VIS Polymodal association cortex

Inferior parietal lobe

AUD

VIS

SOM

AUD SOM

Limbic system

θ

α

Nucleus reticularis thalami

LG

Thalamic

MG

Relay

VPL

Nuclei

Mesencephalic reticular formation SOM AUD VIS Sensory input

θ

α

FIGURE 9.5 Flow of oscillatory information in alpha and theta frequency channels in the auditory pathway, reticular formation, limbic system, and association areas of the cortex. The alpha and theta indicate strong enhancements.

reaches the cortex over association areas (Shepherd, 1988; Figure 9.1). Based on these findings, it is conceivable that the responses in the lower frequency ranges (theta and delta) may reflect the responsiveness of various brain areas dealing with association processes involved in global associative cognitive performance. The first tentative interpretation of our results led us to formulate that the alpha response component mostly handles the primary sensory processing of signals, whereas the theta and/or slower responses are mostly involved in association and cognition (Basar ¸ et al., 1991). © 2004 by CRC Press, LLC

TF1663_C09.fm Page 180 Wednesday, May 12, 2004 8:17 PM

δ θ δ θ

Association cortex

α

θ

Primary sensory cortex

VIS Polymodal association cortex

Inferior parietal lobe

AUD

VIS

SOM

AUD SOM

Limbic system

δ θ

Nucleus reticularis thalami

LG

Thalamic

MG

Relay

VPL

Nuclei

Mesencephalic reticular formation SOM AUD VIS

θ

α

Sensory input

FIGURE 9.6 Flow of oscillatory information in theta and delta frequency channels in the visual (VIS) and auditory (AUD) pathways and limbic system. The theta and delta indicate strong enhancements.

This interpretation is also supported by results obtained with different paradigms. Especially slow frequencies contribute to differences between evoked potentials (EPs) obtained in an omitted stimulus paradigm and EPs recorded in a session without cognitive load. In a time prediction task, selective averaging of responses to the last stimulus before omission showed increased delta–theta amplitudes (Demiralp and Basar ¸ , 1992). This is noteworthy because the increase was most prominent in frontal and parietal electrodes, which are closely related to association areas of the brain (see Chapter 6).

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 181 Wednesday, May 12, 2004 8:17 PM

9.3 HIERARCHICAL CATEGORIZATION OF DIFFERENT LEVELS OF MEMORY 9.3.1 FUSTER’S VIEW

OF

MEMORY NETWORKS: A MILESTONE

IN

NEUROSCIENCE

According to Fuster (1995), the cognitive functions of the frontal cortex and other parts of the neocortex consist of activation and processing within and between networks of representation or memory networks. These networks are widely distributed and highly specific, defined by their synaptic structure and connectivity. Thus, the memory code is relational and all memory is associative. Further, Fuster states that the cortical networks of memory, by any anatomical definition, extend across modules and areas. In our opinion, one of Fuster’s most important concepts is the fact that memory networks overlap and are diffusely interconnected with one another. Thus, a single neuron or group of neurons anywhere in the cortex can be part of many networks and thus many memories. This is why it is virtually impossible by any method to localize a memory. The networks of executive or motor memory are distributed in the cortex of the frontal lobe, and like the perceptual networks of posterior post-rolandic cortex, are hierarchically organized. This statement is in accordance with the described functioning of selectively distributed oscillations that cannot be exactly localized.

9.3.2 TENTATIVE MODEL RELATED

TO

EEG ACTIVATION

The memories of all living beings incorporate survival functions ranging from simplest reflexes to higher nervous activities including episodic and semantic memory processes. The escape of Aplysia illustrated in Figure 9.8 is a typical example of a simple survival action involving a combination of reflex, decision, and movement. Accordingly, based on this hierarchy of memory functions, we can categorize all levels of living and survival processes and tentatively introduce three different levels of memory states based on the scheme of Figure 9.7. (The dynamic character of Figure 9.7 is presented in our home page [http://braindynamics.deu.edu.tr/basar.htm].) The reader can follow the transition between memory states there. Exact boundaries between these levels cannot be described. Level I includes inborn (built-in) memories, i.e., networks that are active during retrieval processes that are genetically coded and are usually not altered or altered little. Level II includes dynamic memory states that are activated and interactive with integrative functions. Level III includes longer-term memory activation.

9.3.3 INBORN (BUILT-IN) NETWORKS (LEVEL I) In order to establish a hierarchical classification of inborn memories we propose six levels of inborn memory that can be classified as parts of the physiological (or fundamental–functional) memory: 1. 2. 3. 4. 5. 6.

Simple reflexes Complex reflexes Stereotypic fixed action patterns Phyletic memory Feature detectors Living system settings

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 182 Wednesday, May 12, 2004 8:17 PM

Reflexes PHYSIOLOGICAL MEMORY (PERSISTENT)

I

A. Reflexes B. Complex reflexes

e.g., simple Achilles reflex

Complex reflexes Stereotypic fixed action pattern

Several segments are involved

Feature detectors

C. Phyletic memory D. Feature detectors E. Living systems settings F. Stable fragments of newly learned percepts (semantic and episodic)

Il Learning in entire CNS: evolving memory

Ill

Obligatory reactions and/or reflexes

Stable for longer periods

b) Perceptual memory Simple percepts (genetically coded or in-built)

Learned percepts from procedural or working memory

c) Motor memory

Dynamic working memory APLR (attention perception, learning, and remembering) (matching + learning, search and learning, process)

e.g. Simple echoic memory Simple iconic memory Electroception

Persistent memory, working memory and longer acting memory are produced also in an inseparable way due to standing information exchange and matching. What is learned during the working memory process is often transferred to longer acting memory and also to persistent memory. Examples: 1) Development of a waves in expectation experiments 2) Development of responses in P300 experiments. Theta appears and is phase locked as the experiment processes. Basar and Stampfer papers: P300 develops at the end of the experiments (compare to Basar and Stampfer 1985).

Learning during psychophysiological processing e.g., motor learning, procedural memory, evolving memory

Longer acting memory system e.g., episodic + semantic

What is learned following a dynamic process can be transferred to persistent memory

Quasi-stable

Persistent (static)

Dynamic

FIGURE 9.7 (See color insert following page 200 for a better understanding of this illustration.) Draft of a detailed scheme of memory levels, hierarchy, and transitions between memories. Level I (top) is persistent memory, of which the essential part is physiological or fundamental memory composed of inborn (or built-in) memories including monosynaptic simple reflexes, (e.g., Achilles reflex); complex reflexes involving multiple segments of the spinal cord, stereotypic fixed action patterns (e.g., flight reaction of aplysia); and phyletic (echoic and iconic) memories and electroception ability. Living systems settings such as blood pressure, smooth muscle reaction, and heart beat belong to physiological memory as does motor memory. The yellow background shows static components; such memory types are persistent. The green-yellow background is associated with quasistable longer-term memory states. Newly learned percepts acquired after activation of procedural or working memory are quasi-stable and acquired during life. They are not inborn, but exist for long periods. Over time, they may be replaced or forgotten. Perceptual memory is between Level I and Level II. Level II covers dynamic processing and the working memory state. Dynamic changes in the APLR alliance are strongly associated with evolving memory. Following motor learning or procedural memory, new engrams can be created and are then transferred to perceptual or motor memory as indicated by an arrow. Following learning during procedural or evolving memory states, newly memorized information is transferred to longer-term memory (Level III). Semantic and episodic memories are categorized at this quasi-stable level indicated by gray and yellow. Newly learned material following a dynamic process is sometimes transferred to persistent memory (Level III to Level I) as the arrow indicates. Persistent memory is not explicitly positioned; it appears in Figure 9.12 as a separate block. © 2004 by CRC Press, LLC

TF1663_C09.fm Page 183 Wednesday, May 12, 2004 8:17 PM

FIGURE 9.8 Escape response of Aplysia californica to starfish Astrometis sertulifera. On contact with the starfish (A), the animal withdraws (B), turns away (C), and escapes with rapid pedal waves (D through F). (Modified from Byrne, J.H. et al., Journal of Neurophysiology, 41, 402, 1978.)

9.3.3.1 Reflexes Memory, learning, and integrative functions of the CNS arise from operations of networks of adaptive elements. Physiologically, the elements are neurons and the networks may be considered reflex pathways that link inputs and outputs by way of specifiable transforms. Organisms contain many reflex systems. Some are as simple as those of the spinal cord described by Sherrington; others are much more complex. L.P. Pavlov (1927) and B.F. Skinner (1938) gained insights into reflex processes supporting learning and memory in the last century. They appreciated the fundamental significance of the reflex as a mediator of behavior. Sherrington (1948) provided physiological documentation of the neural pathways that connect stimulus with response and showed that a knee jerk elicited by tapping the tendon below the patella is mediated by two types of neurons. 9.3.3.2 Stereotypic Fixed Action Patterns Figure 9.8 shows an example of a fixed pattern of escape actions employed by Aplysia californica in contact with a starfish. 9.3.3.3 Phyletic Memory and Oscillatory Response Codes Fuster (1995) states that it is useful for understanding the formation and topography of memory to think of the primary and sensory motor areas of the cortex that can be called the phyletic memory © 2004 by CRC Press, LLC

TF1663_C09.fm Page 184 Wednesday, May 12, 2004 8:17 PM

or memory of the species. The primary sensory and motor cortices may be considered funds of memories acquired by a species through evolution. We can call the formation a memory because, like personal memory, it represents acquired and stored information that can be retrieved (recalled) by sensory stimuli or a need to act. In the structures of primary systems, phyletic memory contains the innate capacity to recall and respond to elementary features of sensation and movement that are common to the repertoires of all members of the species. As noted in Chapter 6, alpha, theta, and gamma responses are also manifestations of phyletic memory because they are inborn and possibly wired responses. This fact is clearly indicated in the last column of the Table 7.1 which describes the hierarchy of activated memories. 9.3.3.4 Feature Detectors The primary features of stimuli such as heat, force, light, sound, and chemical substances are selectively transduced at the peripheral ends of sets of sensory (afferent) nerve fibers. Different groups of these sensory fibers respond selectively at lower thresholds than others do to different forms of impinging energy. This tuning is often called feature detection and is accomplished during evolution of species by the development of specific transducer mechanisms for different forms of energy, either in the nerve endings or in complex sensory organs in which the afferent fibers terminate. Examples are the mammalian retina and cochlea and the pressure transducers of primate hand skin (see the Appendix 1 glossary and the statements of Sokolov, 2001 and Mountcastle, 1998 in Chapter 7). Living system settings are ensembles of detectors and all types of mechanisms that serve to maintain survival functions in living systems, for example, normative values of blood pressure, respiratory rhythms, cardiac pacing, and body temperature. Mechanisms that are important to maintain healthy life qualities within certain limits should be categorized as persistent memory components because damage to these settings strongly affects higher nervous activities and all levels of memory activation. 9.3.3.5 Living System Settings Living system settings are ensembles of detectors and all types of mechanisms that serve living systems to maintain survival functions such as normative values of blood pressure, respiratory rhythms, cardiac pacemakers, and body temperature. Such mechanisms that are important to maintain the body within the limits of healthy life qualities should be also categorized into the level of persistent memory since damage to these settings strongly affects higher levels of nervous activity and all levels of memory activation. Gebber et al. (1995) rewiewed a series of articles from their laboratory on the 10-Hz rhythm in sympathetic nerve discharges of cats and offered a hypothesis on its functional significance: The rhythm is ubiquitous to the discharges of sympathetic nerves with different cardiovascular targets and arises from a system of coupled nonlinear brain stem oscillators, each of which has a selective relationship with a different portion of the spinal sympathetic outflow. The 10-Hz rhythmic discharges of sets of sympathetic nerves are differentially related and the pattern of relationships in one experiment can be the reverse of that in the next. The authors hypothesize that nonuniform and dynamic coupling of the central circuits controlling different sympathetic nerves is the basis for the formulation of complex cardiovascular response patterns that include differential changes in regional blood flows.

Are they tuning and matching effects between cognitive 10-Hz oscillations and sympathetic discharges? This question may be answered in the future. However, the fundamental findings of Gebber’s group indicate that the physiological settings of the circulatory system, phyletic memory networks (sensory alpha responses), and brain stem responsiveness share common oscillatory

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 185 Wednesday, May 12, 2004 8:17 PM

characteristics in the 10-Hz frequency range. A link between cognitive and vegetative processes may possibly be supplied with general 10-Hz oscillations.

9.3.4 PHYSIOLOGICAL

OR

FUNDAMENTAL MEMORY

The physiological memory is the ensemble of memories or living system settings necessary for all vital functions: Physiological memory is a fundamental memory; the functioning of the CNS is impossible without it (Figure 9.7). It is genetically coded and differentiated among species. It comprehends sensory memories as echoic and iconic memories and comprehends reflexes, stereotypic action patterns, and perceptual memory. Based on results cited in Chapter 4, Chapter 6, and Chapter 7 and Figure 9.3, Figure 9.5, and Figure 9.6, sensory and perceptual memory machineries are manifested with multiple oscillations in the alpha, beta, gamma, theta, and delta frequency bands. According to Fuster (1995) the number of networks equals the number of memories and, in theory, their numbers are infinite. All living systems, depending on level of evolution and type of species, have several types of reflexes and also differentiated forms of sensory memory. To make this differentiation understandable, we will mention a few examples. The hagfish has no form (pattern) vision, whereas the ray has vision; hagfish and lampreys do not have any acoustic abilities. Lampreys have good vision and elasmobranches (rays) have electroception.* Mammalians do not have electroception ability, but have well-developed hearing and vision. Reflexes vary from the simplest Achilles’ to stereotypical behavior reflexes. All these types of fundamental reflexes are components of physiological functioning that are extremely important for survival. [For information about the physiology of ray electroception, see Chapter 7 in Basar ¸ (1999) and Bullock and Basar ¸ (1988).] We recently reviewed several classifications devised by distinguished memory investigators in order to find a strategy to explore the electrophysiology of distributed memories. We concluded that recording of event-related oscillations (EROs) (EEGs and event-related potentials [ERPs]) was a good method to attack the problem both at the neurophysiological and psychological levels because oscillations have similar frequency codes in the whole brain (Basar ¸ , 1999; Basar ¸ et al., 2000) and because cognitive inputs evoke oscillatory responses similar to sensory oscillatory responses in the same frequency channels of EEGs. Only prolongation, amplitudes, and topological distributions are different. Cross-modality experiments explained in Sections 9.2.2 through 9.2.4 are important for explaining built-in sensory memories. Accordingly, interactions in sensory cognitive processing are rich and separation of these processes is almost impossible. Remembering and memory are manifestations of various and multiple functional processes, depending on the complexity of the input to the CNS. The electrical response to a simple light flash is based on simple memory processes at the lowest hierarchy order. Figure 9.2 and Figure 9.5 indicate that sensory cortices and thalamic nuclei react to sensory stimulation with alpha responses. This means that occipital alpha response is an inborn or built-in response for iconic memory, reacting to simple light stimulation; conversely, the alpha response of

* Private communication by T.H. Bullock (June 11, 2002): T. H. Bullock consulted Glenn Northcutt — a real authority on fishes, including lampreys (e.g. petromyzon) and the other branch of the cyclostomes or agnathans, namely the hagfish (e.g. myxine, bdellostoma), which have extremely reduced eyes and no lens. Northcutt writes: “To the best of my knowledge, hagfish has no pattern (form) vision and almost certainly can only detect flux. Lampreys on the other hand have a large well differentiated eye and I suspect fairly good vision but I don't know of a single behavioral study on this. Of course shark (and ray) visual abilities have been well documented and reviewed a number of times.” To the best of my knowledge, there is no evidence that the ears of hagfish and lampreys have any acoustic abilities at all. Bullock writes further: “I tried to get AEPs (auditory evoked potentials as well as electroreceptor evoked potentials (EEPs) from both, without any sign of AEPs from either group but good VEPs from lampreys, none from hagfish. Rays and sharks show AEPs and EEPs with a wide range of sensitivity between species. Elasmobranchs have good VEPs, probably with specialization between species.”

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 186 Wednesday, May 12, 2004 8:17 PM

the auditory cortex is a built-in physiological response to simple auditory stimulation. Again, the early gamma response is also a component of phyletic memory response. 9.3.4.1 Changes of Sensory Memory: Spontaneous and Evoked Alpha Activity at Occipital Sites in Three Age Groups The visual cortex needs visual experience during a critical period of early life in order to achieve its full functional development. In the absence of critical experience, the cortex remains defective. As a consequence, normal vision cannot be attained unless the subject has considerable training. We included in Chapter 5 an example related to aging. The literature contains many results that indicate changes of oscillatory responses after acquiring new experiences. The example in Figure 5.6 explains that after learning and training early in life, alpha activity and alpha responsiveness do change. In early life, there are no alpha responses to simple light stimulation. Accordingly the functioning of networks sensitive to light and sound (Basar ¸ , 1998) is susceptible to changes. The phyletic memory and the physiological memory are not completely stable throughout a lifetime. During a memory process — a short- or longer-term one — the perception of a sensory input is matched with information already stored in the neural tissue. When a simple light evokes alpha and gamma responses in the cortex, elementary oscillatory responses are associated strongly with several memory processes at different hierarchical levels (see section on alpha responses in Chapter 6). The topology of the memories, depending on the modality of the input, must be different (see examples cited in sections on cross-modality experiments and measurements in cortical and subcortical structures). During perception, the brain performs a matching process with inborn sensory memories. Even complex percepts are interwoven with simple sensations. Since processing of simple sensations and complex percepts is inseparable, the physiological memory and perceptual memory must be also inseparable and constitute a single entity. We propose the existence of an alliance of physiological and perceptual memories. This consideration has a crucial consequence. Although physiological and perceptual memories are presented separately in Figure 9.7, they overlap during functional processing. Based on these facts, physiological and perceptual memories do not constitute a clear-cut functional hierarchy or absolute separation: The illustration should, in reality, show a continuum of physiological and perceptual memory. It should not present an absolute functional hierarchy or separation. We will tentatively show in Chapter 10 that homogeneous functional processes depict continuities for the transitions among functional states, in accordance with the functioning of selectively distributed oscillations described earlier (see Fuster’s statement in Section 9.3.1). By discussing the elements of Figure 9.7, we will show that it is impossible to define most memory types or levels as fully separated individual entities with rigid (frozen) boundaries in times and topological spaces as in the case of perceptual memories. On the contrary, learned reflexes, simple sensory cognitive processes, and all types of memories have dynamic features. They evolve steadily as life continues. The human semantic memory is altered or extended after it is subjected to a learning process. Studies demonstrating these dynamical processes have rarely been performed; the experiments described in Chapter 8 are such experiments. Therefore, the results and interpretations must be considered preliminary information. Accordingly, multiple distributed memories cannot be treated in detail and perfect classifications of all levels of distributed memories cannot be yet provided and may never be achieved. Therefore, Figure 9.7 presents a proposal related to dynamic and evolving memory in accordance with the workshop character of this chapter.

9.3.5 WORKING MEMORY (LEVEL II) During processing of many complex tasks, it may be necessary to hold information in temporary storage before the task can be completed. The system used for this purpose is referred to as working © 2004 by CRC Press, LLC

TF1663_C09.fm Page 187 Wednesday, May 12, 2004 8:17 PM

memory (Baddeley, 1996). Working memory is a temporary ad hoc activation of an extensive network of short- or long-term perceptual components that are, like other perceptual memories, retrievable and expandable by a new stimulus or experience. Fuster (1995) states that working memory has the same cortical substrate as the kind of short-term memory traditionally considered a gateway to long-term memory. Based on these functional statements, a simple or complex light stimulation or a combination of light stimulation and a task or event (see checkerboard experiments), should evoke oscillatory responses with different time hierarchies (Figure 6.13, Figure 6.15, and Figure 6.22). Functional or oscillatory network modules are distributed both in the cortex and throughout the whole brain (Basar ¸ , 1998b; Chapters 4 and 6, this volume). The detection of a target signal during a P300 type of experiment requires also a type of working memory. The subject must sustain knowledge related to the target (nature of the target, frequency, shape, color, or other factor) during the experiment. The linkage of P300 amplitude and latency measures and working memory processes has been already established (Sanquist et al., 1980; Howard and Polich, 1985; Pratt et al., 1989; Fabiani et al., 1990; Scheffers and Johnson, 1994). The matching process after the detection of the target should rely in this case on working memory or the success of this type of memory P300 versus N100. Based on Chapters 3, 4, and 6, during working and evolving memory states, alpha, delta, and gamma oscillations are differentially activated. An ERP is a compound neuroelectric signal that is rich in functional information (Bullock, 1993) and related to a large spectrum of activities ranging from single percepts to complicated memory processes. Furthermore, in the analysis of integrative brain functions it is indispensable to consider a specific ERP in a given brain structure and take into account that distributed ERPs are interrelated due to the apparent strong parallel processing in the whole brain (see Chapter 6, this volume). Accordingly, it is necessary to analyze the responsiveness of the entire brain in order to understand a specific function manifested by neuroelectric activity of a given structure. For example, when we consider or analyze cognitive processes, the most marked ERPs usually are recorded in fronto-parietal areas or in various association cortices. However, it is necessary to take into account recordings from other areas as well, e.g., from sensory cortices that may indicate parallel processing (Basar ¸ and Schürmann, 1994; Basar ¸ , 1998a,b). 9.3.5.1 Perceptual Memory Several types of analysis categories are crucial in the functional interpretation of ERPs including analysis of the stimulus itself: What can a stimulus evoke in the brain? It can evoke simple percepts, complex sensory percepts, bimodal percepts, memory-related functions, and other activities. According to Fuster (1995, p.10), perceptual memory is acquired through the senses. It comprises all that is commonly understood as personal memory and knowledge, i.e., representations of events, objects, persons, animals, facts, names, and concepts. In the hierarchy of memories, Fuster further notes that memories of elementary sensations are at the bottom; abstract concepts, although originally acquired by sensory experience, become independent from it in cognitive operations and appear at the top. This means that perceptual memory belongs partially to built-in memory. It evolves throughout life by adding new stored percepts and becomes richer in stored information. Therefore, perceptual memory comprises elements of phyletic and fundamental memories and also elements of semantic and episodic memories. Again, it is not possible to define exact boundaries in the hierarchies of perceptual, semantic, and episodic memories. In order to process percepts. the brain needs both built-in networks of elementary sensations and new information obtained throughout life. We include perceptual memory in the category of quasi-stable memories (Figure 9.7) because, according to the descriptions above, perceptual memory is partly inborn. Complex percepts acquired during life may not be completely stabilized and are quasi-stable. For example, the grandmother percept is shaped by several events during life and is not inborn (see Chapter 8). Section 8.2.2 through Section 8.2.4 showed that alpha, gamma, and © 2004 by CRC Press, LLC

TF1663_C09.fm Page 188 Wednesday, May 12, 2004 8:17 PM

theta responses to simple light or simple sound may be generated from built-in networks because cross-modality experiments demonstrated that alpha sensory response was evoked only by adequate stimuli in the cortex and thalamus (Section 8.2.2). Perceptual memory is distributed not only in primary areas of the cortex and thalamus, but also in the whole brain based on recordings of alpha responses in the whole brain (Figure 9.2 and Figure 9.6). Based on these considerations, we classify simple perceptual memory as part of physiological memory. However, complex percepts that show individual differences are less linked to the ensemble of physiological memory. Complex percepts overlap or are interwoven with ensembles of subsets of evolving memory. Recognition and evolving memory are dynamic processes that are manifested by multiple oscillations. A reason for mentioning working memory in this section is the fact that this memory function has been linked to the prefrontal region of the brain, i.e., part of the dorsolateral frontal lobe that comprises the anterior convexity of the cerebral hemispheres (Goldman-Rakic and Friedman, 1991; Fuster, 1991). However, most studies of working memory have been done in nonhuman primates, making it difficult to draw direct neuroanatomical comparisons to humans. Nonetheless, strong evidence from nonhuman primates indicates that the prefrontal region plays a crucial role in working memory, and it is likely that this relationship will, at least to an extent, apply to humans as well (Baddeley, 1986, Chapter 10). Based on experiments cited in Chapter 8, all association areas and sensory cortices are involved during working memory processes, although frontal memory-related theta oscillations seemingly play a major role.

9.3.6 INCORPORATION OF OSCILLATORY CODES IN PHYSIOLOGICAL MEMORY CONSISTING OF PHYLETIC, SENSORY AND PERCEPTUAL MEMORIES Figure 9.7 tentatively shows that physiological memory incorporates phyletic, sensory, and perceptual memories. Are these memory types or activated memory states interwoven with similar frequency codes? Several sections of this book indicate that the mammalian brain responds to simple visual and auditory stimuli with selectively distributed alpha, theta, beta, gamma, and delta responses (Table 7.1; Figure 9.3, Figure 9.4, Figure 9.5, and Figure 9.6). As we explained in Chapter 6, alpha, theta, and gamma responses are also manifestations of phyletic memory because they are inborn and possibly wired responses. This fact is clearly indicated in the fourth column of Table 7.1 which lists the tentative hierarchy of activated memories. According to the results cited in Chapter 4, Chapter 6, and Chapter 7 and the flow charts (Figure 9.3, Figure 9.5, and Figure 9.6) in this chapter, sensory and perceptual memory machineries are manifested by multiple oscillations in the alpha, beta, gamma, theta, and delta frequency bands. Our empirical evaluation suggests that all these memory types are interwoven and/or tuned with the frequency codes of EEG oscillations. This may facilitate the transition between memory states and communication. Strong links or alliances of all integrative functions in the brain could be enhanced.

9.3.7 WHAT

IS

MOTOR MEMORY?

According to Fuster (1995), motor memory consists of representations of motor action in all its forms, from skeletal movement to spoken language. Motor memory, like perceptual memory, is evoked and acquired through the senses, and once acquired, it is largely presented in the neocortex and frontal lobe. The most automatic and firmly established aspects of motor memory are represented outside the neocortex. The basal ganglia and cerebellum are the most fundamental structures related to motor memory. Perception and motor action are interrelated, and both are parts of many representational networks.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 189 Wednesday, May 12, 2004 8:17 PM

9.3.8 DEVELOPMENT

OF

PROCEDURAL MEMORY

THROUGHOUT

LIFE

According to Baddeley (1986), procedural memory refers to the acquisition of skills, whether perceptual–motor, such as riding a bicycle or driving a car, or cognitive, such as skilled reading or problem solving. Skills clearly comprise an important area of learning and represent archetypal examples of procedural learning — learning how rather than learning that. Baddeley notes that skills can be divided into two types: continuous (each component of the skill serves as a cue to the next, as in cycling or steering a car) and discontinuous (a series of discrete stimulus–response links are involved, as in typing). In general, continuous skills seem to involve little or no forgetfulness, whereas forgetting clearly occurs during execution of discontinuous tasks. The basal ganglia have been linked to various forms of nondeclarative memory, particularly those types of memory that depend on a motor act for their realization or what has been termed procedural memory (e.g., riding a bicycle or skating). The cerebellum, another motor-related structure situated behind the brainstem at the base of the brain, participates along with the basal ganglia in many types of procedural learning and memory operations (Glickstein, 1993; Thompson, 1986 and 1990; Chapter 4, this volume). As shown in Figure 9.7, perceptual memory and procedural memory belong to the category of quasi-stable memories. It should be noted that oscillatory response activity changes during aging, as described in Chapter 4.

9.4 DYNAMIC MEMORY IN WHOLE BRAIN: MEMORY STATES INSTEAD OF MEMORIES 9.4.1 ALPHA, THETA,

AND

DELTA OSCILLATORY PROCESSES

DURING

APLR

Attention, perception, learning, and remembering (APLR) are simultaneous and interwoven processes, as described in other chapters in relation to EEG oscillations. We again refer to Figure 9.9 to emphasize that evolving memory is accompanied by the dynamic shaping of theta oscillations. The interplay between delta and theta activities was completely changed at the end of the experiment. The dynamic shaping of the alpha oscillations during learning and dynamic memory experiments is clearly shown in Figure 3.8 through Figure 3.11. Per Figure 9.6, during induced cognitive behavior, theta and delta response susceptibilities become dominant. Hayek (1952) presented his concept of cortical memory network in the context of the main topic, which was not memory. Significantly, his topic was perception as the source of and product of memory. We introduced the APLR alliance as a consequence of measurements. The alliance is not a theoretical construct, but it is derived from empirical evidence.

9.4.2 ARE DYNAMIC EEG TEMPLATES CREATED DURING PROCESSING OF THE APLR ALLIANCE? DO THEY BUILD (VIRTUAL) SHORT-TERM STORAGE OF NEWLY LEARNED MATERIAL? Learnable sequences described in Section 3.4.4 showed that oscillatory activity did change when subjects acquired new knowledge in the course of learning tasks.* These changes are manifested by newly created EEG templates (Figure 3.8, Figure 3.11, and Figure 9.9). Such changes are probably consequences of matching processes and are at least temporarily stored in the brain. During processes of working memory, oscillatory components also work, often in a superimposed

* Learnable sequences are time intervals in which stimuli alternated in some predictable order produced smaller P300 responses than irregular sequences that were unfamiliar and unpredictable. The findings of Basar ¸ and Stampfer (1985) and Donchin et al. (1973) suggest that feed forward from memory can influence P300 amplitude. If memory correctly predicts the input, the P300 response is reduced: if not, a mismatch is registered and a large P300 wave develops.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 190 Wednesday, May 12, 2004 8:17 PM

FIGURE 9.9 (A) Filtered EEG–ERP epochs following randomly applied target tones are filtered in the 3.5to 8-Hz frequency band. Sweeps 2 through 80 are shown. The prolonged and enhanced 3.5- to 8-Hz (theta) deflection is observed after Sweep 49. (B) Filtered EEG–ERP epochs to repetitively applied rate target tones. Filter limits are 8 to 13 Hz. Two groups of sweeps (3 through 13 and 65 through 79) are illustrated separately to show relevant changes in EEG and ERP activities. (Modified from Basar, ¸ E. and Stampfer, H.G., International Journal of Neuroscience, 26, 161, 1985.)

way.* The oscillations work in parallel in the whole brain as coherence analysis has indicated (Figure 6.25). Attention, perception, learning, and memory work in synergy and in a reciprocally activating and interwoven way as the data demonstrate (see Chapter 7). This description, which is experimentally founded, indicates that attention, perception, learning, and dynamic memory are inseparable functions representing a single entity rather than separable processes. In the processing of complex functions, the brain waves work longer, have multiple components, and accordingly, during complex signal processing, delay or prolongation of the oscillatory activities is observed The difficulty of recognizing a target in the oddball paradigm or in the omitted sound paradigm is manifested by delay and prolongation (see Section 3.3, Chapter 6, and Chapter 8).

9.4.3 ARE ALL BRAIN FUNCTIONS LINKED

WITH

MEMORY?

According to Goldman-Rakic (1996), working memory is the ability to hold an item of information transiently in the mind in the service of comprehension, thinking, and planning. Working memory encompasses both storage and processing functions. It serves as a workspace for holding items of information that can be recalled, manipulated, and/or associated with other ideas and incoming information. As noted in Chapters 3, Chapter 5, and Chapter 6, experiments with cognitive loading show that oscillatory prestimulus and poststimulus activities reflect constant work of the brain in progress. Alpha activity occurring prior to expected cognitive targets goes to a state of order. The time course * Superposition principle in Chapter 6 and Chapter 7.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 191 Wednesday, May 12, 2004 8:17 PM

of the alpha activity is ordered or aligned; amplitudes of oscillations are increased during cognitive demands and memory loads. Further, the alpha oscillations are phase-locked to the expected target signals (Section 3.2). If we combine these results with the statement of Goldman-Rakic (1996), our main scope is considering all brain functions and memory as a a single entity. Perhaps we should consider a memory and brain function alliance as a more general categorization of the APLR alliance and say that all brain functions are memory-linked or all memory is function. This means that the qualitative and quantitative properties of the alpha activity changed and evolved during experimental sessions, thus giving the message that alpha activity is involved with constant work that reflects multiple parallel and reverberating components. Attention, perception, learning, and memory are interrelated and interwoven. The dynamic features of our experiments (Chapter 3, Chapter 4, and Chapter 6) give the impression that all these functions evolve together with a unified trace running along the time axis. Procedural memory, working memory, and motor learning are categorized at Level II in Figure 9.7. Motor memory is more difficult to categorize solely on one level, and appears in the illustration at the junction of Level I and Level II.

9.5 COMPLEX MEMORY OR MULTIPLE MATCHING: EVOLVING MEMORY AND APLR ALLIANCE By “memory” I will mean any set of events that makes available to an organism something of a situation after that situation no longer obtains. “Novel” I will define as any aspects of a situation which differ sufficiently from prior situations to produce recordable physiological changes in the organism. By the term “thought” I will refer to the active uncertainty produced when an ordered set of memories mismatches the current novelties of the situation [emphasis added]. K.H. Pribram, 1963, p. 54 Neural memory is special in several ways; among them is its capacity not only to retain information but also to utilize it for adaptive purposes. In this sense neural memory becomes connatural with learning, from which it is operationally difficult to distinguish it, although the term learning usually refers to the process of acquiring memory. J.M. Fuster, 1995, p.9

Figure 9.7 also takes into account the processes explained in the above statements by Fuster (1995) and Pribram (1963). The experiments described in this book extend the concepts of Hayek and Fuster related to the inseparability of perception and memory and propose that the processes of attention, perception, learning, and memory are difficult to distinguish and are interwoven and inseparable. Helmholtz (1962) introduced the notion of mental constructs thought to be generated by past experience and stored in or recalled from memory. Accordingly, percepts are thought to be produced by the comparison of recalled and evoked neural images. What is going on in the brain when such images are evoked? Do these images also evoke electrical potentials in parallel to recalling? Mountcastle (1998) went a step further: “Perhaps what we perceive are patterns of neural activity recalled from the memory for the matching operation, rather than the activity evoked directly by sensory stimuli themselves.” It will come as no surprise that direct neurophysiological evidence for the processes of unconscious inference has been hard to obtain. Such propositions are appended as assumptions to several major unsolved problems of neuroscience, for example, how are experiences stored in and recalled from memory? How are neural populations matched and compared? How does the chosen match flow through the conscious experience? Nevertheless, unconscious © 2004 by CRC Press, LLC

TF1663_C09.fm Page 192 Wednesday, May 12, 2004 8:17 PM

experience and matching identification remain themes of later psychological theories of perception (Mackay, 1970). What is happening during everyday cognitive processes such as recognizing familiar objects? The basic idea is that after a sensory code is established, the information in longer-term or persistent memory needed to identify the perceived object is accessed. If the matching process yields a positive result, the object is recognized. This, in turn, leads to the creation of a short-term memory (STM) code. In this case, pathways from the bottom up, similar or identical to those that serve to retrieve information from long-term memory (LTM) are activated (see Klimesch, 1999). It is also evident that during a process of evolving memory, the APLR alliance undergoes a complex multiple matching process that develops along the following steps: 1. What will be learned must be matched (compared) with the stored percepts (or events) of earlier experiences. 2. In a new sequence, attention will be paid to newly learned percepts. 3. In the ensuing matching process, these percepts will be matched or rematched with new sensory–cognitive inputs. The results of EEG experiments clearly show this type of electrical evolution process that manifests a type of reverberation or recurrent circuit (Chapter 3). If the brain is able to match a coming input with a newly learned experience, temporarily created EEG traces are needed. Without such temporary traces, recurrent inputs during an oddball (OB) experiment cannot be compared with already learned material.

9.5.1 MEMORY ACTIVATION: WORKING MEMORY ORGANIZATION OF MEMORY STATES

AND

HIERARCHICAL

Fuster (1997) notes that components of perceptual memory are retrievable and expandable by new stimuli or experiences. Working memory is presumably the same cortical substrate as the kind of STM traditionally considered the gateway to LTM. Fuster’s description includes a crucial remark: Hierarchical organization, however, does not imply that the various individual memories are rigidly stocked and stored in separate cortical domains. Rather, different types of memories — for example, episodic, semantic or procedural — are probably interlinked in mixed networks that span different levels of perceptual and motor hierarchies. Experiments cited in Chapter 3 and Chapter 8 verified the validity of the foregoing statement. In auditory and visual memory task experiments, the EEG oscillations manifested high degrees of plasticity because networks are susceptible activation by superimposed frequency codes, i.e., by multiple oscillations showing varied degrees of enhancements during working memory processes.

9.5.2 MULTIPLE AND COMPLEX MATCHING PROCESSES: RECIPROCAL ACTIVATION OF ALPHA, DELTA, THETA, AND GAMMA CIRCUITS IN WHOLE BRAIN 9.5.2.1 Reentry? According to Damasio (1997), memory depends on several brain systems working in concert across many levels of neural organization and memory is a constant work in progress. If this is so, what type of processing should occur at the EEG level during this constant work in progress? In order to elucidate the role of EEG-related memory, we propose the existence of complex, recurrent (or reverberating) mechanisms. These proposed mechanisms are experimentally founded and revealed perpetual changes of alpha, theta delta, gamma, and beta oscillations and their dynamical superposition. The temporarily created templates seemingly provide dynamic (storage) copies to be compared with new input signals. New sensory input to the brain is matched with oscillatory networks © 2004 by CRC Press, LLC

TF1663_C09.fm Page 193 Wednesday, May 12, 2004 8:17 PM

including traces of phyletic memories and memories acquired throughout life. We must emphasize that response oscillations strongly depend on the state of prestimulus oscillations during working memory processes (see Chapter 3 and Chapter 5). According to Tononi et al. (1992), the notion of reentry extends concepts such as simple feedback, feedforward, and recurrent circuits. Reentry is inherently parallel, has a statistical nature, and has distributed characteristics. As a result it can occur simultaneously within and across several different areas via multiple parallel and reciprocal connections (as seen in cortico-cortical, corticothalamic, and thalamo-cortical radiations). It can occur also via more complex arrangements linking the cortex with the HI or the basal ganglia. Basar ¸ and Stampfer (1985) and Basar ¸ (1988) mentioned the possibility of such recurrent and reentrant neural populations by means of the recording of EEG oscillations without using the reentry expression (see Chapter 1 for data on reentry and Chapter 3 for results of Basar ¸ and Stampfer). The gross electrical activity of a neural population selected to fire (be activated) shows a high degree of variability and plasticity. During experiments with cognitive load and memory activation, the EEG prior to stimulation was altered, thus giving rise to new types of responses because ERPs are controlled by EEG activities preceding stimulation (Basar ¸ et al., 1998; Rahn and Basar ¸ , 1993a and b; Barry et al., 2003; Chapter 4, this volume). According to Barry et al. (2003), stimulation creates states of preference as explained in Chapter 3 and Chapter 5. It is noteworthy that the control of prestimulus activity upon responsiveness or excitability exhibits complex behavior. The alpha, delta, and theta activities that create preference states depend on the modality of sensory–cognitive input and also on the site of the cortex (Basar ¸ et al., 1998; Barry et al., 2003). As a first step, the selectively distributed oscillations in the cortex, thalamus, HI, and other areas of the brainstem were demonstrated in the cat brain by Basar ¸ et al. (1985a and b). They used a different terminology then (synchronized selectivity in the whole brain). Accordingly, it can be hypothesized that during sensory–cognitive processing, oscillatory networks at different levels of the brain are activated as recurrent or reentrant circuits. This behavior can be measured during various types of exogenous or endogenous stimuli as the plasticity and evolving behavior of oscillatory prestimulus or poststimulus activity in the experiments described in Chapter 3. The reentrant signals may play an essential role in creating learnable sequences. This is shown by enhancement, alignment of oscillations, and the transition to preferred phase angles prior expected stimuli shows (see Section 3.4.4). The loops in Figure 9.7 also indicate this recurrent behavior. Measurements of EEG oscillations provide the only possibility of verifying the presence of reentrant circuits in evolving memory and related types of behavior. Basar ¸ (1980 and 1998) and Basar ¸ et al.(1997c) demonstrated that a brain structure showing spontaneous oscillatory activity is susceptible to reacting or is responsive in the same frequency channel as spontaneous oscillations. This is based on the principle of brain response susceptibility* first proposed by Sato (1963) and Sato et al. (1971 and 1977).

9.5.3 PROLONGED OSCILLATIONS, DELAYS, COMPLEX MATCHING

AND

COHERENT STATES

DURING

In Section 9.2, results of experiments on selectively distributed oscillatory alpha and theta networks in the whole brain were explained by means of schematic illustrations. Sensory cognitive inputs activate not only in the cortex, but also in whole brain alpha, theta, or gamma oscillations as serial and heavily parallel processes (see Chapter 6). Accordingly, reciprocal activation of oscillatory networks and complex matching take place in the whole brain. We tentatively extend, therefore, the definition of active memory stated by Fuster (1995) by using the term whole brain instead of whole

* Superposition principle cited in Chapter 6 and Chapter 7.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 194 Wednesday, May 12, 2004 8:17 PM

cortex. At any given time in the awake organism, a widely distributed and changing representational network is active in the whole brain. The activation in the whole brain during memory processes has been clearly demonstrated in Chapter 4. The active memory is manifested by EEG oscillations. The extent and topography of active memory are determined by several factors including (1) present or recent sensory inputs, (2) neuronal ensembles that by association are activated by those inputs, and (3) prevalent EEG oscillatory activities. As the coherence results cited in Chapter 6 demonstrate, widely distributed networks are largely activated in parallel. We denoted this process as a selectively distributed increase of coherence. Table 7.1 tentatively indicates activated memory components in several hierarchical processes. Fuster notes that (1) working memory, also called operant memory, is an operant concept of active memory and (2) active memory is a state rather than a system of memory. According to results presented in Chapter 6 through Chapter 8, the APLR alliance is an operating ensemble and accordingly is activated during dynamic experiments as discussed in Chapter 3, particularly Section 3.2 and Section 3.3. Dynamic changes in the APLR alliance are manifested by the activities of innumerable selectively distributed networks in the brain that are activated upon input to the CNS. The constant dynamic interactions among APLR are represented by selective multiple oscillations and longdistance coherence in various frequency channels (references) that take place upon sensory cognitive stimulation and/or during experiments related to the APLR alliance. By novelty or incertitude of sensory–cognitive inputs, the number of mismatches certainly increases because new percepts are reflected in the newly stored oscillatory templates, not in the earlier oscillatory templates. Therefore, for all brain structures, matching takes time and the process of remembering requires longer duration. For example, I see my secretary every morning and it is no problem to remember her name and who she is. When I must recall the name of a person I have not seen for years, the remembering process requires more time. We have shown that the oscillatory activities in alpha, gamma, theta, and delta reflect the work of the brain. In cases of novelty or incertitude, the human brain works with delays or prolonged behaviors and needs more time (works longer). Parallel to this, we also observe delays or prolongations of oscillations. P300 oscillatory responses (superposition of oscillatory responses) in alpha, theta, and gamma have longer durations, whereas EPs resulting from simple sensory inputs have shorter and damped oscillatory responses (second theta window, alpha prolongation by difficult tasks, delta delay in P300). The response susceptibility of a brain structure depends mostly on its own intrinsic rhythmic activity (Basar ¸ , 1980, 1983a and b, 1992; Narici et al., 1990). A brain system can react to external or internal stimuli producing those rhythms or frequency components that are already present in its intrinsic (natural) or spontaneous activity. If, at a given frequency range, spontaneous brain rhythms are missing, they will be absent in the event-related oscillations (EROs), and vice versa. (P300 40-Hz gamma prolongation by difficult memory tasks, Section 4.3.1). Similarly, when incertitude is decreased and the matching does not require longer duration, the oscillations are shorter as shown in Chapter 3, Chapter 6, and Chapter 8. When the whole brain is involved in the remembering process or in dynamic reciprocal acting* of the APLR alliance, reverberation of signals among brain structures upon sensory stimulation is

* The response susceptibility of a brain structure depends mostly on its own intrinsic rhythmic activity (Basar, ¸ 1980, 1983a and b, 1992; Narici et al., 1990). A brain system can react to external or internal stimuli producing those rhythms or frequency components that are already present in its intrinsic (natural) or spontaneous activity, i.e. if in a given frequency range the spontaneous brain rhythms are missing, they will be absent in the evoked rhythmicities and vice versa. The concept of response susceptibility is strongly connected with the rule of excitement states of neuronal populations as suggested by Basar ¸ (1980, 1983a and b, 1992). According to this rule, if a neuronal population is able to produce spontaneous activity in a given frequency range, then this structural group can be brought to a state of excitement in the same frequency range by sensory stimuli. This means excitability is related to spontaneity. As an immediate consequence, common features and response susceptibilities of brain structures are related to general common tuning and resonances in various structures of the brain in alpha, theta, delta, beta, and gamma frequency bands.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 195 Wednesday, May 12, 2004 8:17 PM

most probable. Therefore, the process of matching takes longer; cross-talks or reverberations may circulate among different structures.* Relevant increases of coherence of theta and alpha responses in the whole brain involving cortex, brainstem, hippocampus, and thalamus were recorded upon sensory–cognitive inputs to the brain (Section 6.5). Since the signal at each electrode location mostly reflects network activity under the electrode, the coherence between two electrodes should measure interactions between two neural populations. In cases of superbinding of all frequencies, we should see a relevant increase of overall coherence, and this is the case as experimental results demonstrate (Basar ¸ et al., 1979; Basar ¸ , 1990). A hypothetical scheme to describe complex matching is discussed below (Figure 9.10).

9.5.4 COMPLEX MATCHING 9.5.4.1 Matching of Multiple Oscillations in Whole Brain The foregoing sections explained the notion of matching and discussed the first steps of Helmholtz and the critical statements of Mountcastle (1998) and Klimesch (1999). In the following section, we hypothesize that matching processes take place via superposition (parallel processing) and serial processing by different brain areas. Further, we assume that the information flow uses multiple frequency codes of EEG oscillations. Resonance phenomena show that various brain structures including long-distance structures are tuned with the same frequency codes. Upon endogenous or exogenous stimulation, links or coherences are enhanced within multiple frequency windows: This means that most brain areas are tuned to be activated or resonate with EEG frequency codes (last section of Chapter 6). Therefore, we are on solid ground when we hypothesize multiple frequency codes for matching processes. We do not exclude the possibility of other matching codes, for example, the activation of feature detectors by simple impulse stimulation. A preliminary hypothetical scheme to describe complex matching and the flow of oscillationcoded information is presented in Figure 9.10.** For the sake of simplicity only alpha (red), theta (green), and gamma (blue) codes are represented. A more complete illustration would include delta, beta, alpha 1, and alpha 2 codes. Figure 9.10 has been modified from Flohr’s scheme that roughly describes information flow in the CNS. It is important to note that the frequency-coded information flow occurs as (1) serial processing and (2) parallel processing. With matching processing, inborn alpha, theta, and gamma networks facilitate adequate information flow in given frequency channels. The received information is matched in all networks in order to determine whether signals from peripheral organs are adjusted or tuned with the alpha, theta, gamma, delta, and beta frequency codes (see Chapter 6 for information on the concept of frequency coding). A simple matching is a comparison with a unique frequency code: for example, alpha matching. In the proposed complex matching procedure, comparisons or matching with all frequency codes (alpha, beta, gamma, etc.) are fully or partly activated as parallel or serial processing, depending on the nature of performed integrative functions. Because oscillatory networks are selectively distributed, the participation of individual frequency codes shows varying degrees of amplitudes. Once an ample alpha signal is brought to ignition, all structures can be excited, giving rise to a huge reciprocal alpha resonance. High loads of exogenous or endogenous sensory–cognitive input produce great resonance in the brain (see Basar ¸ , 1980; Basar ¸ , 1999b; and Chapter 6, this volume concerning resonance phenomena). In cases of serial processing, delays usually occur around 500 ms — the maximal time interval needed for information of flow from one structure to another (Libet, 1991; Miller, 1991; * One could describe this also as mutual effect. ** 300-ms reaction time of hippocampus cited in Chapter 5.

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 196 Wednesday, May 12, 2004 8:17 PM

Primary sensory cortices

Association cortex Built-in

Built-in

VIS Inferior parietal lobe

Polymodal association cortex

VIS AUD AUD SOM SOM

Limbic system

Thalamus LG Nuceus reticularis thalami

MG VPL

Mesencephalic reticular formation SOM AUD VIS Sensory input α information flow γ information flow θ information flow

FIGURE 9.10 (See color insert.) Preliminary hypothetical scheme to describe complex matching and flow of oscillation-coded information. This figure can be understood only by comparing the color illustration.

Basar ¸ , 1999). Late alpha oscillations along with alpha, theta, and gamma delays have been described in all chapters of this book. Alpha, theta, and gamma signals can also circulate among diverse structures when endogenous or exogenous inputs contain difficult or complicated information to be processed. Accordingly, a process of multiple matching of circulating signals of all frequencycoded signals and a great number of brain structures in the whole brain is needed. This may be the cause of prolonged oscillations and the appearance of late alpha, theta, or gamma windows after 300 to 500 ms. In such cases, the whole brain may work longer. Serial and parallel processing may occur at all levels in a reciprocal activating manner. Figure 9.11 hypothetically illustrates the multiple matching processes during oddball tasks, tasks requiring working memory, and other cognitive tasks. As the experiments cited in Chapter 3 and Chapter 6 have shown, ERPs and prestimulus EEG segments are dominated by delta and theta oscillations. Accordingly thick lines were added to the figure to show the dominance of theta and © 2004 by CRC Press, LLC

TF1663_C09.fm Page 197 Wednesday, May 12, 2004 8:17 PM

Primary sensory cortices

Association cortex Built-in

Built-in

VIS Inferior parietal lobe

Polymodal association cortex

VIS AUD AUD SOM SOM

Limbic system

Thalamus LG Nuceus reticularis thalami

MG VPL

Mesencephalic reticular formation SOM AUD VIS Sensory input α information flow γ information flow θ information flow δ information flow

FIGURE 9.11 (See color insert.) Preliminary hypothetical scheme of complex matching and flow of oscillation-coded information. This figure can be understood only by comparing the color illustration.

delta resonance. Compare Figure 9.11 with Figure 9.6 showing the dominance of theta and delta oscillations in oddball experiments.

9.6 LONGER-ACTING MEMORY AND TRANSITION TO PERSISTENT MEMORY IN WHOLE BRAIN 9.6.1 EVOLVING MEMORY: MULTIPLE LEVEL FUNCTIONING

IN

CNS

We introduced evolving memory and memory building as expressions to describe activation of memory states or processes augmented with new learned (memorized) episodes created by constant and reciprocally active APLR alliances. Evolving memory includes interplays in APLR operations © 2004 by CRC Press, LLC

TF1663_C09.fm Page 198 Wednesday, May 12, 2004 8:17 PM

manifested as selectively distributed oscillations in the whole brain. We have seen that the oscillatory theta or alpha responses related to cognitive inputs evolve during the APLR process. As an example of the augmentation of knowledge or learned material, we mention the occurrence of regular and increased alpha activities or types of alpha templates cited in Chapter 3 (Figure 3.8 through Figure 3.11).

9.6.2 LEVEL III ACTIVITIES PORTRAYED

IN

FIGURE 9.7

Level III is portrayed at the bottom of the scheme of memory hierarchies shown in Figure 9.7. According to Section 4.2 (Figure 9.9), event-related changes in EROs provide relevant changes and extensions in electrical manifestations of evolving memory. Newly learned material is transMEMORY

Persistent Memory (Inborn or acquired during life) I

Phyletic memory, physiological memory, perceptual memory (Inborn + Acquired during life)

II (Learning in CNS) + Perceptual Memory + Evolving memory + Procedural Memory (Manifested with selectively distributed multiple oscillations)

III

Partially transferred

Dynamic memory Short-term memory (STM) and /or working memory

Longer-Acting Memory (acquired during life) Semantic + Episodic Memory

Memory consists of Persistent memory, Dynamic memory and longer acting memory

FIGURE 9.12 (Refer to the color insert for a complete understanding of this illustration.) Global scheme indicating memory levels and transitions. Persistent memory is indicated as a separate block and colored yellow as an adjunct to Figure 9.7. This scheme is a simpler version. Again, red background indicates dynamic processes dominated by the working memory system and states of the APLR alliance. The working memory system and procedural memory are involved with whole brain work. Learned memory traces are then transferred to longer, more stable, or quasi-stable activated memory. Persistent memory is explicitly described. It is composed of physiological memory, stabilized traces of dynamic memory, and quasi-stable longer-acting memory, It can be partly transferred to persistent memory (cf. Figure 9.7). In order to follow the dynamics of transition between memory states, see our home page: http://braindynamics.deu.edu.tr/basar.htm

© 2004 by CRC Press, LLC

TF1663_C09.fm Page 199 Wednesday, May 12, 2004 8:17 PM

formed to LTM for longer times in comparison to working memory. Astonishingly, the durations of storage in working memory and long-term memory are not clearly defined in the literature. In our opinion, longer-acting memory is a better phrase than long-term memory for distinguishing working memory from persistent memory. The next step in the hierarchy of memories shown in Figure 9.7 is the transition of memory traces acquired via experience — and that are temporarily stored in longer-acting memory — to persistent memory. According to memory levels described in this chapter, persistent memory includes inborn memory as physiological memory (an ensemble of submemories: echoic, iconic, olfactory, and other memories) and stabilized parts of longer-acting memory acquired throughout life (Figure 9.12). The issue of how new information acquired during enhanced coherence in the whole brain is transferred to and stored in persistent memory surpasses the scope of this book and remains unclear. However, it is important to note that networks of persistent memory operate with the same oscillatory dynamics of evolving memory, i.e., they use the same basic oscillatory alpha, beta, theta, and other codes (see Section 6.8.1 through Section 6.8.3). This indicates that frequency codes may be transferred to persistent memory or may play an essential role during the transition. They are probably invariant building blocks partially contributing to the development of memory traces (or engrams) in which the principle of susceptibility plays a part in facilitating signal transfer. According to Tranel and Damasio (1995), we have good reason to believe that, depending on the type of memory under consideration, storage may be selectional or instructional, i.e., it may depend more or less on selection from a preexisting repertoire of neuron circuit states (cf. Edelman, 1987; Shenoy et al., 1993).

© 2004 by CRC Press, LLC

TF1663_book.fm Page 201 Friday, May 7, 2004 8:47 AM

Trends in Memory 10 New Dynamics: Concluding Remarks For providing a reasonable summary of the neural substrates of human memory, it is necessary to review a large number and variety of neural structures. In fact, it could be argued that virtually the entire human brain is concerned with memory, of one kind or another, at one level or another. D. Tranel and A.R. Damasio, 1995 From those of general assumptions, especially from distributed nature of cortical networks, it follows that we cannot rightfully consider the cognitive functions of the prefrontal cortex in isolation from those of the rest of the frontal cortex or, for that matter, from the totality of the neocortex and the subjacent anatomical stages of the executive hierarchy. J.M. Fuster, 1995

10.1 THE EMPHASIS OF THIS BOOK: FROM A RESEARCH PROGRAM TO A THEORY ON WHOLE-BRAIN WORK Electroencephalogram (EEG)-related memory research has a short history of only 30 years. The influential work of Hans Berger may be considered a forerunner of this book since the importance of EEG oscillations in memory processing constitutes its core leitmotif. New proposals that emerge in this book are based on extensive observations of experiments conducted on the brains of humans and behaving animals. The need to construct an EEG-related framework for the understanding of memory processing has been the most important consequence of the application of the Brain Dynamics Program (see Chapter 1) within the last 30 years. This book indicates that mechanisms of supersynergy and superbinding constitute important frameworks in electrical activation and functioning of the brain. The existence of these mechanisms is in accord with the principles of cooperativity described by Hebb (1949) and equipotentiality described by Lashley (1929). Furthermore, the interwoven functionality of perception and memory indicated by Hayek (1952) is extended via the concepts of interaction of the attention, perception, learning, and remembering (APLR) alliance and evolving memory. In Chapter 11 we introduce a new proposal we have designated whole-brain work with the intent of merging all these concepts into a common denominator.

10.2 DISTRIBUTED MEMORY IN THE WHOLE BRAIN Fuster (1995) noted that the cognitive functions of the cortex of the frontal lobe, as with any part of the neocortex, consist of activation and processing within and between networks of representation or memory networks. Those networks are widely distributed and highly specific; they are defined by their synaptic structures and connectivities. Thus, the memory code relational and all memory is associative. Memory networks overlap and are interconnected. Thus, one neuron or group of

© 2004 by CRC Press, LLC

TF1663_book.fm Page 202 Friday, May 7, 2004 8:47 AM

neurons anywhere in the cortex can belong to many networks and thus many memories. This is why it is virtually impossible to localize memory by any method. Electroencephalogram (EEG) analysis implies that selectively distributed oscillatory networks are also activated during memory processing (Chapter 4, Chapter 6, and Chapter 8). To extend observations by Fuster (1995), the experimental data of this book indicate that the memory function is selectively distributed through the whole brain — not only in the neocortex. Occipital theta prolongation and occipital alpha enhancement in the so-called grandmother experiments showed that all areas of the human cortex are strongly involved in memory processing, depending on the nature of percepts (Chapter 8). Activation of the APLR alliance is selectively distributed throughout the whole brain as animal experiments indicate. Sensory cortex, the hippocampus (HI), reticular formation (RF), and cerebellum* are strongly involved in reciprocal activation of the APLR alliance. Event-related oscillations (EROs) (alpha, theta, and gamma) are also recorded in intracortical structures such as the hippocampus and reticular formation (see Chapter 3 and Chapter 4). Several authors proposed that working memory function is linked to the prefrontal region of the brain, i.e., part of the dorsolateral frontal lobe that comprises the anterior convexity of the cerebral hemispheres (Goldman-Rakic and Friedman, 1991; Fuster, 1991). However, most studies of working memory have been done in nonhuman primates, making it difficult to draw direct neuroanatomical comparisons with humans. Nonetheless, strong evidence from nonhuman primates indicates that the prefrontal region plays a crucial role in working memory, and it is likely that this relationship applies to humans as well, at least to some extent (see Baddeley, 1986, Chapter 10). Our empirical findings strongly demonstrate that working memory is also selectively distributed, as shown by the grandmother experiments in humans and experiments with omitted stimuli in cats. By bringing these results together with the statement of Goldman-Rakic cited above, we arrive at the main scope of this volume by considering all brain functions and memory as a single entity. Perhaps we should refer to a memory and brain function alliance as a more general categorization of the APLR alliance and say that all brain functions are memory linked or all memory is function (see Section 9.5). Lashley (1929) described distributed memory in the whole brain without using the modern techniques of functional magnetic resonance imaging (fMRI) and EEG oscillations. However, since the measurements made then lacked the refinement of experiments performed with EROs, selective distribution could not have been considered. We must also note that Hebb’s ideas (1949) related to reverberations should be expanded to include the whole brain.

10.3 CORRELATION OF BRAIN OSCILLATIONS WITH MULTIPLE BRAIN FUNCTIONS All brain tissues including isolated ganglia of invertebrates, brains of lower vertebrates, and human brains react to sensory and cognitive inputs with oscillatory activity within common and almost invariant frequency channels. Experimental results showed that the degrees of synchrony, amplitudes, durations, and phase lags vary continuously, but similar oscillations are always present in activated brain tissues. Chapter 6 described approximately 50 types of oscillatory activities with definitive or tentative explanations of their functional relations (this book contains more examples, but not all are summarized here). In light of all these results, it is evident that it is not possible to assign only one function to a given type of oscillatory activity. EEG oscillations are involved with multiple functions and act as universal operators or codes of brain functional activity. Complex and integrative brain functions are manifested via the superposition of several oscillations and frequency stabilization, and also with the following parameters: degree of prolongation, enhancement, delay, time locking, and phase locking in several time windows. * See observations of Leiner et al. cited in Section 4.6.4.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 203 Friday, May 7, 2004 8:47 AM

The event-related potential (ERP) oscillatory components we designated alpha, theta, delta, beta, and gamma responses are correlated to various functions. However, the complexity of an event is not reflected only by ERPs. In a number of experiments, spontaneous EEGs can be considered internally evoked potentials (EPs) whose inputs come from yet unknown sources (Chapter 3).

10.3.1 ARE ALL MEMORY STATES TUNED OSCILLATIONS?

WITH

FREQUENCIES

OF

EEG

Figure 9.7 tentatively presents the incorporation of phyletic, sensory, and perceptual memories in physiological memory. Are these memory types or activated memory states interwoven with similar frequency codes? Experiments described in several parts of this book indicate that the mammalian brain responds to simple visual and auditory stimuli with selectively distributed alpha, theta, beta, gamma, and delta responses (see Table 7.1 and illustrations in Chapter 9). As we explained in Chapter 6, alpha, theta, and gamma responses are also manifestations of phyletic memory because they are inborn and possibly wired responses, as shown in the fourth column of Table 7.1 which illustrates the hierarchy of activated memories. According to results cited in Chapter 4, Chapter 6, and Chapter 7 and flow charts in Chapter 9, sensory and perceptual memory machineries are manifested with multiple oscillations in the alpha, beta, gamma, theta, and delta frequency bands. Our empirical evaluation suggests that all these memory types are interwoven and/or tuned with the frequency codes of EEG oscillations. This ability may facilitate the transition between memory states and communication and strong links or alliances with all integrative functions in the brain could be progressed rapidly.

10.4 GESTALTS AND THE GRANDMOTHER PERCEPT The preliminary but tenable consequences of the grandmother experiments described in Chapter 8 are not yet fully known. However, we will indicate the most important and relevant consequence: the fact that the whole brain and all oscillations are activated in the course of recognizing the faces of a subject’s grandmother and an anonymous person (whose face was unknown at the beginning of the experiment). The ensemble of responses behaves as a three-dimensional construct consisting of temporal, spatial, and frequency spaces. The responses to the faces are not represented solely by one location and unique frequency within the brain. Further, the responses are differentiated along the time axis. Delay and prolongation of multiple oscillations are selectively distributed in the whole cortex (Figure 8.3 a and Figure 8.3b). According to Fuster (1997), one neuron or group of neurons anywhere in the cortex can participate in many networks and thus many memories. This is why it is virtually impossible for any method to localize a memory. It does not make sense to talk about frontal control of vision for a subject who has lost the ability to see; injuries of the temporal cortex affect hearing and speech. Accordingly, multiple neural populations are always involved in cognitive memory control, as was demonstrated by the grandmother experiments: Great changes in the frontal lobes and important changes in posterior areas were recorded after subjects’ observations of known and unknown faces. Increased frontal and occipital alpha responses are signs of episodic memory. Enormous increases in occipital delta responses in comparison to frontal delta responses, upon presentation of pictures of the subjects’ grandmothers and of an anonymous face are signs of visual signal detection. The increases of gamma during a working memory task (Basar ¸ , 2001) and in recognition (Burgess, 2000) are linked with activation of multiple cortical areas. We also strongly consider the important roles of the reticular formation and the cerebellum in all memory- and cognition-related tasks (see Section 4.4). Analyses of experiments related to face recognition indicate that transition from semantic memory to episodic memory is manifested by plasticity in the EEG oscillations. Frontal theta oscillations gained the shape of responses to a © 2004 by CRC Press, LLC

TF1663_book.fm Page 204 Friday, May 7, 2004 8:47 AM

known face after many presentations of the unknown picture, indicating that subjects became familiar with the anonymous face (Figure 8.7).

10.5 ACTIVATED MEMORY MANIFESTED BY EEG OSCILLATIONS Fuster (1995) believed that active memory may consist of reverberating activity through the circuits linking neurons within and between the component assemblies of a network that represent the various associated features of the memory. Furthermore, he states that the morphological evidence of recurrent and reciprocal connections within and between cortical areas is formidable. Do APLR alliance activation and reciprocal activation play essential roles in sustained EEG oscillations? This seems probable according to the experiments discussed in Chapter 3 and Chapter 6. New experiments should be designed to yield more evidence to answer this question. Preliminary studies of EEG oscillations are strongly in favor of such a hypothesis, again supporting the concept of Hebb mechanisms at the EEG level.

10.5.1 PLAUSIBILITY EXPERIMENTS

OF

HEBB’S REVERBERATING ACTIVITY BASED

ON

EEG

Hebb (1949) proposed a physiologically concrete hypothesis involving temporary electrical activity of an interconnected neuron net. He specified that the activity within such an assembly was maintained by reverberating circuits and further postulated that the reverberating activity led to structural changes within the assembly that alone constituted long-term memory (LTM). Based on the oscillatory dynamics of EEG data, the reverberation hypothesis remains plausible, but not yet possible to demonstrate. We again mention Figure 3.8 through Figure 3.11 in which alpha oscillations seem to be consolidated. It was clearly demonstrated that alpha activity almost built an oscillatory template after a number of applied cognitive tasks were applied. The evolving EEG process explained in previous sections supports the idea that a structural change is possible, at least at the levels of oscillatory EEG networks. We further refer to observation of delays in the delta, theta, and gamma responses and prolongation of alpha responses during experiments with difficult cognitive and memory task experiments (Chapter 3 and Chapter 4). Delay or prolongation of oscillations should be expected according to the hypothesis of Hebb described in Chapter 1. Nothing can be said here about structural changes at the synaptic level. The following extensions of the above description are necessary: 1. The recurrent excitation through reentrant circuits (and/or resonance) may play a critical role in the sustained activation of a memory network (neural population). 2. The active memory states are accompanied by high amplitude and longer standing oscillations in all EEG frequency windows (see Chapter 3 and Sections 9.5 and 9.6). 3. The manifestation in increase, delay or prolongation of oscillations provides a metric to describe the activated memory in distinct large neural populations.

10.6 MODEL RELATED TO MEMORY STATES All brain functions are inseparable from memory functions and the functions of attention, perception, learning, and remembering (APLR alliance) are interrelated. Memory-related oscillations are selectively distributed in the brain as the grandmother experiments demonstrated. These oscillations have dynamic properties; they evolve after exogenous and endogenous inputs to the brain. The reciprocal acting functions of APLR alliance are categorized as levels of evolving memory. The memory states do not have exact boundaries along the time–space connection; the hierarchical order is embedded in a continuum. Evolving memory has a controlling role in integrative brain functions. The hierarchy of memories is not manifested with separable states because memory © 2004 by CRC Press, LLC

TF1663_book.fm Page 205 Friday, May 7, 2004 8:47 AM

manifests rapid transitions. Therefore, we suggest the designation of memory states instead of separate memories. Based on observations explained in this book, physiological memory is a type of fundamental memory (Section 9.4.2) and a continuous hierarchy and transitions exist among all memory levels.

10.6.1 ACTIVE MEMORY

AND

REVERBERATION HYPOTHESIS

Activation of oscillatory networks and complex matching takes place in the whole brain (Figure 9.10). At any given time in an awake organism, a widely distributed and changing representational network is active in the whole brain — the active memory manifested by EEG oscillations. According to oscillatory dynamics of EEG-data, the reverberation hypothesis remains plausible, but not possible to demonstrate. We refer to Figure 3.8 through Figure 3.11. The figures reveal that the alpha oscillations first seem to be in a state of dynamic development and are later consolidated (as for P300 experiments). This probably indicates that new combinations of frequency codes are also transferred to persistent memory and that the frequency codes play essential roles during the transition. The frequency codes may be invariant building blocks that partly contribute to the development of memory traces (or engrams). In this case, the response susceptibility of brain networks with almost invariant frequency codes aids a major facilitation of signal transfer (Chapter 9). We emphasize the existence of multiple generators giving rise to P300 responses in the hippocampus (HI), in other brain structures such as the cortex, and also in the brain stem (Chapter 4). Accordingly, during states of APLR alliance, the whole brain is involved in multiple functions. The close anatomical relationships between the hippocampus and neocortical association areas, especially the frontal and parietal association areas, suggest that the interaction of the hippocampus with the neocortical association areas in the theta frequency band may be the basis of the theta response system involved in focused attention and expectancy. As explained in Chapter 4, cats might also possess a type of attentive stage that is somewhat comparable to responses in human experiments with the same parameters and designs. Does the reverberation of oscillations by selectively distributed auto-excitation provide a memory buffer during working memory? Although not yet demonstrated, interactive multiple processing during reverberations in the APLR alliance may be considered hypothetically as a type of memory buffer.

10.6.2 MEMORY STATE

AS

CONTINUUM

Some conventional definitions of memory and dynamic processes have been merged in this book. The first relevant step is the introduction of the hierarchy of memories as a continuum. The second step is manifested with empirically founded reciprocal actions in the APLR alliance. Since the processing of simple sensations and complex percepts is an inseparable function, the physiological and perceptual memories must also be inseparable; they may constitute a single entity and it is best to hypothesize an alliance of physiological and perceptual memories. The following features in the hierarchy of memory states are relevant: 1. The APLR alliance is introduced in order to explain dynamic changes in memory processing. The observed dynamic changes due to the alliance are manifested by the activities of selectively distributed brain networks. These networks are innumerable and they are selectively activated upon input to the central nervous system (CNS). We already stated that attention, perception, learning, and memory are separate processes that are difficult to distinguish and separate. 2. Dynamic memory processing is interwoven with evolving memory. We introduced evolving memory or memory building as a description of the activation of memory states or © 2004 by CRC Press, LLC

TF1663_book.fm Page 206 Friday, May 7, 2004 8:47 AM

processes augmented with newly learned (memorized) episodes created by incessant reciprocally activation within the APLR alliance. The evolving memory includes interplays in APLR operations manifested as selectively distributed oscillations in the whole brain. It is a perpetuum. During sensory–cognitive processing, the oscillatory networks at different levels of the brain are activated as the recurrent or reentrant circuits proposed by Edelmann (1978) and Tononi et al. (1992). This behavior can be measured during various types of exogenous or endogenous stimuli as the plasticity and evolving behavior of oscillatory prestimulus or poststimulus activity in the experiments described in Chapter 3, Chapter 4, and Chapter 8. 3. Re-entrant signals may play an essential role in creating learnable sequences, as enhancement, alignment of oscillations, and their transition to preferred phase angles prior to expected stimuli show (see Chapter 5 and Section 3.9 in Chapter 3).

10.6.3 MULTIPLE MATCHING OF RECOGNITION

WITH

EEG FREQUENCY CODES

AS AN

ESSENTIAL

Section 2.6 of Chapter 2 emphasizes the model of Sokolov (1975) for the orienting response: expectation cells in the hippocampus that fire according to the expected input; sensory reporting cells that fire according to actual stimulus; and comparator cells that fire in the case of a discrepancy between the other two. Associative learning always requires an expectation. Bullock (1988) further states that expectations are at least as widespread in the animal kingdom as habituation and associative learning. In light of the pioneering ideas of Helmholtz (1962) and statements of Sokolov and Bullock, we propose the following. During a process of evolving memory, the APLR alliance undergoes a complex multiple matching process that develops according to the following steps: 1. What is learned must be matched (compared) with the stored percepts (or events) of earlier experience. 2. In a new sequence, attention will be paid to newly learned percepts. 3. In the ensuing matching process, these percepts will be rematched with new sensory–cognitive inputs (see also Figure 9.10 and Figure 9.11). The results of EEG experiments clearly show this type of electrical evolution process that manifests reverberations or recurrent circuits (see Chapter 3 and Chapter 9).

10.6.4 LONGER ACTING MEMORY

AND

PERSISTENT MEMORY

According to Section 9.5 and Section 9.6, changes in EROs provide relevant changes and extensions in the electrical manifestations of evolving memory. It is clear that newly learned material is transformed to long-term memory (LTM) for longer time intervals in comparison to working memory. Durations of storage in working memory and LTM are not clearly defined. Accordingly we introduced longer acting memory as an expression to replace LTM and distinguish it from the working memory and persistent memory continuum in the hierarchy of memories (see Figure 9.7). In auditory and visual memory task experiments, EEG oscillations manifested high degrees of plasticity, since networks are susceptible to activation with superimposed frequency codes, i.e., with multiple oscillations showing varied degrees of enhancements (see Section 9.5.2).

10.7 IMPORTANCE OF EEG ANALYSIS In order to investigate the working memory function and semantic or episodic memory, the observation of patients with memory loss is required. It is impossible to record single-cell changes in the electrical activity of such patients. Also, recording changes in the electrical activities of © 2004 by CRC Press, LLC

TF1663_book.fm Page 207 Friday, May 7, 2004 8:47 AM

healthy persons during learning, working memory states, and following learning is not possible by means of single cell-recording. Studies using fMRI have important limitations in temporal resolution. All transitions or substantial changes in types of memories or memory stages can be measured only by means of EEG or MEG (magnetoencephalography) oscillations that have the best temporal resolutions. Moreover they are noninvasive methods. Chapter 3, Chapter 6, and Chapter 8 indicate that one can realize all types of measurement and analysis for describing all states of memory that are impossible to realize with other tools. The memory state transitions and the transfers between states constitute an interwoven and reverberating continuum of functions. With EEG, we can examine dynamic analysis for transitions during states of memory and their reciprocal activation. Such measurements are impossible to realize with other tools. Analysis of brain oscillations seems the most efficient procedure to measure these dynamic processes and especially the electrophysiological manifestations of interactions and reciprocal actions in the APLR alliance. Mountcastle (1998) claims that the dynamic patterns of cooperative activity of neurons and neural populations located in many nodes — sometimes widely separated — cannot be predicted from knowledge of the activity patterns of any single class of neurons or the population in any single node of a distributed system. As quoted in the preface, the last lines of Mountcastle’s book state: Thus a major task for neuroscience is to devise ways to study and to analyze the activity of distributed systems in waking brains, including particularly human brains, and to seek direct correlations and explanations of the relevant behavior in terms of those patterns of neural activity.

© 2004 by CRC Press, LLC

TF1663_C11.fm Page 209 Wednesday, May 12, 2004 8:21 PM

and Whole-Brain Work: 11 Memory Draft of a Theory Based on EEG Oscillations Erol Basar ¸ and Sirel Karakas¸ The whole burden of philosophy seems to consist in this, from the phenomena of motions to investigate the forces of nature, and from these forces to demonstrate the other phenomena. Sir Isaac Newton, Principia Philosophiae, 1686

11.1 INTEGRATION OF PROPOSALS RELATED TO WHOLE-BRAIN WORK Chronological evolution of our conceptual framework evolved in the last 20 to 25 years. Developments in the past 5 years have been particularly rapid. We introduced the various conclusions involved in our conceptual framework in detail in Chapter 3, Chapter 7, and Chapter 10 to enable readers to see the chain of thought in the same sequence in which it was developed. This chapter aims to integrate the previously presented perspectives, principles, and postulates, all of which emphasize the superbinding concept. This chapter is thus an extension of the neurons–brain theory. We tentatively propose that the functions in the whole brain, especially those revealed during behavioral studies, can be studied in the most appropriate way through EEGs and field potentials (see the statement of Mouncastle (1998) in the preface and in Chapter 10). This book emphasizes the three important principles of whole-brain work: Principle 1 — Oscillations that are selectively distributed in the whole brain are activated during cognitive tasks. This principle may explain to an extent Karl Lashley’s conclusion (1929) that the brain operates as a “whole.” Principle 2 — The whole brain is activated in all percept- and memory-related processes and cooperation among distant structures of the brain occurs. The cooperation may be measured by means of coherences and phase differences. The cooperation is also selective; it is demonstrated, for example, in the selective distribution of coherence functions among brain structures and the variances of the respective values between 0 and 1 (see Figure 6.25). The demonstration of the principle of selective cooperation requires the analysis of oscillations in several neural populations and frequency windows. Such analyses and related findings were instrumental in the refinement of the whole-brain and cooperation concepts. The new tools and concepts related to analysis of electrical activity of the brain during sensory–cognitive processing may also provide new interpretations of the statements of Lashley and Hebb concerning brain functioning.

© 2004 by CRC Press, LLC

TF1663_C11.fm Page 210 Wednesday, May 12, 2004 8:21 PM

Principle 3 — The supersynergy machinery consists of an ensemble of at least six submechanisms that act in synergy upon sensory–cognitive input. The mechanisms of supersynergy and superbinding are supported by large amounts of data obtained from animal and human brains. The extended theory along with basic principles and issues presented in the next section integrates the mechanisms of supersynergy, superbinding, and reciprocal interaction of attention, perception, learning, and remembering (APLR alliance) into the earlier neurons–brain theory (Basar ¸ , 1999). It is proposed that integrative brain function is based on the coexistence and cooperative action of these interwoven and interacting submechanisms. Accordingly, integrative brain functions represent wholistic behavior, thus leading to the concept of whole-brain work. Why do we use the whole-brain work phrase? The descriptions of experiments in this book clearly show that all integrative functions that have been analyzed using EEG oscillations are processed by the whole brain. Accordingly, the concepts of all functions and the whole brain are merged into the concept of all works of the brain or the shorter whole-brain work phrase. Whole has a double meaning: One meaning refers to wholistic functions in the whole brain. The second meaning relates to fusion in function, space, and time. The whole-brain work slogan represents the integration of all functions in the whole brain and the functional duration throughout which the brain works. We think that this overall integration is more forcefully represented by whole-brain work than by integrative brain function. Why do we include work in the phrase? Work has an additional dynamic component that is not implied by function. Thus, as represented by the prolongation of oscillations, the brain works longer during cognitive processing (Section 9.5.3). However, as noted earlier, we consider this book a workshop, first for presenting the new ideas; second, to provide a certain knowledge level; and finally, to enable the reader to discuss these new issues. We also consider this book a draft or a proposition that should over time undergo a maturation process by way of discussions of members of the interested scientific community.

11.2 WHOLE-BRAIN WORK THEORY: HOW TO APPROACH BRAIN FUNCTIONS BY MEANS OF EEG OSCILLATIONS The theory of whole-brain work consists of four structural and/or functional levels, each of which contains certain principles and/or critical issues.

11.2.1 LEVEL A: TRANSITION FROM SINGLE NEURONS TO OSCILLATORY DYNAMICS 1. The neuron is the basic signaling element of the brain. 2. Since morphologically different neurons or neural networks are excitable upon sensory–cognitive stimulation, the type of structural element does not play a major role in the frequency tuning of oscillatory networks. This makes the suggestion that all brain networks communicate by means of specific frequency codes of electroencephalogram (EEG) oscillations tenable. In fact, neural populations in the cerebral cortex, hippocampus (HI), and cerebellar cortex are all tuned to the same frequency ranges as EEG oscillations.1 3. Intrinsic oscillatory activities of single neurons form the basis of the natural frequencies of neural assemblies. Oscillatory activities of the neural assemblies and brain consist of gamma, alpha, beta, theta, and delta frequencies — natural frequencies that thus constitute real responses of the brain.2,3 4. Feature detectors (see Section 7.7.1),4 place cells, and memory cells (Fuster, 1995) are empirically established constructs. However, a crucial turning point arose during the grandmother experiments when large groups of neural populations were selectively © 2004 by CRC Press, LLC

TF1663_C11.fm Page 211 Wednesday, May 12, 2004 8:21 PM

5.

6.

7.

8.

9.

10.

activated upon complex semantic and episodic inputs to the brain.5,6 These experiments and similar studies led to the replacement of the neurons with neural assemblies when describing integrative functions of the brain (for material on theta cells, see Section 4.4.5). The emphasis on neural assemblies is the critical issue that discriminates the neurons–brain theory from Sherrington’s neuron doctrine and Barlow’s new perception doctrine.7 Sokolov (2001) excellently described and also constructively criticized the roles of feature detectors. However, integrative functioning of the brain needs the selectively distributed and selectively coherent neural populations8 in concert with the feature detectors. The brain has response susceptibilities that primarily depend on its intrinsic rhythmic activity.9 A brain system responds to external or internal stimuli by those rhythms or frequency components that are among its intrinsic (natural) rhythms and thus part of its spontaneous activity. Accordingly, if a given frequency range does not exist in the spontaneous activity, it will also be absent in the evoked activity. Conversely, if activity in a given frequency range does not exist in the evoked activity, it will also be absent in the spontaneous activity. An inverse relation exists between EEGs and event-related potentials (ERPs). The amplitude of an EEG thus serves as a control parameter for responsiveness that the brain manifests in the form of evoked or event-related potentials.10 An EEG is a quasi-deterministic or chaotic signal and should not be considered as simple background noise. This characteristic and the concept of response susceptibility lead to the conclusion that the oscillatory activity that forms the EEG governs the most general transfer functions in the brain.11 Oscillatory neural tissues that are selectively distributed in the whole brain are activated upon sensory–cognitive input. The oscillatory activities of neural tissues may be described through a number of response parameters. Different tasks and the functions they elicit are represented by different configurations of parameters. Due to this characteristic, the same frequency range is used by the brain to perform multiple functions. The response parameters of oscillatory activities are as follows: • Enhancement or amplitude • Delay or latency12 • Blocking or desynchronization13 • Prolongation or duration14 • Degree of coherence between different oscillations15 • Degree of entropy16 The number of oscillations and the ensemble of parameters obtained under a given condition increase as the complexity of the stimulus increases or the recognition of the stimulus becomes difficult.17

11.2.2 LEVEL B: SUPERBINDING

OF

NEURAL ASSEMBLIES (SUPERSYNERGY)

The submechanisms and/or related processes of supersynergy are as follows: 11. Simple binding involves temporal coherence between cells in cortical columns. This has been demonstrated by several authors.18 12. Each function is represented in the brain by the superposition of various oscillations in the frequency ranges of the EEG. The values of the oscillations on a number of response parameters (see Item 9 above) along with the comparative polarities and phase angles of different oscillations produce function-specific configurations. Neuron assemblies do not obey the all-or-none rule that single neurons obey.19 © 2004 by CRC Press, LLC

TF1663_C11.fm Page 212 Wednesday, May 12, 2004 8:21 PM

13. The superposition principle indicates synergy of the alpha, beta, gamma, theta, and delta oscillations during performance of sensory-cognitive tasks. Thus, according to the superposition principle, integrative brain function is obtained through the combined actions of multiple oscillations. 14. The response susceptibility of the brain activates resonant communications by facilitating electrical processing between networks.20 This may be interpreted as a general tuning process between neural populations and feature detectors.21 15. Parallel processing in the brain shows selectivity produced by varying degrees of spatial coherences that occur over long distances between brain structures and neural assemblies.22 16. Temporal and spatial changes of entropy in the brain demonstrate that oscillatory activity is a controlling factor in functioning.23 The mechanisms of superposition, activation of selectively distributed oscillatory systems, and selectively distributed long distance coherences are collectively labeled superbinding. The supersynergy mechanism additionally includes entropy, and EEG oscillations as control parameters in brain responsiveness.

11.2.3 LEVEL C: INTEGRATION, ALLIANCE

AND INTERPLAY IN

MEMORY

17. All brain functions are inseparable from memory functions.24 As with all integrative brain functions, memory is manifested as multiple and superimposed oscillations. The superposition of multiple oscillations, each of which is characterized by response parameters such as delay and prolongation (Item 9 above) lead to a configuration that is specific to the type of memory. 18. The attention, perception, learning, and remembering functions (APLR alliance) are interrelated. As the grandmother experiments demonstrated,25 memory-related oscillations are selectively distributed in the brain. These oscillations have dynamic properties; they evolve upon exogenous and endogenous inputs. Memory states have no exact boundaries along time or space. A hierarchical order takes place on a continuum but the boundaries of memory states merge into each other. Functions from the simplest sensory memories to the most complex semantic and episodic memories are manifested in distributed multiple oscillations in the whole brain. 19. In our theoretical framework, we introduced evolving memory and memory building. The critical factor in memory building is the APLR alliance that shows a constant reciprocal activation of its four subprocesses. Evolving memory has a controlling role in integrative brain functions.26 The hierarchy of memories is not manifested with separable states because memory manifests rapid transitions. Therefore we suggest using the term memory states rather than memory stores. This explanation does not apply, however, to persistent memory that may be inborn or obtained through over-learned engrams or habits.

11.2.4 LEVEL D: CAUSALITY27

AND

BRAIN RESPONSIVENESS

To discover the cause of an event is to discover something among its temporal antecedents such that, if it had not been present, the event would not have occurred. In memory processing as described in this book, brain responsiveness is controlled at least by three groups of causal factors: Level D is entirely descriptive and is not sequential to the principles of levels A through C. For that reason, the following paragraphs do not follow the numerical sequence above.

© 2004 by CRC Press, LLC

TF1663_C11.fm Page 213 Wednesday, May 12, 2004 8:21 PM

1. Genetically fixed causal factors — The brain and central nervous system (CNS) ganglia contain genetically coded networks. The inborn phyletic memory networks play essential roles in the responsiveness of neural populations, for example: A. Occipital networks in the mammalian brain that respond to light stimulation with enhanced 12-Hz oscillations.28 Note that temporal auditory areas do not react to light stimulation; they respond to auditory stimuli with 10-Hz enhanced oscillations. B. The ray brain reacts with 10-Hz oscillations to electric stimuli (electroception) in the range of a few microvolts, whereas the human brain does not have this ability.29 C. The brain contains selectively distributed gamma networks in addition to alpha networks. These networks show obligatory responses to sensory stimuli. D. Reflexes are genetically coded. The so-called prepotent responses30 of reflexive actions partially represent this type of causality. E. The results of Sokolov (1975) related to orienting responses and genetically fixed causal factors must be emphasized. There are expectation cells that fire upon expected input; sensory-reporting cells that fire in response to a stimulus; and comparator cells that fire whenever a discrepancy between stimuli occurs.31 F. We also strongly indicate here that recent results by the group of Porjesz´ et al. (2002) show that human genetic factors play prominent roles in the generation of eventrelated oscillations (EROs) in humans during cognitive processes (see Section 5.7). 2. Dynamic causality due to reentry and dynamic behavior of EEG oscillations during reciprocal activation of the APLR alliance — Changes in prestimulus EEG oscillations32 and oscillatory behavior during reciprocal activation of the APLR alliance33 affect the future responsiveness of the brain. Possible reentries;34 reciprocal activation of attention, perception, learning and remembering; and recurrent inputs change the causalities of brain responses in a dynamic manner.35 Present behavior influences the immediately following behavior. Represented by oscillations, this adaptive behavior also reflects the plasticity of the brain. Oscillatory plasticity is an additonal causal factor of responsiveness. In auditory and visual memory task experiments, EEG oscillations manifested high degrees of plasticity because neural networks are susceptible to activation by superimposed frequency codes, i.e., with multiple oscillations showing varied degrees of enhancements.36 The existence of a significant difference in major operating oscillations in occipital and frontal areas strongly supports the possibility that spontaneous, evoked, or induced alpha or theta rhythms have fundamentally different functional operations. During some functional states, major operating rhythms can change their functional roles. Thus, as shown in Chapter 3, the nature of the experiment and the task conditions may influence the weights of the functional components on brain rhythms. This behavior of brain oscillations reflects the dynamic plasticity of responsiveness and the top-down processes in oscillatory brain responses. Recent results of Karakas¸ et al. (2003) showed that early sensory gamma responses showed individual differences.37 The existence or absence of gamma responses could be predicted from a battery of neuropsychological tests that measured attention, perception, learning and memory — components of the APLR alliance. Such complex cognitive processes are probably multicausal and include the impacts of both genetic codes and environmental conditions. 3. Age as a causal factor — Healthy aging is a causal factor in brain responsiveness, as demonstrated by analysis of brain oscillations.38

© 2004 by CRC Press, LLC

TF1663_C11.fm Page 214 Wednesday, May 12, 2004 8:21 PM

11.3 NEWTONIAN CAUSALITY, CHAOTIC DYNAMICS, AND BRAIN LANGUAGE …logics and mathematics in the central nervous system, when viewed as languages, must be structurally essentially different from those languages to which our common experience refers. John von Neumann, 1966

The exposition and interpretation of EEG oscillation analysis suffer from a lack of general rules and a framework. With the exception of the conventional analysis of neural activity with single unit recordings, the concept of causality and the need for general rules have been somewhat neglected in the neural sciences field as a whole. Therefore, in closing, we will remark on causality and some leading general approaches and concepts in the physical and mathematical sciences. In all natural sciences, the general questions and problems of causality are based on Newtonian dynamics — the metaphor of all natural sciences. However, recent developments in the dynamics of quantum physics and the new approach to chaos led to a different understanding to the Newtonian causality. In his highly popular book on chaos, Gleick (1987), an advocate of the new science, went so far as to say, “Twentieth-century science will be remembered for just three things: relativity, quantum mechanics, and chaos.” Chaos is the century's third great revolution in the physical sciences. Like the first two, chaos cuts away at the tenets of Newtonian physics. Can chaos also be useful and of great importance in brain research? “EEG is not noise, but is a quasi-deterministic signal.” Such a statement resulted from the need of neurophysiologists to interpret the findings of their experiments (Basar ¸ , 1990). The new developments that emanated from research on chaos in brain function are certainly fascinating. However, these efforts cannot be considered isolated research endeavors. Between 1985 and 2000 when findings on chaotic EEGs appeared, noteworthy progress also occurred in studies of oscillatory phenomena and neural network resonance at the cellular level. Also to be noted are the fruitful results obtained from application of the concept of entropy to brain processes (see Chapters 5 and 7 and Rosso et al., 2000, 2001, and 2002). The conceptual renaissance exemplified by the statement that “EEG is not simple noise” represents a watershed. With a few exceptions, predictions and mathematical descriptions of brain behavior have not been in the mainstream of brain research, Therefore, the “big bang” arising from applying chaotic dynamics to brain activity appeared too early — before brain scientists were prepared to use these concepts. Isaac Newton was not only interested in describing motions of planets, but he also wanted to find the mechanisms of gravitation of the planets. Galileo Galilei observed the oscillations of clocks and also wanted to learn about their machineries. Albert Einstein was interested in describing tracks of the molecules as in the case of Brownian motion, then analyzed the causes of Brownian motion. Furthermore, Einstein sought the causes of gravitation and also wanted to understand the causes of dissipating energy. To establish what has happened in the galactic system, Einstein predicted the existence of black holes by combining facts about astrophysical events with accumulated data on the motion of stars and the laws of physics. He tried to describe the natures of stars and the galaxy, including black holes and other phenomena that were not visible via conventional observation. According to the first law of motion in the Newtonian system, free motion is uniform motion in a straight line. When a force is applied to a body, it causes the body to deviate from this free motion. All observed motions can be divided into free (inertia) components and components due to acting force. The second law states that the force acting on a body is always proportional to the product of its mass and acceleration. Newton never regarded force simply as a name for this mathematical product. As a natural scientist, he eschewed speculation about the nature of forces, thinking it sufficient for scientific purposes to be able to calculate and observe their effects. © 2004 by CRC Press, LLC

TF1663_C11.fm Page 215 Wednesday, May 12, 2004 8:21 PM

The relevance of the prestimulus EEG as a causal factor in attention, perception, learning, and remembering processes has important parallels to Newton’s first law of motion. According to this law, the state of a moving body is a causal factor for the further evolution of the movement. As a parallel, the state of the brain as reflected in the prestimulus EEG is the causal factor for later brain responses. The usefulness to brain research of metaphors in quantum physics including Feynmann’s diagrams (Basar ¸ , 1980 and 1999) has been discussed earlier. It would be an interesting endeavor to include these concepts and approaches into the framework of whole-brain work. This step, however, requires extensive theoretical preparation. Analyzing trajectories (EEG signals) reflecting the activity of neuronal populations can be considered somewhat similar to the analysis of motion. As the brain dynamics expression implies, we intend to also elucidate the causes or mechanisms that give rise to the trajectories of electrical signals in the brain. Similar to trajectories of missiles and trajectories in Brownian motion, EEG trajectories have provided useful information about neural mechanisms that give rise to different transitions. The EEG seems to serve as a fundamental trajectory that is causally related to memory building and integrative brain function. The application of the Newtonian perspective or Einstein’s approach to searching the mechanisms behind EEG trajectories has already started. Does the brain have its own language that is essentially different from other types of languages as John von Neumann (1966) proposed? It probably does. As Table 7.1 shows, the mumbled language that forms brain waves may be an important candidate for the language used by the brain. In fact, the brain does converse and its alpha, beta, gamma, delta, and theta frequency ranges are the phonemes. Superimposed oscillatory responses are the words and the selectively distributed parallel processing pathways represent the syntax of brain language. The whole-brain work that follows supersynergy is equivalent to a sentence in the language of the brain.

NOTES 1. 2. 3. 4. 5. 6.

7. 8.

9. 10. 11. 12. 13. 14.

For review see Basar, ¸ 1998, 1999; Llinas, 1988; Singer, 1989; Eckhorn, 1988; Steriade, 1990, 1992 See Basar ¸ et al., 2001a See Basar ¸ et al., 2001b,c See the proposal by Sokolov in Chapter 7, Section 7.1 See theta cells in Chapter 4 PRELIM, Section 4.5 “Although single neurons may occasionally function as a group, no pontifical neuron or single-neuron decision unit (Bullock, 1992) will ever be found at the highest levels of a system of any large degree of plasticity.” (Edelman, 1978) Barlow, 1995; see also Chapter 7, Section 7.1 This interpretation of Sokolov is important to discussion of the possible tuning of feature detectors with the oscillatory activity. “Frequency code makes it possible to extended specificity in time domain producing ‘frequency and phase selective tuning’ of feature detectors. The EEG is resulted from neuronal oscillations dominated under particular conditions. It means that frequencies are tools for more precise neuronal tuning. An important role in such a tuning of endogenous pacemaker oscillations results in frequency and phase- tuning of the feature detector. Thus the feature detector becomes statedependent tuned.” (Chapter 7, Section 7.1) See Basar, ¸ 1980, 1983a,b, 1992; Narici et al., 1990 See Chapter 5 and Basar, ¸ Rahn, and Basar, ¸ 1993a, Chapters 8 and 9; Basar ¸ et al., 2003. View of Basar, ¸ 1998 and Barry et al., 2003, reentrant circuits, Chapter 1 See Chapter 7, Section 8.2. For chaos analysis see Basar, ¸ 1990 See Chapters 3, 6, for the second theta response window, delay of theta, delay of gamma and P300-40 Hz response see Chapter 4 See Pfurtscheller et al., 1997 See Chapters 3, 5, 6, 8 and Chapter 9, Section 5

© 2004 by CRC Press, LLC

TF1663_C11.fm Page 216 Wednesday, May 12, 2004 8:21 PM

15. Chapter 6, Figures 6.25A and B, Basar ¸ et al., 1999; Koscis et al., 2001; Miltner, 1999; Schürmann et al., 2000 16. Rosso et al., 2000, 2001, 2002…. 17. Chapter 3, Chapter 6 Basar, ¸ 1980, 1999; Basar ¸ et al., 1999, 2000, 2001 18. Binding, see Eckhorn et al., 1988; Gray and Singer, 1989; and Chapter 7 19. Chapters 6, 7 Superposition principle with, new interpretations by Karakas¸ et al., 2000; Klimesch et al., 2000; Chen et al., 2000 20. For response susceptibility see Chapter 9, Section 5.2 and Basar ¸ et al., 1997 21. See the related interpretation of Sokolov in Chapter 7, Section 7.1. 22. Basar, 1980, 1983; Basar ¸ et al., 1999; Miltner, 1999; Schürmann et al., 2000; Kocsis et al., 2001 23. Quiroga et al., 1999; Yordanova et al., 2002; beim Graben et al., 2000, 2001 24. Fuster, 1995, 1997; Hayek, 1952; Chapter 6 and Chapter 10 25. Chapter 8, Basar ¸ et al., 2003a and b 26. See view of Basar ¸ and Barry et al., 2003, and also reentrant circuits (Edelman, 1978, and Tononi et al., 1992 in Chapters 1 and 9) 27. A cause, in the sense of dynamic memory and brain dynamics, must clearly be prior in time to its effect. It need not be contiguous in space and time; but since we know from our everyday experience that the transmission of movement from one body to another only occurs when the bodies are in contact, common sense always assumes that cause and effect are linked by a continuous substance (Modified from Encyclopedia Britannica). Prestimulus EEG and Internal ERPs (endogenous oscillations, mostly presenting top down activity) play a crucial role in brain’s reaction. 28. Several examples in Chapter 6 29. See Chapters 5 and 9 30. Miller, Nature Reviews, Oct. 2000, p. 61 31. See Chapter 4, Section 2.6 32. See Chapter 5, Barry et al., 2003 33. See Chapters 3 and 9. For definition of APLR alliance see Chapter 2 34. See Section 5.4 in Chapter 1 and Edelman, 1978 35. See Chapter 3, Section 5.2.1 Basar ¸ and Stampfer, 1985 36. Experiments in Chapter 3, especially the learnable sequences in Section 4.4 37. See Section 3.3.2 in Chapter 6 38. See footnote nr. 36, and Basar ¸ et al. (1997) for changes in activities in frontal lobes and occipital lobes

© 2004 by CRC Press, LLC

TF1663_book.fm Page 241 Friday, May 7, 2004 8:47 AM

Appendix: Relevant Mathematical Methods A.1 METHOD OF TRANSIENT RESPONSE ANALYSIS Transient response analysis is a common method from the general systems theory. It studies system responses in the time domain by application of step or impulse functions at the input point. The advantage of the method is the observer immediately sees the responses of the system under study when sudden changes (jumps or steps) in input function occur. A practical method of determining evoked potential (brain system transient response) is averaging; the mean value as a function of time is taken from EEG records that follow a number of identical stimulus presentations. The averaged values of time-unlocked activities and the noise (unrelated brain response activities) tend toward zero, whereas the average of the evoked potential (e.g., the event-locked and repeatable signal) tends to remain constant. This approach is the most common and leads to the accumulation of enormous amounts of data (for mathematical descriptions and extended survey information, see Basar ¸ , 1980; Regan, 1989). The greatest disadvantage of the method is that information about the distinct components of the system in the transient response is obscure. When the system contains two, three, or more components, the observer cannot distinguish the different components without further mathematical analysis. Physiologists usually prefer transient response analysis, but peak identification of distinct components (subsystems) in the time domain is often erroneous. Simple-looking system transient responses sometimes have large numbers of components and vice versa. A large number of peaks in the transient response may not necessarily reveal the existence of a large number of system components. See examples presented in Basar ¸ , 1998 and 1999. Another important disadvantage is that certain information about dynamic properties of the brain is lost by applying averaging. The methods that follow were introduced in our Brain Dynamics Research Program and are intended to help overcome these shortcomings.

A.2 FREQUENCY CHARACTERISTICS AND TRFC METHODS Before describing methods, we would like to explain the theoretical bases of the analyses. When the transfer properties of a system are studied, the investigator is often confronted with the resonance phenomenon. Resonance is a response that may be expected of underdamped systems when a periodic signal of a characteristic frequency is applied to the system. The response is characterized by a surprisingly large output amplitude for relatively small input amplitudes, i.e., the gain is large. Resonance phenomena or forced oscillations can be analyzed in the direct empirical way as follows. A sinusoidal signal of frequency f is applied to the system. After a certain period sufficient for the damping of the transient, only forced oscillations that have the frequency of the input signal will remain. The amplitude of the applied signal (input), the amplitude of the forced oscillations (output), and the phase difference between input and output are then measured. By gradually increasing the frequency from f = 0 to f = f0, the output amplitude relative to the input amplitude and the phase differences will be measured as functions of frequency called amplitude and phase characteristics, respectively. Although this approach reveals the natural frequencies of the system, only a small number of scientists have investigated EEG responses using sinusoidally modulated light or sound signals (for

© 2004 by CRC Press, LLC

TF1663_book.fm Page 242 Friday, May 7, 2004 8:47 AM

details on pioneering experiments with sinusoidally modulated light, see Van der Tweel, 1961). Difficulties result from the requirement to record evoked responses to sinusoidal signals over at least three decades of stimulation frequencies, with the evoked responses in each stimulation frequency averaged for at least 100 single stimulus applications. Another difficulty comes from the frequent changes in brain activity stages. They may change within a few minutes and have limited duration that is not sufficient for the application of many sinusoidal stimuli of different frequencies. Frequency data can be obtained also via the transient response frequency characteristics (TRFC) method: According to general systems theory, all information about frequency characteristics of a linear system is contained in the transient response of the system and vice versa. In other words, knowledge of the transient response of the system allows one to predict how the system will react to different stimulation frequencies if the stimulating (input) signal is sinusoidally modulated. If the step response c(t) of the system — in our case the sensory evoked potential — is known, the frequency characteristics G( jw ) of this system can be obtained with a Laplace (or one-sided Fourier) transform of the following form: G( jw ) =

•

Ú

{ } exp(- jwt)dt

d c (t ) dt

0

G( jw ) =

or,

•

Ú exp(- jwt)d {c(t)} 0

G( jw ) =

•

Ú exp(- jwt)l(t) dt 0

where G( jw ) = frequency characteristics of the system; c(t) = step response of the system; l(t ) = impulse response; w = 2 pf = angular frequency, and f = frequency of the input signal. The frequency characteristics G( jw ) including the information on amplitude changes of forced oscillations and the phase angle between output and input is also called the frequency response function. It is a special case of the transfer function and is, in practice, identical with the transfer function (Bendat and Piersol, 1968). For numerical evaluation, a fast Fourier transform (FFT) is used. Let X n be a discrete time series X n = X ( nDt ),T = (( N - 1) Dt ) . Then the Fourier transform Yk of X n is: Yk = Y (w k ) =

N -1

ÂX

n

exp( -i2 pN -1 nk );

w k = 2 pkT -1

n =0

where Yk = a k + ibk are the complex Fourier coefficients, the geometric mean of which is the amplitude spectrum. Although this transform is valid only for linear systems, it can be applied to nonlinear systems as a first approach because the errors due to system nonlinearities are smaller than errors resulting from the length of measurements in sinusoidal stimulation experiments given the rapid transitions of brain activity from one stage to another (Basar ¸ , 1980). In the mathematical literature, the TRFC method is simply called the one-sided Fourier transform or the Laplace transform. We use the TRFC method expression to indicate that the method reveals all characteristics in the time and frequency domains. A physiologist is used to observing © 2004 by CRC Press, LLC

TF1663_book.fm Page 243 Friday, May 7, 2004 8:47 AM

experimental parameters in the time domain by determining transient responses of the studied system. This method allows him or her to obtain the most frequently used transient responses. Moreover, it is possible to analyze the frequency content or the components in the frequency domain by computing the amplitude frequency characteristics from the same transient response. Therefore, we find the TRFC method a more useful description. The methodology for evaluating EPs, AFCs, and digitally filtered data was previously described (Basar ¸ , 1980). The essential steps are as follows: 1. Recording of EEG–EP epochs — With every stimulus presented, segments of EEG activities preceding and following stimulus application are digitized and recorded. This operation is repeated for each trial. 2. Selective averaging of EPs — The stored raw single EEG–EP or EEG–ERP epochs are selected with specified criteria after the recording session. EEG segments showing movement artifacts, sleep spindles, or slow waves are eliminated. 3. Amplitude frequency characteristics (AFCs) — These are computed according to the formula above. 4. Adaptive digital filtering — The method will be described in the next section.

A.3 RESPONSE ADAPTIVE FILTERING Response adaptive filtering is the ideal theoretical method of filtering the transient response of a system in such a way that selective blocking of one or more components (or subsystems) is obtained. Ideal filters are defined as transmission elements which, within a given frequency range, transfer the input signal without change in amplitude and with a fixed (independent of frequency) time shift. Outside this frequency range, a filter has zero transmission (or vice versa, depending on whether the filter has a band-pass or band-stop characteristic). Filters are not physically realizable, but they should be considered useful analytical tools when a contribution made to a signal by a frequency band is to be deduced without distortion. Let us assume a system G( jw ) results from the interconnections of subsystems, G1 ( jw ) , G2 ( jw ) , G3 ( jw ) , …, GK ( jw ) , …, G N ( jw ) in such a way that G( jw ) = G1G2 G3 ...GK ...G N . If we already know the amplitude frequency characteristics of the system G( jw ) under study, and want to know how the transient system response would be affected if one or more components of the system were missing, we first determine the frequency band limits of the component to be eliminated (or component that should be removed from the system). The procedure has the following steps: 1. The amplitude characteristics | G( jw ) | of the system under study are obtained by means of the Laplace (or one-sided Fourier) transform using the transient evoked response c(t):

(

•

) Ú exp(- jwt)d {c(t)}

| G( jw ) |= L d {c (t )} / dt =

0

2. Frequency band limits of theoretical filters are determined according to the frequency and bandwidth of amplitude maxima in the amplitude frequency characteristics | G( jw ) |. © 2004 by CRC Press, LLC

TF1663_book.fm Page 244 Friday, May 7, 2004 8:47 AM

3. After determination of ideal filter characteristics in the frequency domain GKF ( jw ) , the weighting function gKF(t) of the filter is computed by means of the inverse Fourier transform:

gFK (t ) = F

-1

{

}

GKF ( jw ) =

1 2p

+•

Ú(G

KF

(

)) ( )

( jw ) exp - jwt exp jwt dw

-•

By taking t to be equal to zero, any fixed or frequency dependent time shift (inevitable in the case of a real electrical filter) can easily be avoided. 4. The experimentally obtained transient evoked response c(t) is theoretically filtered by means of the convolution integral using the weighting function gKF(t) of an adequately determined ideal filter:

c F (t ) = g KF (t ) * c(t ) =

Úg

KF

(t) c(t - t) dt

where, cF(t) is the filtered evoked response. Since the time response is available in the form of discrete data with a sampling interval of Dt, the integrals in the above equations can be replaced with iterative summation. Evaluation of these integrals is achieved by using the fast Fourier transform (FFT) algorithm. The method of response adaptive digital filtering has a very important advantage in the study of biological systems. It is usually very difficult to remove or attenuate subsystems from a biological system under investigation. However, if the frequency characteristics of a system are known, we can do so theoretically by using the theoretical isolation method — a version of the method of selective blocking by application of pharmacological agents or surgical ablation techniques. Although some electronic filtering methods have already been used in the study of brain waves and evoked potentials, the theoretical isolation method presented here allows us to choose amplitude and phase frequency characteristics of the filters separately. The investigator can apply ideal filters without phase shifts. It is also possible to use filters with exact characteristics and change them adequately according to the amplitude characteristics of the system under study. Therefore, the use of theoretical filters is much simpler and more flexible than the use of electronic filters. Theoretical filters are designed as digital filters and can be applied because they introduce no phase shift in the signal (Basar ¸ and Ungan, 1973; Basar ¸ , 1980, Cook and Miller, 1992; Farwell et al., 1993). However, filter characteristics, especially for narrow filter pass-bands and abrupt amplitude changes typical for averaged EPs, should be chosen so as to avoid the production of filter-related oscillations (Wastell, 1979; de Weerd, 1981; Farwell et al., 1993). We should mention that the choice of filters can be made independently of frequency characteristics but such choices would be arbitrary. Adaptive filtering, however, aims at a component analysis in the study of a given brain response. Important examples of how powerful this method can be are given in Basar ¸ (1998 and 1999).

A.3.1 COMBINED ANALYSIS: EEG

AND

EP COMPARISON

The theoretical background for developing the combined analysis procedure is the concept of the EEG as an active signal in the brain. A spontaneous EEG is regarded as a signal that determines or governs brain responses. Within this framework, we need a technique providing for analysis of

© 2004 by CRC Press, LLC

TF1663_book.fm Page 245 Friday, May 7, 2004 8:47 AM

both spontaneous (ongoing) and evoked EEG activity. The methodology for comparing the brain's spontaneous activity and EPs can be briefly described as follows: 1. A sample of the spontaneous activity of the studied brain structure just prior to stimulus is recorded. 2. A stimulation signal is applied to the experimental subject (animal or human). Visual, acoustical, somatosensory, etc. inputs may serve as stimulation; for example, an auditory step function in the form of a tone burst with a frequency of 2000 Hz and intensity of 80 dB sound pressure level (SPL). 3. Single-sweep evoked response following the stimulation is recorded. As a result, the EEG activities prior and following stimulation are stored together as a combined record. 4. The operations explained in the three steps above are repeated about 100 times. The number of trials depends on the nature of the experiment and the behavior of the subject. 5. The stored single sweeps are averaged using a selective averaging method (Basar ¸ , 1980; Basar ¸ et al., 1975a; Ungan and Basar ¸ , 1976). 6. The selectively averaged EP is transformed to the frequency domain with the Fourier transform to determine amplitude frequency characteristics | G( jw ) | of the studied brain structure. 7. The frequency band limits of the amplitude maxima in the amplitude frequency characteristics | G( jw ) | are determined according to which of the cut-off frequencies of the digital pass-band filters are justified. 8. The stored EEG–EPs are filtered within the properly chosen pass bands (Step 7 above). 9. The maximal amplitudes of the filtered EPs and the so-called enhancement factor for each EEG–EP are evaluated.

A.3.2 DEFINITION

OF

ENHANCEMENT FACTOR (EHF)

In a given experimental EEG–EP record, the enhancement factor (EHF) is the ratio of the maximal time-locked response amplitude (max) to the rms value of the spontaneous activity just prior to the stimulus, with both signals (spontaneous and evoked activities) filtered within the same frequency pass bands: EHF =

max 2 2 rms

REFERENCES Bailey, C.H., Giustetto, M., Huang, Y., Hawkins, R. D. and Kandel, E.R (2000) Is heterosinaptic modulation essential for stabilizing Hebbian plasticity and memory, Nature Review Neuroscience, I: 11-20. Basar ¸ , E. (1998) Brain Function and Oscillations. I. Brain Oscillations: Principles and Approaches, (Berlin Heidelberg, Springer). Basar ¸ , E. (1980) EEG-Brain Dynamics. Relation between EEG and Brain Evoked Potentials, Amsterdam: Elsevier. Basar ¸ , E., Gönder, A., Özesmi, C., Ungan, P. (1975a) Dynamics of brain rhythmic and evoked potentials. I. Some computer methods for the analysis of electrical signals from the brain, Biological Cybernetics, 20: 137–145. Basar ¸ , E., Ungan, P. (1973) A component analysis and principles derived for the understanding of evoked potentials of the brain: Studies in the hippocampus, Kybernetik, 12: 133–140. Basar ¸ , E. (1999) Brain Function and Oscillations. II. Integrative Brain Function. Neurophysiology and Cognitive Processes, (Springer, Berlin Heidelberg). Bendat, J.S., Piersol; A. G. (1968) Mesasurement and Analysis of Random Data, John Wiley, New York. © 2004 by CRC Press, LLC

TF1663_book.fm Page 246 Friday, May 7, 2004 8:47 AM

Cook, E.W. III, Miller, G.A. (1992) Digital filtering: Background and tutorial for psychophysiologists 2, Psychophysiology, 29: 350–367. de Weerd, J.A. (1981) A posteriori time-varying filtering of averaged evoked potentials. I. Introduction and conceptual basis, Biological Cybernetics, 41: 211–222. Eichenbaum, H. (2000) A cortical-hippocampal system for declarative memory, Nature Reviews Neuroscience, USA, I: 41–50. Farwell, L.A., Martinerie, J.M., Bashore, T.R., Rapp, P.E., Goddard, P.H. (1993) Optimal digital filters for long-latency components of the event-related brain potential, Psychophysiology, 30: 306–315. Fuster, J.M. (1995) Memory in the Cerebral Cortex, A Bradford Book The MIT Press, Cambridge, Massachusetts, London England,1–358. Miller, E. K. (2000) The prefrontal cortex and cognitive control, Nature Reviews Neuroscience, 1:59–65. Mountcastle, V. B. (1998) The cerebral cortex, Perceptual Neuroscience, Harvard University Press. Regan, D. (1989) Human Brain Electrophysiology. Evoked Potentials and Evoked Magnetic Fields in Science and Medicine, Elsevier, New York Amsterdam, London. Sokolov, E.N. (2001) in Toward new theories of Brain function and brain dynamics, E. Basar ¸ and M. Schürmann (eds) International Journal Psychophysiology, 39: 87–89. Ungan, P. and Basar ¸ , E. (1976) Comparison of Wiener filtering and selective averaging of evoked potentials, Electroencephalography and Clinical Neurophysiology, 40: 516–520. Van der Tweel, L.H. (1961) Some problems in vision regarded with respect to linearity and frequency response, Annals of the New York Academy of Science, 89: 829–856. Wastell, D.G. (1979) The application of low-pass linear filters to evoked potential data: filtering without phase distortion, Electroencephalography and clinical Neurophysiology, 46: 355–356.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 219 Wednesday, May 12, 2004 8:23 PM

References Abeles, M. and Prut, Y. (1996), Spatio-temporal firing pattern in the frontal cortex of behaving monkeys, Journal of Physiology, Paris, 90: 249–250. Adey, W.R. (1966), Neurophysiological correlates of information transaction and storage in brain tissue, in Stellar, E. and Sprague, J.M., Eds., Progress in Physiological Psychology, Academic Press, New York, 1–43. Adey, W.R. (1989), Cell membranes, electromagnetic fields and intercellular communication, in Basar, ¸ E. and Bullock, T.H., Eds., Brain Dynamics: Progress and Perspectives, Springer, Berlin, 26–42. Adey, W.R., Dunlop, C.W., and Hendrix, C.E. (1960), Hippocampal slow waves: distribution and phase relationships in the course of approach learning, Archives of Neurology, 3: 74–90. Adrian, E.D. (1941), Afferent discharges to the cerebral cortex from peripheral sense organs, Journal of Physiology, London, 100: 159–191. Adrian, E.D. (1942), Olfactory reactions in the brain of the hedgehog, Journal of Physiology, London, 101: 459–473. Adrian, E.D. (1951), Rhythmic discharges from the thalamus, Journal of Physiology, London, 113: 9–108. Andersen, P. and Andersson, S.A. (1968), Physiological Basis of the Alpha Rhythm, Appleton–Century–Crofts, New York. Apostol, G. and Creutzfeldt, O.D. (1974), Crosscorrelation between the activity of septal units and hippocampal EEG during arousal, Brain Research, 67: 65–75. Arieli, A., Sterkin, S., Grinvald, A., and Aertsen, A. (1996), Dynamics of ongoing activity: explanation of the large variability in evoked cortical responses, Science, 272: 1868–1871. Atkinson, R.C. and Shiffrin, R.M. (1968), Human memory: a proposed system and its control processes, in Spence, K.W., Ed., The Psychology of Learning and Motivation: Advances in Research and Theory, Academic Press, New York, 89–195. Babloyantz, A. (1988), Chaotic dynamics in brain activity, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing of the Brain, Springer, Berlin, 196–202. Baddeley, A.D. (1986), Working Memory, Oxford University Press, Oxford. Baddeley, A.D. (1986), Modularity, mass-action and memory (editorial), Quarterly Journal of Experimental Psychology, 38: 527–533. Baddeley, A.D. (1992), Working memory, Science, 255: 556–559. Baddeley, A.D. (1994), Working memory and conscious awareness, in Collins, A., Gathercole, S., Conway, M., and Morris, P., Eds., Theories of Memory, Lawrence Erlbaum Associates, Mahwah, NJ, 11–28. Baddeley, A. (1996), The fractionation of working memory, Proceedings of the National Academy of Science of the USA, 93: 13468–13472. Baddeley, A. (1997), Human Memory, Psychology Press, New York. Baddeley, A. and Hitch, G. (1974), Working memory, in Bower, G.A., Ed., The Psychology of Learning and Motivation, Academic Press, New York, 8: 47–89. Baddeley, A.D. and Wilson, B.A (1985), Phonological coding and short-term memory in patients without speech, Journal of Memory and Language, 24: 490–502. Baddeley, A.D. and Wilson, B.A. (1988), Frontal amnesia and the dysexecutive syndrome, Brain and Cognition, 7: 212–230. Baddeley, A., Wilson, B.A., and Watts, F.N. (1995), Handbook of Memory Disorders, John Wiley & Sons, New York. Bailey, C.H. and Chen, M. (1983), Morphological basis of long-term habituation and sensitization in aplysia, Science, 220: 91–93. Bailey, C.H., Giustetto, M., Huang, Y., Hawkins, R.D., and Kandel, E.R (2000), Is heterosynaptic modulation essential for stabilizing Hebbian plasticity and memory? Nature: Reviews in Neuroscience, I: 11–20. Baird, B. (1999), An oscillating cortical model of auditory attention and electrophysiology, Neurocomputing, 26–27: 319–328.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 220 Wednesday, May 12, 2004 8:23 PM

Barlow, H.B. (1972), Single units and sensation: a neuron doctrine for perceptual psychology, Perception, 1: 371–394. Barlow, H.B. (1995), The neuron doctrine in perception, in Gazzaniga, M., Ed., The Cognitive Neurosciences, MIT Press, Cambridge, MA, 415–436. Barry R.J., De Pascalis V., Hodder, D., Clarke, A.R., and Johnstone, S.J. (2003a), Preferred EEG brain states at stimulus onset in a fixed interstimulus interval auditory oddball task and their effects on ERP components, International Journal of Psychophysiology, 47:187–198. Barry, R.J., Johnstone, S.J., and Clarke, A.R. (2003b), Review of electrophysiology in attention deficit/hyperactivity disorder II. Event-related potentials, Clinical Neurophysiology, 114: 184–198. Bartley, S.H. and Bishop, G.H. (1933), The cortical response to stimulation of the optic nerve in the rabbit, American Journal of Physiology, 103: 159–172. Basar, ¸ E. (1972), A study of the time and frequency characteristics of the potentials evoked in the acoustical cortex, Kybernetik, 10: 61–64. Basar, ¸ E. (1976), Biophysical and Physiological Systems Analysis, Addison-Wesley, Reading, MA. Basar, ¸ E. (1980), EEG Brain Dynamics: Relation between EEG and Brain-Evoked Potentials, Elsevier, Amsterdam. Basar, ¸ E. (1983a), Toward a physical approach to integrative physiology I. Brain dynamics and physical causality, American Journal of Physiology, 14: 510–533. Basar, ¸ E. (1983b), Synergetics of neuronal populations: a survey of experiments, in Basar, ¸ E., Flohr, H., Haken, H., and Mandell, A., Eds., Synergetics of the Brain, Springer, Berlin, 183–200. Basar, ¸ E. (1988), EEG dynamics and evoked potentials in sensory and cognitive processing by the brain, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing by the Brain, Springer, Berlin, 30–55 Basar, ¸ E., Ed. (1990), Chaos in Brain Function, Springer, Berlin. Basar, ¸ E., Ed. (1992), Brain natural frequencies are causal factors for resonances and induced rhythms, in Basar, ¸ E. And Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 425–467. Basar, ¸ E. (1997), Towards a renaissance of alphas, International Journal of Psychophysiology, 26: 1–3. Basar, ¸ E. (1998), Brain Function and Oscillations I. Brain Oscillations: Principles and Approaches, Springer, Berlin, 153–167. Basar, ¸ E. (1999), Brain Function and Oscillations II. Integrative Brain Function. Neurophysiology and Cognitive Processes, Springer, Berlin. Basar, ¸ E. (2004), Macrodynamics of electrical activity in the whole brain, International Journal of Bifurcation and Chaos, 14(2): 1–20. Basar, ¸ E. and Bullock, T.H. Eds. (1989), Brain Dynamics: Progress and Perspectives, Springer, Berlin. Basar, ¸ E. and Bullock, T.H. Eds. (1992), Induced Rhythms in the Brain, Birkhäuser, Boston. Basar, ¸ E. and Özesmi, C. (1972), The hippocampal EEG activity and a systems analytical interpretation of averaged evoked potentials of the brain, Kybernetik, 12: 45–54. Basar, ¸ E. and Schürmann, M. (1994), Functional aspects of evoked alpha and theta responses in humans and cats: occipital recordings in cross-modality experiments, Biological Cybernetics, 72: 175–183. Basar, ¸ E. and Stampfer, H.G. (1985), Important associations among EEG dynamics: event-related potentials, short-term memory and learning, International Journal of Neuroscience, 26: 161–180. Basar, ¸ E. and Ungan, P. (1973), A component analysis and principles derived for the understanding of evoked potentials of the brain: studies in the hippocampus, Kybernetik, 12: 133–140. Basar, ¸ E., Gönder, A., Özesmi, C., and Ungan P. (1975a), Dynamics of brain rhythmic and evoked potentials I. Some computational methods for the analysis of electrical signals from the brain, Biological Cybernetics, 20: 137–143. Basar, ¸ E., Gönder, A., Özesmi, C., and Ungan, P. (1975b), Dynamics of brain rhythmic and evoked potentials II. Studies in the auditory pathway, reticular formation, and hippocampus during the waking stage, Biological Cybernetics, 20: 145–160. Basar, ¸ E., Gönder, A., Özesmi, C., and Ungan, P. (1975c), Dynamics of brain rhythmic and evoked potentials III. Studies in the auditory pathway, reticular formation, and hippocampus during sleep, Biological Cybernetics, 20: 161–169. Basar, ¸ E., Gönder, A., and Ungan, P. (1976a), Important relation between EEG and brain evoked potentials I. Resonance phenomena in subdural structures of the cat brain, Biological Cybernetics, 25: 27–40. Basar, ¸ E., Gönder, A., and Ungan, P. (1976b), Important relation between EEG and brain evoked potentials II. A system analysis of electrical signals from the human brain, Biological Cybernetics, 25: 41–48. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 221 Wednesday, May 12, 2004 8:23 PM

Basar, ¸ E., Demir, N., Gönder, A., and Ungan, P. (1979a), Combined dynamics of EEG and evoked potentials I. Studies of simultaneously recorded EEG–EP-grams in the auditory pathway, reticular formation and hippocampus of the cat brain during the waking stage, Biological Cybernetics, 34: 1–19. Basar, ¸ E., Durusan, R., Gönder, A., and Ungan, P. (1979b), Combined dynamics of EEG and evoked potentials II. Studies of simultaneously recorded EEG-Programs in the auditory pathway, reticular formation and hippocampus of the cat brain during sleep, Biological Cybernetics, 34: 21–30. Basar, ¸ E., Gönder, A., and Ungan, P. (1980), Comparative frequency analysis of single EEG-evoked potential records, Journal of Biomedical Engineering, 2: 9–14. Basar, ¸ E., Flohr, H., Haken, H., and Mandell, A.J. Eds. (1983), Synergetics of the Brain, Springer, Berlin. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Rosen, B., and Schütt, A. (1984), A new approach to endogenous event-related potentials in man: relation between EEG and P300 wave, International Journal of Neuroscience, 24: 1–21. ˘ Basar, ¸ E., Rosen, B., Basar ¸ -Eroglu, C., and Greitschus, F. (1987), The associations between 40–Hz EEG and the middle latency response of the auditory evoked potential, International Journal of Neuroscience, 33: 103–117. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Röschke, J., and Schütt, A. (1989a), The EEG is a quasi-deterministic signal anticipating sensory–cognitive tasks, in Basar, ¸ E. and Bullock, T.H., Eds., Brain Dynamics, Springer, Berlin, 43–71. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Röschke, J., and Schult, J. (1989b), Chaos and alpha reparation in brain function, in Cotterill, R., Ed., Models of Brain Function, Cambridge University Press, Cambridge, 365–395. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Rahn, E., and Schürmann, M. (1991), Sensory and cognitive components of brain resonance responses: analysis of responsiveness in human and cat brain upon visual and auditory stimulation, Acta Otolaryngologica, 491: 25–35. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Parnefjord, R., Rahn, E., and Schürmann, M. (1992a), Evoked potentials: ensembles of brain induced rhythmicities in the alpha, theta and gamma ranges, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 155–181. Basar, ¸ E., Dössel, O., Fuchs, M., Rahn, E., Saermark, K., and Schürmann, M. (1992b), Evoked alpha responses from frontal-temporal areas in multichannel SQUID systems, Proceedings of IEEE Symposium on Neuroscience and Technology, Lyon, 28–33. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Demiralp, T., and Schürmann, M. (1993), Compound P300–40 Hz response of the human brain, Electroencephalography and Clinical Neurophysiology, 87: 14. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Demiralp, T., and Schürmann, M. (1995), Time and frequency analysis of the brain s distributed gamma-band system, IEEE Engineering in Medicine and Biology, 14: 400–410. Basar, ¸ E., Hari, R., Lopes da Silva, F.H., and Schürmann, M. Eds. (1997a), brain alpha activity: new aspects and functional correlates, International Journal of Psychophysiology, 26: 1–482. ˘ Basar, ¸ E., Schürmann, M., Basar ¸ -Eroglu, C., and Karakas¸, S. (1997b), Alpha oscillations in brain functioning: an integrative theory, in Basar, ¸ E., Hari, R., Lopes-Da Silva, F.H., and Schürmann, M., Eds., Brain Alpha Activity: New Aspects and Functional Correlates. Special issue of the International Journal Psychophysiology 26: 5–29. ˘ Basar, ¸ E., Yordanova, J., Kolev, V., and Basar ¸ -Eroglu, C. (1997c), Is the alpha rhythm a control parameter for brain responses? Biological Cybernetics, 76: 471–480. Basar, ¸ E., Rahn, E., Demiralp, T., and Schürmann, M. (1998), Spontaneous EEG theta activity controls frontal visual evoked visual evoked potential amplitudes, Electroencephalography and Clinical Neurophysiology, 108: 101–109. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Karakas¸, S., and Schürmann, M. (1999a), Are cognitive processes manifested in event-related gamma, alpha, theta and delta oscillations in the EEG? Neuroscience Letter, 15: 165–168. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Karakas¸, S., and Schürmann, M. (1999b), Oscillatory brain theory: a new trend in neuroscience, IEEE Engineering in Medicine and Biololgy, 18: 56–66. ˘ ˘ Basar, ¸ E., Demiralp, T., Schürmann, M., Basar ¸ -Eroglu, C., and Ademoglu, A. (1999c), Oscillatory brain dynamics, wavelet analysis, and cognition, Brain Language, 66: 146–183. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Karakas¸, S., and Schürmann, M. (2000), Brain oscillations in perception and memory, International Journal of Psychophysiology, 35: 95–124. ˘ Basar, ¸ E., Basar ¸ -Eroglu, C., Karakas¸, S., and Schürmann, M. (2001a), Gamma, alpha, delta, and theta oscillations govern cognitive processes, International Journal of Psychophysiology, 39: 241–248. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 222 Wednesday, May 12, 2004 8:23 PM

Basar, ¸ E., Özgören, M., and Karakas¸, S. (2001b), A brain theory based on neural assemblies and superbinding, in Wahrnehmen und Erkennen, Reuter, H., Schwab, P., and Gniech, K.D., Eds., Pabst Science Publishers, Lengerich, 11–24. Basar, ¸ E., Özgören, M., and Karakas¸, S. (2001c), Superbinding in integrative brain function and memory, in Proceedings of 23rd Annual International Conference of IEEE Engineering in Medicine and Biology Society. ˘ A. (2001d), Event-related oscil˘ Basar, ¸ E., Schürmann, M., Demiralp, T., Basar ¸ -Eroglu, C., and Ademoglu, lations are real brain responses: wavelet analysis and new strategies, International Journal of Psychophysiology, 39: 91–127. ˘ Basar, ¸ E., Özgören, M., Karakas¸, S., and Basar ¸ -Eroglu, C. (2004), Super-synergy in the brain: the grandmother percept is manifested by multiple oscillations, International Journal of Bifurcation and Chaos, 14(2): 1–38. ˘ Basar, ¸ E., Özgören, M., Basar ¸ -Eroglu, C., and Karakas¸, S. (2003a), Superbinding: spatio-temporal oscillatory dynamics versus Sherrington s neuron doctrine, Theory in Biosciences, 121: 370–385. ˘ Basar ¸ -Eroglu, C. (1990), Eine vergleichende Studie corticaler und subcorticaler ereigniskorrelierter Potentiale des Katzengehirns, Thesis (Habilitationsschrift), Medizinische Universität zu Lübeck. ˘ Basar ¸ -Eroglu, C. and Basar, ¸ E. (1991), A compound P300–40 Hz response of the cat hippocampus, International Journal of Neuroscience, 60: 227–237. ˘ Basar ¸ -Eroglu, C. and Demiralp, T. (2001), Event-related theta oscillations: an integrative and comparative approach in the human and animal brain, International Journal of Psychophysiology, 39: 167–195. ˘ Basar ¸ -Eroglu, C., Basar, ¸ E., and Schmielau, F. (1991a), P300 in freely moving cats with intracranial electrodes, International Journal of Neuroscience, 60: 215–226. ˘ Basar ¸ -Eroglu, C., Schmielau, F., Schramm, U., and Schult, J. (1991b), P300 response of hippocampus with multielectrodes in cats, International Journal of Neuroscience, 60: 239–248. ˘ Basar ¸ -Eroglu, C., Basar, ¸ E., Demiralp, T., and Schürmann, M. (1992), P300 response: possible psychophysiological correlates in delta and theta frequency channels, a review, International Journal of Psychophysiology, 13: 161–179. ˘ Basar ¸ -Eroglu, C., Basar, ¸ E., and Zetler, G. (1996a), Effects of ceruletide and haloperidol on auditory-evoked potentials in the cat brain, International Journal of Neuroscience, 85: 131–146. ˘ Basar ¸ -Eroglu, C., Strüber, D., Kruse, P., Basar, ¸ E., and Stadler, M. (1996b), Frontal gamma band enhancement during multistable visual perception, International Journal of Psychophysiology, 24: 113–125. ˘ Basar ¸ -Eroglu, C., Brand, A., Çabuk, A., and Hildebrandt, H. (2000), Increase of gamma band of ERPs by working memory, International Journal of Psychophysiology, 37: S26. ˘ Basar ¸ -Eroglu, C., Brand, A., and Hildebrandt, H. (2002), Brain oscillations in working memory, International Journal of Psychophysiology, 45: 36. Beim-Graben, P. (2001), Estimating and improving the signal-to-noise ratio of time series by symbolic dynamics, Physical Review E., 64: 1–15. Beim-Graben, P., Saddy, J.D., and Schlesewsky, M. (2000), Symbolic dynamics of event-related brain potentials, Physical Review E., 62: 4. Bendat, J.S. and Piersol, A.G. (1968), Mesasurement and Analysis of Random Data, John Wiley & Sons, New York. Berger, H. (1929), Über das Elektrenkephalogramm des Menschen I. Bericht, Archiv Fuer Psychiatrie und Nervenkrankheiten, 87: 527–570. Berndl, K., Dewitz, W., Grüsser, O.-J., and Kiefer, R.H. (1986a), A test movie to study elementary abilities in perception and recognition of mimic and gestural expression, European Archive of Psychiatry and Neurological Science, 235: 276–281. Berndl, K., von Cranach, M., and Grüsser, O.-J. (1986b), Impairment of perception and recognition of faces, mimic expression and gestures in schizophrenic patients, European Archive of Psychiatry and Neurological Science, 235: 282–291. Berndl, K., Grüsser, O.-J., Martin, M., and Remschmidt, H. (1986c), Comparative studies on recognition of faces, mimic and gestures in adolescent and middle-aged schizophrenic patients, European Archive of Psychiatry and Neurological Science, 236: 123–130. Best, P.J. and Ranck, J.B., Jr. (1982), Reliability of the relationship between hippocampal unit activity and sensory–behavioral events in the rat, Experimental Neurology, 75: 652–664.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 223 Wednesday, May 12, 2004 8:23 PM

Bishop, G.H. (1933), Cyclic changes in excitability of the optic pathway of the rabbit, American Journal of Physiology, 103: 213–224. Bishop, P.O., Jeremy, D., and McLeod, J.G. (1953), Phenomenon of repetitive firing in lateral geniculate of cat, Journal of Neurophysiology, 16: 443–447. Black, A.H. (1975), Hippocampal electrical activity and behaviour, vol. II, in Isaacson, R.I. and Pribram, K.H., Eds., The Hippocampus, Plenum, New York, 129–168. Blanco, S., Figliola, A., Quian-Quiroga, R., Rosso, O.A., and Serrano, E. (1998), Time-frequency analysis of electroencephalogram series III. Wavelet packets and information cost function, Physical Review E, 57: 932–940. Bliss, T.V. and Lomo, T. (1973), Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path, Journal of Physiology, London, 232: 331–356. Bötzel, K., Grüsber, O.J., Haussler, B., and Nauman, A. (1989), The search for face-specific evoked potentials in Basar, ¸ E. and Bullock, T.H., Eds., Brain Dynamics, Springer, Berlin, 449–468. Brandt, M.E. and Jansen, B.H. (1991), The relationship between prestimulus alpha amplitude and visual evoked potential amplitude, International Journal of Neuroscience, 61: 261–268. Brandt, M.E, Jansen, B.H., and Carbonari, J.P. (1991), Prestimulus spectral EEG patterns and visual evoked response, Electroencephalography and Clinical Neurophysiology, 80: 16–20. Brazier, M.A. (1968), Studies of the EEG activity of limbic structures in man, Electroencephalography and Clinical Neurophysiology, 25: 309–318. Bremer, F. and Bonnet, V. (1950), Interprétation des réactions, rhythmiques prologuées des aires sensorielles de l écorce cérébrale, Electroencephalography and Clinical Neurophysiology, 2: 384–400. Bressler, S.L. (1990), The gamma wave: a cortical information carrier? Trends in Neuroscience, 13: 161–162. Bressler, S.L. and Kelso, J.A. (2001), Cortical coordination dynamics and cognition, Trends in Cognitive Science, 1: 26–36. Bullock, T.H. (1988a), Compound potentials of the brain, ongoing and evoked: perspectives from comparative neurology, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing by the Brain, Springer, Berlin, 3–18. Bullock, T.H. (1988b), The comparative neurology of expectation: stimulus acquisition and neurobiology of anticipated and unanticipated input, in Atema, J., Popper, P.R., and Tavolga, W.N., Eds., Sensory Biology of Aquatic Animals, Springer, Berlin, 195–222. Bullock, T.H. (1992), Introduction to induced rhythms: a widespread, heterogeneous class of oscillations, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 1–26. Bullock T.H., Ed. (1993), How do brains work? Birkhäuser, Boston. Bullock, T.H. and Basar, ¸ E. (1988), Comparison of ongoing compound field potentials in the brain of invertebrates and vertebrates, Brain Research Review, 13: 57–75. Bullock, T.H. and McClune, M.C. (1989), Lateral coherence of the electrocorticogram: a new measure of brain synchrony, EEG and Clinical Neurophysiology, 73: 479–498. Burgess, A.P and Gruzelier J.H. (2000), Short duration power changes in the EEG during recognition memory for words and faces, Psychophysiology, 37: 596–606. Burgess, A.P. and Ali, L. (2002), Functional connectivity of gamma EEG activity is modulated at low frequency during conscious recollection, International Journal of Psychophysiology, 46: 91–100. Buzsáki, G. (1985), Electroanatomy of the hippocampal rhythmic slow activity (RSA), in the behaving cat, in Buzsáki, G. and Vanderwolf, C.H., Eds., Electrical Activity of the Archicortex, Akademiai Kiado, Budapest, 143–164. Byrne, J.H., Castellucci, V.F., Carew, T.J., and Kandel, E.R. (1978), Stimulus-response relations and stability of mechanoreceptor and motor neurons mediating defensive gill withdrawal reflex in Aplysia, Journal of Neurophysiology, 41: 402–417. Cajal, R. (1911), Histologie du système nerveux de l’homme et des vertébrés, translated by L. Azoulay. Consejo superior de investigaciones cientificas. Instituto Ramon y Cajal, Madrid, edn 1972. Castellucci, V.F. and Kandel, E.R. (1974), A quantal analysis of the synaptic depression underlying habituation of the gill-withdrawal reflex in Aplysia, Proceedings of the National Academy of Science of the USA, 71: 5004–5008. Chang, H.T. (1950), The repetitive discharges of corticothalamic reverberating circuit, Journal of Neurophysiology, 13: 235–257. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 224 Wednesday, May 12, 2004 8:23 PM

Chen, A.C.N. and Hermann, C.S. (2001), Perception of pain coincides with the spatial expansion of electroencephalographic dynamics in human subjects, Neuroscience Letter, 297: 183–186. Cohen, J.D., Perlstein, W.M., Braver, T.S. Nystrom, L.E., Noll, D.C., and Smith, E.E. (1997), Temporal dynamics of brain activation during a working memory task, Nature, 386: 604–608. Cook, E.W., III and Miller, G.A. (1992) Digital filtering: background and tutorial for psychophysiologists 2, Psychophysiology, 29: 350–367. Courtney, S.M., Ungerleider, L.G., Keil, K., and Haxby, J.V. (1997), Transient and sustained activity in a distributed neural system for human working memory, Nature, 386: 608–611. Craik, F.1.M. and Watkins, M.J. (1973), The role of rehearsal in short-term memory, Journal of Verbal Learning and Verbal Behavior, 12: 599–607. Creutzfeldt, O.D. (1983), Cortex Cerebri, Springer, Berlin. Crick, F. and Koch, C. (1998), Consciousness and neuroscience, Cerebral Cortex, 8: 97–107. Curio, G., Mackert, B.M., Burghoff, M., Koetitz, R., Abraham-Fuchs, K., and Harer, W. (1994), Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system, Electroencephalography and Clinical Neurophysiology, 91: 483–487. Damasio, A.R. (1989), Time-locked multiregional retroactivation: a systems-level proposal for the neural substrates of recall and recognition, Cognition, 33: 25–62. Damasio, A.R. (1994), Descartes Error: Emotion, Reason, and the Human Brain, Grosset–Putnam, New York. Damasio, A.R (1997), Memory’s lanes, Geary, J. (Ed.), Time, 149, 18: 39–45. Damasio, A.R. and Damasio, H. (1994), Cortical systems for retrieval of concrete knowledge: the convergence zone framework, in Koch, C. and Davis, J.L., Eds., Large-Scale Neuronal Theories of the Brain, MIT Press, Cambridge, MA, 61–74. Demiralp, T. and Basar, ¸ E. (1992), Theta rhythmicities following expected visual and auditory targets, International Journal of Psychophysiology, 13: 147–160. ˘ Demiralp, T., Basar ¸ -Eroglu, C., Rahn, E., and Basar, ¸ E. (1994), Event-related theta rhythms in cat hippocampus and prefrontal cortex during an omitted stimulus paradigm, International Journal of Psychophysiology,18: 35–48. ˘ A., Schürmann, M., Basar ˘ Demiralp, T., Ademoglu, ¸ -Eroglu, C., and Basar, ¸ E. (1999a), Detection of P300 waves in single trials by the wavelet transform (WT), Brain and Language, 66: 108–128. Demiralp T., Yordanova, J., Kolev, V., Ademoglu, A., Devrim, M, and Samar, V.J. (1999b), Time-frequency analysis of single-sweep event-related potentials by means of fast wavelet transform, Brain Language, 66: 129–145. Desmedt, J., Ed. (1979), Cognitive Components in Cerebral Event-Related Potentials and Selective Attention, Karger, Basel. Desmedt, J.E. and Tomberg, C. (1994), Transient phase locking of 40 Hz electrical oscillations in prefrontal and parietal human cortex reflects the process of conscious somatic perception, Neuroscience Letter, 168: 126–129. De Weerd, J.A. (1981), A posteriori time-varying filter of averaged evoked potentials I. Introduction and conceptual basis, Biological Cybernetics, 41: 211–222. Dinse, H.R., Krüger, K., Akhavan, A.C., Spengler, F., Schöner, G., and Schreiner, C.E. (1997), Low-frequency oscillations of visual, auditory and somatosensory cortical neurons evoked by sensory stimulation, International Journal of Psychophysiology, 26: 205–227. Donald, N.W. (1983), Neural selectivity in auditory attention: sketch of a theory, in Gaillard, A.W.K. and Ritter, W., Eds., Tutorials in Event-Related Potential Research, Elsevier, Amsterdam. Donchin, E. and Coles M.G.H. (1988), Is the P300 component a manifestation of context updating? Brain and Behavior Science, 11: 357–374. Donchin, E., Ritter, W., and McCallum, W. (1978), Cognitive psychophysiology: endogenous components of the ERP, in Callaway, E., Tueting, P., and Koslov, S., Eds., Event-Related Brain Potentials in Man, Academic Press, New York. Donchin, E., Gerbrandt, L.K., Leifer, L.. and Tucker, L.R. (1973), Contingent negative variations and motor response, in McCallum, W.C. and Knott, J.R., Eds., Event-Related Slow Potentials of the Brain: Their Relations to Behavior. Proceedings of the Second International CNV Congress (1971), Elsevier, Amsterdam, 187–190.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 225 Wednesday, May 12, 2004 8:23 PM

Doppelmayr, M., Klimesch, W., Pachinger, Th., and Ripper, B. (1998), Individual differences in brain dynamics: important implications for the calculation of event-related band power measures, Biological Cybernetics, 79: 49–57. Doppelmayr, M., Klimesch, W., Schwaiger, J., Stadler, W., and Röhm, D. (2000), The time locked theta response reflects interindividual differences in human memory performance, Neuroscience Letter, 278: 141–144. Duke, D. and Pritchard, W.. Eds. (1991), Measuring Chaos in the Human Brain, World Scientific, Singapore. Eccles, J.C. (1973), The Understanding of the Brain, McGraw Hill, New York. Eckhorn, R., Bauer, R., Jordan, R., Brosch, W., Kruse, M., Munk, M., and Reitboeck, H.J. (1988), Coherent oscillations: a mechanism of feature linking in the visual cortex? Biological Cybernetics, 60: 121–130. Edelman, G.M. (1978), Group selection and phasic reentrant signaling: a theory of higher brain functions, in Edelman, G.M. and Mountcastle, V.B., Eds., The Mindful Brain, MIT Press, Cambridge, MA, 51–100. Edelman, G.M. (1987), Neural Darvinism, Basic Books, New York. Eeg-Olofsson, O. (1971), The development of the EEG in normal children from age 1 to 15 years: the 14 and 6 Hz positive spike phenomenon, Neuropädiatrie, 3: 11–45. Egner, T. and Gruzelier, J H. (2001), Learned self-regulation of EEG frequency components affects attention and event-related brain potentials in humans, Neuroreport, 12: 4155–4159. Eichenbaum, H. (1999), The hippocampus and mechanisms of declarative memory, Behavioral Brain Research, 103: 123–133. Eichenbaum, H. (2000), A cortical-hippocampal system for declarative memory, Nature: Reviews in Neuroscience, I: 41–50. Elazar, Z. and Adey, W.R. (1967), Spectral analysis of low frequency components in the electrical activity of the hippocampus during learning, Electroencephalography and Clinical Neurophysiology, 23: 225–240. Ellis, H.C. and Hunt, R.R. (1993), Fundamentals of Cognitive Psychology, Brown, Oxford. Engel, A.K., Singer, W. (2001), Temporal binding and the neural correlates of sensory awareness, Trends in Cognitive Sciences, 5: 16–25. Engel, A.K., Fries, P., and Singer, W. (2001), Dynamic predictions: oscillations and synchrony in top-down processing, Nature Reviews Neuroscience, 2: 704–716. Fabiani, M., Karis, D., and Donchin, E. (1990), Effects of mnemonic strategy manipulation in a Van Restorff paradigm, Electroencephalography and Clinical Neurophysiology, 75: 22–35. Farwell, L.A., Martinerie, J.M., Bashore, T.R., Rapp, P.E., and Goddard, P.H. (1993), Optimal digital filters for long-latency components of the event-related brain potential, Psychophysiology, 30: 306–315. Fell, J., Hinrichs, H., and Röschke, J. (1997), Time course of 40 Hz EEG activity accompanying P3 responses in an auditory oddball paradigm, Neuroscience Letter, 235: 121–124. Fell, J., Klaver, P., Lehnertz, K., Grunwald, T., Schaller, C., Elger C.E., and Fernández, G. (2001), Human memory formation is accompanied by rhinal-hippocampal coupling and decoupling, Nature: Reviews in Neuroscience, 4: 1259–1264. Fernandez, G., Brewer, J.B., Zhao, Z., Glover, G.H., and Gabrieli, J.D.E. (1999), Level of sustained entorhinal activity at study correlates with subsequent cued-recall performance functional MRI study with high acquisition rate, Hippocampus, 9: 35–44. Fernandez,G. and Tendolkar, I. (2001), Integrated brain activity in medial temporal and prefrontal areas predicts subsequent memory performance: human declarative memory formation at the system level, Brain Research Bulletin, 55: 1–9. Fessard, A. (1961), The role of neuronal networks in sensory communications within the brain, in Rosenblith, W.A., Ed., Sensory Communication, MIT Press, Cambridge, MA. Flohr, H. (1991), Brain processes and phenomenal consciousness: a new and specific hypothesis, Theory and Psychology, 1: 245–262. Freeman, W.J., Ed. (1975), Mass Action in the Nervous System, New York, Academic Press. Freeman, W.J. (1998), Foreword, in Basar, ¸ E., Ed., Brain Function and Oscillations I. Brain Oscillations: Principles and Approaches, Springer, Berlin. Freeman, W.J. (1999), Foreword, in Basar, ¸ E., Ed., Brain Function and Oscillations II. Integrative Brain Function: Neurophysiology and Cognitive Processes, Springer, Berlin. Freeman, W.J. and Skarda, C.A. (1985), Spatial EEG patterns, non-linear dynamics and perception: the neo-Sherringtonian view, Brain Research Review, 10: 147–175. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 226 Wednesday, May 12, 2004 8:23 PM

Friedman, D. (1984), P300 and slow wave: the effects of reaction time quartile, Biological Psychology, 18: 49–71. Frisk, V. and Milner, B. (1990a), The relationship of working memory to the immediate recall of stories following unilateral temporal or frontal lobectomy, Neuropsychologia, 28: 121–135. Frisk, V. and Milner, B. (1990b), The role of the left hippocampal region in the acquisition and retention of story content, Neuropsychologia, 28: 349–359. Fuster, J.M. (1989), The Prefrontal Cortex: Anatomy, Physiology, and Neuropsychology of the Frontal Lobe, Raven Press, New York. Fuster, J.M. (1991), Role of prefrontal cortex in delay tasks: evidence from reversible lesion and unit recording in the monkey, in Levin, H.S., Eisenberg, H.M., and Bendon, A.L., Eds., Frontal Lobe Function and Dysfunction, Oxford University Press, New York, 59–71. Fuster, J.M. (1995), Memory in the Cerebral Cortex, MIT Press, Cambridge, MA. Fuster, J.M. (1997), Network memory, Trends in Neuroscience, 20: 451–459. Fuster JM. (2000a), Prefrontal neurons in networks of executive memory, Brain Research Bulletin, 52: 331–336. Fuster, J.M. (2000b), Cortical dynamics of memory, International Journal of Psychophysiology, 35: 155–164. Fuster, J.M. (2000c), Executive frontal functions, Experimental Brain Research, 133: 66–70. Fuster, J.M. (2000d), Memory networks in the prefrontal cortex, Progress in Brain Research, 122: 309–316. Galambos, R. and Hillyard, S.A. (1981), Electrophysiological approaches to human cognitive processing, Neuroscience Research Program Bulletin, 20: 141–265. Galambos, R.S., Makeig, P., and Talmachoff, A. (1981), 40 Hz auditory potential recorded from the human scalp, Proceedings of the National Academy of Science of the USA, 78: 2643–2647. Galambos, R., Rose, J.E., Bromley, R.B., and Hughes, J.R. (1952), Microelectrode studies on medial geniculate body of the cat II. Response to clicks, Journal of Neurophysiology, 15: 359–380. Gauthier, P., Hamon, J.F., and Gottesmann, C. (1986), Etude de la variation contingente negative (VCN), et des composantes tardives des potentiels evoques au cours du sommeil chez l homme Revue d’Electroencephalographie et de Neurophysiologie Clinique, 15: 323–330. Gebber, G.L., Zhong, S., and Barman, S.M. (1995), The functional significance of the 10–Hz sympathetic rhythm: a hypothesis, Clinical and Experimental Hypertension, 17: 181–195. Gevins, A., Smith, M.E., McEvoy, L., and Yu, D. (1997), High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing and practice, Cerebral Cortex, 7: 374–385. Giannitrapani, D. (1985), The electrophysiology of intellectual functions, Karger, Basel. Glanzer, M. and Cunitz, A.R. (1966), Two storage mechanisms in free recall, Journal of Verbal Learning and Verbal Behavior, 5: 351–360. Gleick, J. (1987), Chaos: Making New Science, Penguin, New York. Glickstein, M. (1993), Cerebellar function in normal movement and in motor learning, in Andersen, P., Hvalby, O., Paulsen, O., and Hokfelt, B., Eds. (1993), Memory Concepts, Elsevier, Amsterdam, 127–135. Goldman-Rakic, P.S. (1988), Topography of cognition: parallel distributed networks in primate association cortex, Annual Review of Neuroscience, 11: 137–156. Goldman-Rakic, P.S. (1996), Regional and cellular fractionation of working memory, Proceedings of the National Academy of Science of the USA, 93: 13473–13480. Goldman-Rakic, P.S. (1997), Space and time in the mental universe, Nature, 386: 559–560. Goldman-Rakic, P.S. and Friedman, H.R. (1991), The circuitry of working memory revealed by anatomy and metabolic imaging, in Levin, H.S., Eisenberg, H.M., and Benton, A.L., Eds., Frontal Lobe Function and Dysfunction, Oxford University Press, New York, 72–91. Goodin, D.S., Squires, K.C., Henderson, B.H., and Starr, A. (1978), An early event-related cortical potential, Psychophysiology, 15: 360–365. Gönder, A. and Basar, ¸ E. (1978), Evoked frequency stabilization in the electrical activity of the cat brain, Biological Cybernetics, 31: 193–204. Gray, C.M. and McCormick, D.A. (1996), Chattering cells: superficial pyramidal neurons contributing to the generation of synchronous oscillations in the visual cortex, Science, 274: 109–113. Gray, C.M. and Singer, W. (1987), Stimulus specific neuronal oscillations in the cat visual cortex: a cortical function unit, Social Neuroscience, 404: 3.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 227 Wednesday, May 12, 2004 8:23 PM

Gray, C.M. and Singer, W. (1989), Stimulus-specific neuronal oscillations in orientation columns of cat visual cortex, Proceedings of the National Academy of Science of the USA, 86: 1698–1702. Gray, C.M., König, P., Engel, A.K., and Singer, W. (1989), Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties, Nature, 338: 334–337. Griffith, J.S. (1971), Mathematical Neurobiology: An Introduction to the Mathematics of the Nervous System, Academic Press, New York. Grossberg, S. (1999), The link between brain learning, attention, and consciousness, Conscious Cognition, 8: 1–44. Gruzelier J, Ed. (1996), New Advances in EEG and Cognition, International Journal of Psychophysiology, 241–187. Grüsser, O.J. (1984), Face recognition within the reach of neurobiology and beyond it, Human Neurobiology, 3: 183–190. Gurtubay, I.G., Alegre, M., Labarga, A., Malanda, A., Iriarte, J., and Artieda, J. (2001), Gamma band activity in an auditory oddball paradigm studiedwith the wavelet transform, Clinical Neurophysiology, 112: 1219–1228. Haenschel, C., Baldeweg, T., Croft, R.J., Whittington, M., and Gruzelier, J. (2000), Gamma and beta frequency oscillations in response to novel auditory stimuli: a comparison of human electroencephalogram (EEG), data with in vitro models, Proceedings of the National Academy of Science of the USA, 13: 7645–7650. Haig, A.R. Gordon, E., De Pascalis, V., Meares, R.A., Bahramali, H., and Harris, A. (2000), Gamma activity in schizophrenia: evidence of impaired network binding? Clinical Neurophysiology, 111: 1461–1468. Haken, H. (1977), Synergetics: An Introduction, Springer, Berlin. Halgren, E., Stapleton, J.M., Smith, M., and Altafullah, I. (1986), Generators of human scalp P3, in Cracco, R.Q. and Bodis-Wollner, I., Eds., Evoked Potentials, Liss, New York, 269–284. Hartline, P.H. (1987), Multisensory convergence, in Adelman, G., Ed., Encyclopedia of Neuroscience, Birkhäuser, Boston, 706–709. Hawkins, R.D., Abrams, T.W., Carew, T.J., and Kandel, E.R. (1983), A cellular mechanism classical conditioning in Aplysia: activity-dependent amplification of presynaptic facilitation, Science, 219: 400–405. Hayek, F.A., Ed., The Sensory Order, University of Chicago Press, Chicago. Hebb, D.O. (1949), The Organization of Behaviour, John Wiley & Sons, New York. Helmholtz, H. (1962), Handbook of Physiological Optics (translation), Dover, New York. Hernandez-Peón, R. (1961), Reticular mechanisms of sensory control, in Rosenblith, W.A., Ed., Sensory Communication, MIT Press, Cambridge, MA. Hillyard, S.A. and Picton, T.W. (1979), Event-related brain potentials and selective information processing in man, in Desmedt, J.E., Ed., Cognitive Components in Cerebral Event-Related Potentials and Selective Attention, Karger, Basel, 1–50. Howard, L. and Polich, J. (1985), P300 latency and memory span development, Developmental Psychology, 21: 283–289. Inouye, T., Shinosaki, K., Sakamoto, H., Toi, S., Ukai, S., Iyama, A., Katzuda, Y., and Hirano, M. (1991), Quantificationof EEG irregularity by use of the entropy of power spectrum, Electroencephalography and Clinical Neurophysiology, 79: 204–310. Inouye, T., Shinosaki, K., Iyama, A., and Matsumoto, Y. (1993), Localization of activated areas and directional EEG patterns during mental arithmetic, Electroencephalography and Clinical Neurophysiology, 86: 224–230. James, W. (1890), The Principles of Psychology, vol. 2, Henry Holt, New York. Jansen, B.H., and Brandt, M.E. (1991), The effect of the phase of prestimulus alpha activity on the averaged visual evoked response, Electroencephalography and Clinical Neurophysiology, 80: 241–250. Jarcho, L.W. (1949), Excitability of cortical afferent systems during barbiturate anesthesia, Journal of Neurophysiology, 12: 447–457. Jasper, H.H. (1958), The ten–twenty electrode system of the International Federation, Electroencephalography and Clinical Neurophysiology, 10: 371–375. John, E.R. (1967), Mechanisms of Memory, Academic Press, New York. John, E.R. (1988), Electrophysiological and radiographic evidence for the mediation of memory by an anatomically distributed system, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing by the Brain, Springer, Berlin, 56–87. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 228 Wednesday, May 12, 2004 8:23 PM

Johnson, R., Jr. (1988), Scalp-recorded P300 activity in patients following unilateral temporal lobectomy, Brain, 111: 1517–1529. Johnson, R., Jr., Rohrbaugh, J.W., and Parasuraman, R., Eds. (1990), Event-Related Brain Potentials: Basic Issues and Applications, Oxford University Press, New York. ´ B., Almasy, L., Goate, A., Hinrichs, A., Kwon, J., Rice, J.P., Rohrbaugh, J., Beirut, L., Jones, K., Porjesz, Stock, H., Wang, J.C., Wu, W., Bauer, L.O., Chorlian, D.B., Dick, D.M., Edenberg, H., Foroud, T., Hesselbrock, V., Kuperman, S., Nurnberger, J., O’Connor, S.J., Schuckit, M.A., Stimus, A.T., Reich, T., and Begleiter, H. (2003), Linkage and linkage disequilibrium of evoked EEG oscillations with CHRM2 receptor gene polymorphisms: implications for human brain dynamics and cognition, Nature and Neuroscience, submitted. Kahana, M.J., Sekuler, R., Caplan, J.B., Kirschen, M., and Madsen, J.R. (1999), Human theta oscillations exhibit task dependence during virtual maze navigation, Nature, 399: 781–784. ´ B., Jones, K.A., Choi, K., Chorlian, D.B., Padmanabhapillai, A., Rangaswamy, M., Kamarajan, C., Porjesz, Stimus, A., and Begleiter, H. (2003), The role of brain oscillations as functional correlates of cognitive systems: a study of frontal inhibitory control in alcoholism, International Journal of Psychophysiology, in press. Kandel, E.R. and Schwartz, J.H. (1982), Molecular biology of learning: modulation of transmitter release, Science, 218: 433–443. Kandel, E.R., Schwartz, J.H., and Jessel, T.M. (1991), Principles of Neural Science, Elsevier, New York. Karakas¸, S. (1997), A descriptive framework for information processing: an integrative approach, International Journal of Psychophysiology, 26: 353–368. Karakas¸, S. and Basar, ¸ E. (1983), High-frequency components of human visual-evoked potentials, International Journal of Neuroscience, 19:161–172. Karakas¸, S. and Basar, ¸ E. (1998), Early gamma response is sensory in origin: a conclusion based on crosscomparison of results from multiple experimental paradigms, International Journal of Psychophysiology, 31: 13–31. Karakas¸, S., Ungan, P., and Basar, ¸ E. (1980), Human high frequency visual evoked responses, Pflügers Archive (Deutsche Physiologiche Gesellschaft), 53: 384. Karakas¸, S., Erzengin, O.U., and Basar, ¸ E. (2000a), Genesis of human event-related responses explained through the theory of oscillatory neural assemblies, Neuroscience Letter, 285: 45–48. Karakas¸, S., Erzengin, O.U., and Basar, ¸ E. (2000b), A new strategy involving multiple cognitive paradigms demonstrates that ERP components are determined by the superposition of oscillatory responses, Clinical Neurophysiology, 111: 1719–1732. Karakas¸, S., Karakas¸, H., and Erzengin, O.U. (2002), Early sensory gamma represents the integration of bottom-up and top-down processing, International Journal of Psychophysiology, 45: 39. Karakas¸, S., Bekçi, B., and Erzengin, Ö.U. (2003), Early gamma response in human neuroelectric activity is correlated with neuropsychological test scores, Neuroscience Letter, 340: 37–40. Katzir-Katchalsky, A. and Rovland, V. (1974), Dynamics Patterns of Brain Cell Assemblies, MIT Press, Cambridge, MA. Keil, A., Gruber, T., and Muller, M.M. (2001), Functional correlates of macroscopic high-frequency brain activity in the human visual system, Neuroscience Biobehavioral Review, 25: 527–534. Kirschfeld, K. (1992), Oscillations in the insect brain: do they correspond to the cortical gamma waves of vertebrates? Proceedings of the National Academy of Science of the USA, 89: 4764–4768. Kiss, T., Orban G., Lengyel, M., and Erdi, P. (2001), Intrahippocampal gamma and theta rhythm generation in a network model of inhibitory interneurons, Neurocomputing, 38–40 : 713–719. Klatzky, R.L. (1980), Human Memory: Structures and Processes, Freeman, New York. Klimesch, W. (1999), EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis, Brain Research Review, 29: 169–195. Klimesch, W. (2002), personal communication. Klimesch, W., Schimke, H., and Schwaiger, J. (1994), Episodic and semantic memory: an analysis in the EEG theta and alpha band, Electroencephalography and Clinical Neurophysiology, 91: 428–441. Klimesch, W., Doppelmayr, M., Russegger, H., and Pachinger, Th. (1996), Theta band power in the human scalp EEG and the encoding of new information, NeuroReport, 7: 1235–1240. Klimesch, W., Doppelmayr, M, Schimke, H., and Ripper, B. (1997a), Theta synchronization in a memory task, Psychophysiology, 34: 169–176. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 229 Wednesday, May 12, 2004 8:23 PM

Klimesch, W., Doppelmayr, M., Pachinger, Th., and Russegger, H. (1997b), Event-related desynchronization in the alpha band and the processing of semantic information, Cognitive Brain Research, 6: 83–94. Klimesch, W., Doppelmayr, M., Röhm, D., Pöllhuber, D., and Stadler, W. (2000a), Simultaneous desynchronization and synchronization of different alpha responses in the human electroencephalograph: a neglected paradox? Neuroscience Letter, 284: 97–100. Klimesch, W., Doppelmayr, M., Schwaiger, J., Winkler, T., and Gruber, W. (2000b), Theta oscillations and the ERP old/new effect: independent phenomena? Clinical Neurophysiology, 111: 781–793. Klimesch, W., Doppelmayr, M., Stadler, W., Pöllhuber, D., Sauseng, P., and Röhm, D. (2001a), Episodic retrieval is reflected by a process specific increase in human theta activity, Neuroscience Letter, 302: 49–52. Klimesch, W., Doppelmayr, M., Yonelinas, A., Kroll, N.E.A., Lazzara, M., Röhm, D., and Gruber, W. (2001b), Theta synchronization during episodic retrieval: neural correlates of conscious awareness, Cognitive Brain Research, 12: 33–38. Knight, R.T., Hillyard, S.A., Woods, D.L., and Neville, H.J. (1981), The effects of frontal cortex lesions on event-related potentials during auditory selective attention, Electroencephalography and Clinical Neurophysiology, 52: 571–582. Kocsis, B., Viana Di Prisco, G., and Vertes, R.P. (2001), Theta synchronization in the limbic system: The role of Gudden’s tegumental nuclei, European Journal of Neuroscience, 13: 381–388. Kolev, V. and Schürmann, M. (1992), Event-related prolongation of induced EEG rhythmicities in experiments with a cognitive task, International Journal of Neuroscience, 67: 199–213. Kolev, V. and Yordanova, J. (1997), Analysis of phase-locking is informative for studying event-related EEG activity, Biological Cybernetics, 76: 229–235. Kolev, V., Yordanova, J., Schürmann, M., and Basar, ¸ E. (1999), Event-related alpha oscillations in task processing, Clinical Neurophysiology, 110: 1784–1792. Koshino, Y. and Niedermeyer, E. (1975), Enhancement of Rolandic mu rhythm by pattern vision, Electroencephalography and Clinical Neurophysiology, 38: 535–538. Krause, C.M., Sillanmaki, L., Haggqvist, A., and Heino, R. (2001), Test–retest consistency of the event-related desynchronization/event-related synchronization of the 4–6, 6–8, 8–10 and 10–12 Hz frequency bands during a memory task, Clinical Neurophysiology,112: 750–757. Kuhlman, W.N. (1978), Functional topography of human mu rhythm, Electroencephalography and Clinical Neurophysiology, 44: 83–93. Kutas, M., McCarthy, G., and Donchin, E. (1977), Augmenting mental chronometry: the P300 as a measure of stimulus evaluation time, Science, 197: 792–795. Kyllonen, P.C. and Christal, R.E. (1990), Reasoning ability is (little more than), working memory capacity, Intelligence, 14: 389–433. Lang, M., Lang, W., Diekmann, V., and Kornhuber, H.H. (1987), The frontal theta rhythm indicating motor and cognitive learning, Current Trends in Event-Related Potential Research, (EEG Suppl. 40), 322–327. Lashley, K.S., Ed. (1929), Brain Mechanisms and Intelligence: A Quantitative Study of Injuries to the Brain, University of Chicago Press, Chicago. Lehmann, D. (1989), From mapping to the analysis and interpretation of EEG/EP maps, in Maurer, K., Ed., Topographic Brain Mapping of EEG and Evoked Potentials, Springer, Berlin, 53–75. Leiner, H.C., Leiner, A.L., and Dow, R.S. (1989), Reappraising the cerebellum: what does the hindbrain contribute to the forebrain? Behavioral Neuroscience, 103: 998–1008. Leung, L.S. (1992), Fast (beta), rhythms in the hippocampus: a review, Hippocampus, 2: 93–98. Libet, B. (1991), Control of the transition from sensory detection to sensory awarenes in man by the duration of a thalamic stimulus, Brain, 114: 1731–1757. Lieb, J., Sclabassi, R., Crandall, P., and Buchness, R. (1974), Comparison of the action of diazepam and phenobarbital using EEG-derived power spectra obtained from temporal lobe epileptics, Neuropharmacology, 13: 769–784. Llinás, R.R. (1988), The intrinsic electrophysiological properties of mammalian neurons: insights into central nervous system function, Science, 242: 1654–1664. Llinás, R.R. and Ribary, U. (1992), Rostrocaudal scan in the human brain: a global characteristic of the 40 Hz response during sensory input, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 230 Wednesday, May 12, 2004 8:23 PM

MacKay, D.M. (1970), Perception and brain function, in The Neurosciences, Rockefeller University Press, New York, 303–316. Makeig, S., Westerfield, M., Jung, T.P., Enghoff, S., Townsend, J., Courchesne, E., and Sejnowski, TJ. (2002), Dynamic brain sources of visual evoked responses, Science, 295: 690–694. Maltseva, I., Geissler, H.G., and Basar, ¸ E. (2000), Alpha oscillations as an indicator of dynamic memory operations: anticipation of omitted stimuli, International Journal of Psychophysiology, 36: 185–197. Maras, L., Palejev, G., and Radil, T. (1990), Dynamics of evoked responses according to defined patterns of EEG activity, IBRO International Symposium: Mathematical Approaches to Brain Functioning Diagnostics, Prague. McCallum, W.C. (1980), Brain slow potential changes elicited by missing stimuli and by externally paced voluntary responses, Biological Psychology, 11: 7–19. McClelland, J.L., Rumelhart, D.E., and PDP research group (1986), Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vol. 2, Psychological and Biological Models, MIT Press, Cambridge, MA. Mesulam, M.M. (1990), Large scale neurocognitive networks and distributed processing for attention, language, and memory, Annals of Neurology, 28: 597–613. Mesulam, M.M. (1994), Neurocognitive networks and selectively distributed processing, Review of Neurology: Paris, 150: 564–569. Mesulam, M.M., Van Hoesen, G.W., Pandya, D.N., and Geschwind, N. (1977), Limbic and sensory connections of the inferior parietal lobule (area PG), in the rhesus monkey: a study with a new method for horseradish peroxidase histochemistry, Brain Research, 136: 395–414. Miller, E.K. (2000), The prefrontal cortex and cognitive control, Nature: Reviews in Neuroscience,1: 59–65. Miller, R. (1991), Cortico-Hippocampal Interplay and the Representation of Contexts in the Brain, Springer, Berlin. Miltner, W., Braun, C., Arnold, M., Witte, H., and Taub, E. (1999), Coherence of gamma-band EEG activity as a basis for associative learning, Nature, 397: 434–436. Mizuki, Y., Masotoshi, T., Isozaki, H., Nishijima, H., and Inanaga, K. (1980), Periodic appearance of theta rhythm in the frontal midline area during performance of a mental task, Electroencephalography and Clinical Neurophysiology, 49: 345–351. Mizuki, Y., Takii, O., Nishijima, H., and Inanaga, K. (1983), The relationship between the appearence of Q), and memory function, Electroencephalography and Clinical frontal midline theta activity (FmQ Neurophysiology, 56: 56. Molnar M. (1999), The dimensional complexity of the P3 event-related potential: area-specific and taskdependent features, Clinical Neurophysiology, 110: 31–38. Morris, R.G.M., Garrud, P., Rawlins, J.N.P., and O’Keefe, J. (1982), Place navigation impaired in rats with hippocampal lesions, Nature, 297: 681–683. Moruzzi, G. and Magoun, H.W. (1949), Brainstem reticular formation and activation of the EEG, Electroencephalography and Clinical Neurophysiology, 1: 455–473. Motter, B.C. and Mountcastle, V.B. (1981), The functional properties of the light-sensitive neurons of the posterior prietal cortex studied in waking monkeys: foveal sparing and opponent vector organization, Journal of Neuroscience, 1: 3–26. Mountcastle, V.B.( 1976), The world around us: neural command functions for selective attention, Neuroscience Research Program Bulletin, 14: xii–xiv. Mountcastle, V.B. (1992), Preface, in Basar, ¸ E. and Bullock, T.H. Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 217–231. Mountcastle, V. B. (1998), The cerebral cortex, in Perceptual Neuroscience, Harvard University Press, Cambridge, MA. Mountcastle, V.B., Lynch, J.C., Georgopoulos, A.P., Sakata, H., and Acuna, C. (1975), Posterior association cortex of the monkey: command functions for operations in extrapersonal space, Journal of Neuroscience, 1: 1218–1235. Mundy-Castle, A.C. (1951), Theta and beta rhythm in the electroencephalograms of normal adults, Electroencephalography and Clinical Neurophysiology, 3: 477–486. Näätänen, R. (1988), Implications of ERP data for psychophysiological theories of attention, in Renault, B. et al., Eds., Event-Related Potential Investigations of Cognition, Elsevier, Amsterdam, 117–163. Näätänen, R (1992), Attention and Brain Function, Erlbaum, Hillsdale, NJ. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 231 Wednesday, May 12, 2004 8:23 PM

Nakamura, K., Toshima, T., and Takemura, I. (1986), The comparative and developmental study of auditory information processing in autistic adults, Journal of Autism and Developmental Disorders, 16: 105–118. Narici, L., Pizzella, V., Romani, G.L., Torrioli, G., Traversa, R., and Rossini, P.M. (1990), Evoked alpha and mu rhythm in humans: a neuro-magnetic study, Brain Research, 520: 222–231. Newton Isaac (1687) Philosophiae Naturalis Principia Mathematica. Nicolis, G. and Prigogine, I. (1980), Self-Organization in Nonequilibrium Systems, John Wiley & Sons, New York. Niedermeyer, E. (1993), The normal EEG of the waking adult, in Niedermeyer, E. and Lopes-da Silva, F.H., Eds., Electroencephalography: Basic Principles, Clinical Applications and Related Fields, Williams & Wilkins, Baltimore, 131–152. O Connor, T.A. and Starr, A. (1985), Intracranial potentials correlated with an event-related potential, P300, in the cat, Brain Research, 339: 27–38. O Keefe, J. (1976), Place units in the hippocampus of the freely moving rat, Experimental Neurology, 51: 78–109. O Keefe, J. (1979), A review of the hippocampal place cells, Progress in Neurobiology, 13: 419–439. O Keefe, J. and Dostrovsky, J. (1971), The hippocampus as a spatial map: preliminary evidence from unit activity in the freely-moving rat, Brain Research, 34: 171–175. O Keefe, J. and Nadel, L. (1978), The Hippocampus as a Cognitive Map, Clarendon Press, Oxford. Olds, J., Mink, W.D., and Best, P.J. (1969), Single unit patterns during anticipatory behavior, Electroencephalography and Clinical Neurophysiology, 26: 144–158. Owen, A.M., Evans, A.C., and Petrides, M. (1996), Evidence for a two-stage model of spatial working memory processing within the lateral frontal cortex: a positron emission tomography study, Cerebral Cortex, 6: 31–38. Paller, K.A., Zola-Morgan, S., Squire, L.R., and Hillyard, S.A. (1988), P3–like brain waves in normal monkeys and monkeys with medial temporal lesions, Behavioral Neuroscience, 102: 714–725. Pantev, C., Makeig, S., Hoke, M., Galambos, R., Hampson, S., and Gallen, C. (1991), Human auditory evoked gamma-band magnetic fields, Proceedings of the National Academy of Science of the USA, 88: 8996–9000. Pavlov, I.P. (1927), Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex, Oxford University Press, London. Perkel, D.H. and Bullock, T.H. (1968), Neural coding, Neuroscience Research Progress Bulletin, 6: 221–348. Petersen, I. and Eeg-Olofsson, O. (1971), The development of the electroencephalogram in normal children from the age of 1 through 15 years, Neuropädiatrie, 3: 247–304. Petsche, H. and Etlinger, S.C. (1998), EEG and Thinking, Verlagder Osterreichschen Akademie der Wissenschaften, Wien, 377. Petsche, H., Rappelsberger, P., and Pockberger, H. (1987), EEG Veränderungen beim Lesen, in Weinmann, H.M., Ed., Zugang zum Verständnis höherer Hirnfunktionen durch das EEG, Zuckschwerdt, München, 59–74. Petsche, H., Stumpf, C., and Gogolak, G. (1962), The significance of the rabbit septum as a relay station between the midbrain and the hippocampus 1. The control of hippocampal arousal activity by the septum cells, Electroencephalography and Clinical Neurophysiology, 14: 202–211. Pfurtscheller, G., Neuper, C., Andrew, C., and Edlinger, G. (1997), Hand and foot area mu rhythms, International Journal of Psychophysiology, 26: 121–135. Pineda, J.A., Foote, S.L., and Neville, H.J. (1987), Long latency event-related potentials in squirrel monkeys: further characterization of wave form morphology, topographic and functional properties, Electroencephalography and Clinical Neurophysiology, 67: 77–90. Pinsker, H., Kupfermann, I., Castellucci, V.F., and Kandel, E.R. (1970), Habituation and dishabituation of the gill-withdrawal reflex in Apbeia, Science, 167: 1740–1742. Polich, J. and Kok, A. (1995), Cognitive and biological determinants of P300: an integrative review, Biological Psychology, 41: 103–146. ´ B., Almasy, L., Edenberg, H., Wang, K., Chorlian, D.B., Foroud, T., Goate, A., Rice, J.P., O’Connor, Porjesz, S., Rohrbaugh, J., Kuperman, S., Bauer, L., Crowe, R.R., Schuckit, M.A., Hesselbrock, V., Conneally, P.M., Tischfield, J.A., Li, T.K., Reich, T., and Begleiter, H. (2002), Linkage disequilibrium between the beta frequency of the human EEG and a GABAa receptor gene locus, PNAS, 99: 3729–3733. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 232 Wednesday, May 12, 2004 8:23 PM

Posner, M.I. and Petersen, S.E. (1990), The attention system of the human brain, Annual Review of Neuroscience, 13: 25–42. Pratt, H., Michalewski, H.J., Barrett, G., and Starr, A. (1989), Brain potentials in a memory-scanning task I. Modality and task effects on potentials to the probes, Electroencephalography and Clinical Neurophysiology, 72: 407–421. Pribram, K.H. (1963), The new neurology: memory, novelty, thought and choice, in Pribram, K.H., Ed., Brain and Behaviour 3, Memory Mechanisms (1969), Penguin Education, Sydney. Prigogine, I. (1980), From Being to Becoming: Time and Complexity in the Physical Sciences, W.H. Freeman, San Francisco. Pulvermuller, F., Preissl, H., Lutzenberger, W., and Birbaumer, N. (1996), Brain rhythms of language: nouns versus verbs, European Journal of Neuroscience, 8: 937–941. Quiroga, R.Q. and Schürmann, M. (1998), Functions and sources of event-related EEG alpha oscillations studied with the wavelet transform, Clinical Neurophysiology, 110: 643–654. Quiroga, R.Q., Rosso, O.A., and Basar, ¸ E. (1999), Wavelet entropy: a measure of order in evoked potentials, Electrencephalography and Clinical Neurophysiology, 49: 299–303. Quiroga, R.Q., Sakowitz, O.W., Basar, ¸ E., and Schürmann, M. (2001), Wavelet transform in the analysis of the frequency composition of evoked potentials, Brain Research Protocols, 8: 16–24. Rahn, E. and Basar, ¸ E. (1993a), Prestimulus EEG activity strongly influences the auditory evoked vertex responses: a new method for selective averaging, International Journal of Neuroscience, 69: 207–220. Rahn, E. and Basar, ¸ E. (1993b), Enhancement of visual evoked potentials by stimulation during low prestimulus EEG stages, International Journal of Neuroscience, 72: 123–136. Ramon y Cajal, S. (1911), Histologie du système nerveux de l homme et des vertébrés, Vol. 2, Maloine, Paris. Ranck, J.B., Jr. (1973), Studies on single neurons in dorsal hippocampal formation and septum in unrestrained rats I. Behavioral correlates and firing repertoires, Experimental Neurology, 41: 461–531. ´ B., Chorlian, D.B., Wang, K., Jones, K.A., Kuperman, S., Rohrbaugh, J., O’Connor, Rangaswamy, M., Porjesz, S., Bauer, L.O., Reich, T., and Begleiter, H. (2003), Resting EEG in offspring of male alcoholics: beta frequencies, International Journal of Psychophysiology, in press. Ratcliffe, R. (1990), Connectionist models of recognition memory: constraints imposed by learning and forgetting functions, Psychological Review, 97: 285–308. Regan, D. (1989), Human Brain Electrophysiology: Evoked Potentials and Evoked Magnetic Fields in Science and Medicine, Elsevier, Amsterdam. Rémond, A. and Lesèvre, N. (1957), Remarques sur l activité cérébrale des sujets normaux, in Fischgold, H. and Gastaut, H., Eds., Conditionnement et réactivité en electroencephalographie, Electroencephalography and Clinical Neurophysiology, Suppl. 6: 235–255. Ritter, W., Vaughan, H.G., and Costa, L.D. (1968), Orienting and habituation to auditory stimuli: a study of short term changes in averaged evoked responses, Electroencephalography and Clinical Neurophysiology, 25: 550–556. Romani, A., Callieco, R., and Cosi, V. (1988), Prestimulus spectral patterns and the evoked auditory vertex response, Electroencephalography and Clinical Neurophysiology, 70: 270–272. Röschke, J. and Basar, ¸ E. (1988), The EEG is not a simple noise: strange attractors in intracranial structures, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing by the Brain, Springer, Berlin, 203–216. Röschke, J., Mann, K., Riemann, D., Frank, C., and Fell, J. (1995), Sequential analysis of the brain’s transfer properties during consecutive REM episodes, Electroencephalography and Clinical Neurophysiology, 96: 390–397. Rose, S. (1997), Time, 149, 18: 39–45. Rosen, R. (1969), Hierarchical organization in automata theoretic models of the central nervous system, in Leibovic, K.N., Ed., Information Processing in the Nervous System, Springer, Berlin. Rosso, O.A. and María, L. (2002), Characterization of time dynamical evolution of electroencephalographic epileptic records, Physica A: Statistical Mechanics and Its Applications, 312: 469–504. Rosso, O.A., Blanco, S., Yordanova, J., Kolev, V., Figliola, A., Schürmann, M., and Basar, ¸ E. (2001), Wavelet entropy: a new tool for analysis of short duration brain electrical signals, Journal of Neuroscience Methods, 105: 65–75. Rosso, O.A., Martin, M.T., and Plastino, A. (2002), Brain electrical activity analysis using wavelet-based informational tools I, Physica A: Statistical Mechanics and Its Applications, 313: 587–608. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 233 Wednesday, May 12, 2004 8:23 PM

Rosso, O.A., Blanco, S., and Rabinowicz, A. (2003), Wavelet analysis of generalized tonic–clonic epileptic seizures, in Signal Processing (in press; available online). Rosso, O.A., Martin, M.T., and Plastino, A. (2003), Brain electrical activity analysis using wavelet-based informational tools II, Physica A: Statistical Mechanics and Its Applications, 320: 497–511. Ruchkin, D.S. and Sutton, S. (1983), Positive slow wave and P300: association and disassociation, Tutorials in ERP Research: Endogenous and Clinical Neurophysiology, 25: 550–556. Rumelhart, D.E. and McClelland, J.L., Eds. (1986), Parallel Distributed Processing: Explorations in the Microstructures of Cognition, Vol. 1, Foundations, MIT Press, Cambridge, MA. Saermark, K., Mikkelsen, K.B., and Basar, ¸ E. (1992), Magnetoencephalographic evidence for induced rhythms, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 129–145. Sakowitz, O. W., Schürmann, M., and Basar, ¸ E. (2000), Oscillatory frontal theta responses are increased upon bisensory stimulation, Clinical Neurophysiology, 111: 884–893. Sakowitz, O. W., Quiroga, R. Q., Schürmann, M., and Basar, ¸ E. (2001), Bisensory stimulation increases gamma responses over multiple cortical regions, Cognitive Brain Research, 11: 267–279. Sannita, W.G. (2000), Stimulus- specific oscillatory responses of the brain: a time/frequency related coding process, Clinical Neurophysiology, 111: 565–583. Sanquist, Th.F., Rohrbaugh, J.W., Syndulko, K., and Lindsley, D.B. (1980), Electrocortical signs of levels of processing: perceptual analysis and recognition memory, Psychophysiology, 17: 568–576. Sato, K. (1963), On the linear model of the brain activity in electroencephalographic potentials, Folia Psychiatrica et Neurologica Japonica, 17: 156–166. Sato, K., Kitajima, H., Mimura, K., Hirota, N., Tagawa, Y., and Ochi, N. (1971), Cerebral visual evoked potentials in relation to EEG, Electroencephalography and Clinical Neurophysiology, 30: 123–128. Sato, K., Ono, K., Chiba, G., and Fukuta, K. (1977), Component activities in the autoregressive activity of physiological systems, International Journal of Neuroscience, 7: 239–249. Sauseng, P., Klimesch, W., Gruber, W., Doppelmayr, M., Stadler, W., and Schabus, M. (2002), The interplay between theta and alpha oscillations in the human electroencephalogram reflects the transfer of information between memory systems, Neuroscience Letter, 324: 121–124. Sayers, B.McA., Beagley, H.A., and Henshall, W.R. (1974b), The mechanism of auditory evoked EEG responses, Nature, 274: 481–483. Sayers, B.McA., Beagley, H.A., and Riha, J. (1979), Pattern analysis of auditory evoked EEG potentials, Audiology, 18: 1–16. Scheffers, M.K., and Johnson, R., Jr. (1994), Recognition memory and search for attended letters: an eventrelated potential analysis, Journal of Psychophysiology, 8: 328–347. Schürmann, M. and Basar, ¸ E. (1994), Topography of alpha and theta oscillatory responses upon auditory and visual stimuli in humans, Biological Cybernetics, 72: 161–174. ˘ Schürmann, M., Basar ¸ -Eroglu, C., Rahn, E., Braasch, M., Dössel, O., Fuchs, M., and Basar, ¸ E. (1992), A comparative study of alpha responses in human MEG temporoparietal and occipital recordings and cat intracranial EEG recordings, Proceedings of the IEEE symposium on Neuroscience and Technology, Lyon, 132–137. ˘ Schürmann, M., Basar ¸ -Eroglu, C., Kolev, V., and Basar, ¸ E. (1995), A new metric for analyzing single-trial event-related potentials (ERPs): application to human visual P300 delta response, Neuroscience Letter, 197: 167–170. ˘ Schürmann, M., Basar ¸ -Eroglu, C., and Basar, ¸ E. (1997a), A possible role of evoked alpha in primary sensory processing: common properties of cat intracranial recordings and human EEG and MEG, International Journal of Psychophysiology, 26: 149–170. ˘ Schürmann, M., Basar ¸ -Eroglu, C., and Basar, ¸ E. (1997b), Gamma responses in the EEG: elementary signals with multiple functional correlates, NeuroReport, 8: 1793–1796. ˘ Schürmann, M., Demiralp, T., Basar, ¸ E., and Basar ¸ -Eroglu, C. (2000), Electroencephalogram alpha (8–15 Hz), responses to visual stimuli in cat cortex, thalamus, and hippocampus: a distributed alpha network? Neuroscience Letter, 292, 175–178. Schütt; A. and Basar, ¸ E. (1992a), The effects of acetylcholine, dopamine and noradrenaline on the visceral ganglion of Helix pomatia II. Stimulus evoked field potentials, Comparative Biochemistry and Physiology, 102C: 169–176.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 234 Wednesday, May 12, 2004 8:23 PM

Schütt, A., Basar, ¸ E., and Bullock, T.H. (1992b), The effects of acetylcholine, dopamine and noradrenaline on the visceral ganglion of Helix pomatia I. Ongoing compound field potentials of low frequencies, Comparative Biochemistry and Physiology, 102C: 159–168. Schütt, A., Basar, ¸ E., and Bullock, T.H. (1999), Power spectra of ongoing activity of the snail brain can discriminate odorants, Comparative Biochemistry and Physiology, 123A: 95–110. Segal, M. and Olds, J. (1972), Behavior of units in hippocampal circuit of the rat during learning, Journal of Neurophysiology, 35: 680–690. Seung, H.S. (2000), Half a century of Hebb, Nature Neuroscience, 3: 1166–1167. Shallice, T. and Warrington, E.K. (1970), Independent functioning of verbal memory stores: a neuropsychological study, Quarterly Journal of Experimental Psychology, 22: 261–273. Shannon, C.E. and Weaver, W. (1962), A Mathematical Theory of Communication Bell Systems, 9th ed., University of Illinois Press, Champaign–Urbana. Sheer, D.E. (1976), Focused arousal and 40 Hz EEG, in Knight, R.M. and Bakker, D.J., Eds., The Neuropsychology of Learning Disorders, University Park Press, Baltimore, 71–87. Shenoy, K.V., Kaufinan, J., McGrann, J.V., and Shaw, G.L. (1993), Learning by selection in the trion model of cortical organization, Cerebral Cortex, 3: 239–248. Shepherd, G.M. (1988), Neurobiology, Oxford University Press, New York. Sherrington, C., Ed. (1948), The Integrative Action of the Nervous System, Cambridge University Press, Cambridge. Shiffrin, R.M. and Geisler, W.S. (1973), Visual recognition in a theory of information processing, in Solso, R.L., Ed., Contemporary Issues in Cognitive Psychology, Loyola Symposium, Washington, D.C., 53–101. Silva, L.R., Amitai, Y., and Connors, B.G. (1991), Intrinsic oscillations of neocortex generated by layer 5 pyramidal neurons, Science, 251: 432–435. Simson, R., Vaughan, H.G., and Ritter, W. (1976), The scalp topography of potentials associated with missing visual or auditory stimuli, Electroencephalography and Clinical Neurophysiology, 40: 33–42. Singer, W. (1989), Zur Selbstorganisation kognitiver Strukturen, in Pöppel, E., Ed., Gehirn und Bewußtsein, VCH Verlag, Weinheim, 45–59. Singer, W. (1989), The brain: a self-organizing system, in Klivington, K.A., Ed., The Science of Mind, MIT Press, Cambridge, 174–179. Singer, W. and Gray, C.M. (1995), Visual feature integration and the temporal correlation hypothesis, Annual Review of Neuroscience, 18: 555–586. Skinner, B. F. (1999), The Behavior of Organisms: An Experimental Analysis (reprint), B.F. Skinner Foundation, Cambridge, MA. Smith, E.M., Halgren, E., Sokolik, M., Baudena, P., Musolino, A., Liegeouis-Chauvel, C., and Chauvel. P. (1990), The intracranial topography of the P3 event-related potential elicited during auditory oddball, Electroencephalography and Clinical Neurophysiology, 76: 235–248. Sokolov, E.N. (1975), Neuronal mechanisms of the orienting reflex, in Sokolov, E.N. and Vinogradova, O.S., Eds., Neuronal Mechanisms of the Orienting Reflex, Lawrence Erlbaum, Hillsdale, NJ, 217–238. Sokolov, E.N. (2001), Toward new theories of brain function and brain dynamics, International Journal of Psychophysiology, 39, 87–89. Solodovnikov, V.V. (1960), Introduction to the Statistical Dynamics of Automatic Control Systems, Dover, New York. Squire, L.R. (1992), Declarative and nondeclarative memory: multiple brain systems supporting learning and memory, Journal of Cognitive Neuroscience, 4: 232–243. Squires, N.K., Squires, K.C., and Hillyard, S.A. (1975), Two varieties of long-latency positive waves evoked by unpredictable auditory stimuli in man, Electroencephalography and Clinical Neurophysiology, 38: 387–401. Stampfer, H.G. and Basar, ¸ E. (1985), Does frequency analysis lead to better understanding of human event related potentials, International Journal of Neuroscience, 26: 181–196. Stampfer, H.G. and Basar, ¸ E. (1988), An analysis of preparation and response activity in P300 experiments in humans, in Basar, ¸ E., Ed., Dynamics of Sensory and Cognitive Processing by the Brain, Springer, Berlin, 275–286. Steriade, M., Jones, E.G., and Llinas R.R. (1990a), Thalamic Oscillations and Signaling, John Wiley & Sons, New York. © 2004 by CRC Press, LLC

TF1663_Refs.fm Page 235 Wednesday, May 12, 2004 8:23 PM

Steriade, M., Gloor, P., Llinas, R.R., Lopes da Silva, F.H., and Mesulam, M.M. (1990b), Basic mechanisms of cerebal rhythmic activities, Electroencephalography and Clinical Neurophysiology, 76: 481–508. Steriade, M., Curró Dossi, R., and Paré, D. (1992), Mesopontine cholinergic systems suppress slow rhythms and induce fast oscillations in thalamocortical circuits, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhauser, Boston, 251–268. Stryker, M.P. (1989), Is grandmother an oscillation? Nature, 338: 297–298. Sutton, S., Braren, M., Zubin, J.. and John, E.R. (1965), Evoked potential correlates of stimulus uncertainty, Science, 150: 1187–1188. Swanson, L.W. (1981), A direct projection from Ammon’s horn to the prefrontal cortex in the rat, Brain Research, 217: 150–154. Szenthágothai, J. (1983), The local neuronal apparatus of the cerebral cortex, in Buser, P. and Rougeul-Buser, A., Eds., Cerebral Correlates of Conscious Experience, Elsevier, Amsterdam. Takasaka, Y. (1985), Expectancy-related cerebral potentials associated with voluntary time estimation and omitted stimulus, Folia Psychiatrica et Neurologica Japonica, 39: 167–172. Tallon-Baudry, C., Bertrand, O., Peronnet, F., and Pernier, J. (1998), Induced gamma-band activity during the delay of a visual short-term memory task in humans, Journal of Neuroscience, 18: 4244–4254. Tesche, C. D. and Karhu, J. (2000), Theta oscillations index human hippocampal activation during a working memory task, Proceedings of the National Academy of Science of the USA, 97: 919–924. Thompson, R.F. (1986), The neurobiology of learning and memory, Science, 233: 941–947. Thompson, R.F. (1990), Neural mechanisms of classical conditioning in mammals, Philosophical Transactions of the Royal Society of London B, 329: 161–170. Tiitinen, H., Sinkkonen, J., Reinikäinen, K., Alho, K., Lävikäinen, J., and Näätänen, R., (1993), Selective attention enhances auditory 40–Hz transient response in humans, Nature, 364: 59–60. Tononi, G., Sporns, O., and Edelman, G.M. (1992), The problem of neural integration: induced rhythms and short-term correlation, in Basar, ¸ E. and Bullock, T.H., Eds., Induced Rhythms in the Brain, Birkhäuser, Boston, 365–393. Tranel, D. and Damasio, A.R. (1995), Neurobiological foundations of human memory, in Baddeley, A.D., Wilson, B.A., and Watts, F.N., Eds., Handbook of Memory Disorders, John Wiley & Sons, New York, 27–50. Tulving, E. (1972), Episodic and semantic memory, in Tulving, E. and Donaldson, W., Eds., Organization of Memory, Academic Press, New York. Tulving, E. and Schacter, D.L. (1990), Priming and human memory systems, Science, 247: 301–306. Ungan, P. and Basar, ¸ E. (1976), Comparison of Wiener filtering and selective averaging of evoked potentials, Electroencephalography and Clinical Neurophysiology, 40: 516–520. Van der Tweel, L.H. (1961), Some problems in vision regarded with respect to linearity and frequency response, Annals of New York Academy of Science, 89: 829–856. Vanderwolf, C.H. (1969), Hippocampal electrical activity and voluntary movement in the rat, Electroencephalography and Clinical Neurophysiology, 26: 407–418. Vanderwolf, C.H., Kramis, R., Gillespie, L.A., and Bland, B.H. (1975), Hippocampal rhythmic slow activity and neocortical low-voltage fast activity: relations to behavior, in: Isaacson, R.L. and Pribram, K.H., Eds., The Hippocampus, Plenum Press, New York, 101–128. Varela, F., Lachaux, J.P., Rodriguez, E., and Martinerie, J. (2001), The brainweb: phase syncronization and large-scale integration, Nature: Reviews in Neuroscience, 2: 229–232. Vaughan, H.G. and Ritter, W. (1970), The sources of auditory evoked responses recorded from the human scalp, Electroencephalography and Clinical Neurophysiology, 28: 360–367. Vinogradova, O.S. (1970), Registration of information and the limbic system, in Horn, G. and Hinde, R.A., Eds., Short-Term Changes in Neural Activity and Behaviour, Cambridge University Press, Cambridge, 95–148. Vinogradova, O.S. (1975), The hippocampus and the orienting reflex, in Sokolov, E.N. and Vinogradova, O.S., Eds., Neuronal Mechanisms of the Orienting Reflex, John Wiley & Sons, New York, 128–154. Vinogradova, O.S. and Dudaeva, K.I. (1972), Comparator function of the hippocampus, Dokady. Akademii Nauk SSSR, 202: 486–489. Vinogradova, O.S. and Zolotukhina, L.I. (1972), Sensory characteristics of the neurons in the medial and lateral septal nuclei, Zhurnal Vysshei Nervnoi Deyatel’nosti, 22: 1260–1269.

© 2004 by CRC Press, LLC

TF1663_Refs.fm Page 236 Wednesday, May 12, 2004 8:23 PM

Virginia, F. and Milner B. (1990a), The relationship of working memory to the immediate recall of stories following unilateral temporal or frontal lobectomy, Neuropsychologia, 28: 121–135. Virginia, F. and Milner B. (1990b), The role of the left hippocampal region in the acquisition and retention of story content, Neuropsychologia, 28: 349–359. von Neumann, J. and Burks, A.W. (1966), Theory of Self-Reproducing Automata, University of Illinois Press, Champaign–Urbana. von Stein, A. and Sarnthein, J. (2000), Different frequencies for different scales of cortical integration: from local gamma to long range alpha/theta synchronization, International Journal of Psychophysiology, 38: 301–313. Walter, D.O., Etevenon, P., Pidoux, B., Tortrat, D., and Guillou, S. (1984), Computerized topo-EEG spectral maps: difficulties and perspectives, Neuropsychobiology, 11: 264–272. Wastell, D.G. (1979), The application of low-pass linear filters to evoked potential data: filtering without phase distortion, Electroencephalography and Clinical Neurophysiology, 355–356. Westphal, K.P., Grözinger, B., Diekmann, V., Scherb, W., Reeß, J., Leibing, U., and Kornhuber, H.H. (1990), Slower theta activity over the midfrontal cortex in schizophrenic patients, Acta Psychiatrica Scandinavica, 81: 132–138. Wilder, M.B., Farley, G.R., and Starr, A. (1981), Endogenous late positive component of the evoked potential in cats corresponding to P300 in humans, Science, 211: 605–607. Woods, D.L. (1990), The physiological basis of selective attention: implication of event-related potential studies, in Rohrbaugh, J.W., Parasuraman, R., and Johnson, R.W., Jr., Eds., Event-Related Brain Potentials: Basic Issues and Applications, Oxford University Press, Oxford, 178–209. Yordanova, J. and Kolev, V. (1998), Single-sweep analysis of the theta frequency band during an auditory oddball task, Psychophysiology, 35: 116–126. Yordanova, J., Kolev, V., and Demiralp, T. (1997), The phase-locking of auditory gamma band responses in humans is sensitive to task processing, Neuroreport, 8: 3999–4004. Yordanova, J., Kolev, V., Rosso, O.A., Schürmann, M., Sakowitz, O.W., Özgören, M., and Basar, ¸ E. (2002), Wavelet entropy analysis of event-related potentials indicates modality-independent theta dominance, Journal of Neuroscience Methods, 117: 99.

© 2004 by CRC Press, LLC

TF1663__Epilogue.fm Page 217 Wednesday, May 12, 2004 8:22 PM

Epilogue: From EEG–Brain Dynamics to Memory–Brain Dynamics My monograph titled EEG–Brain Dynamics (Basar ¸ , 1980) has been, according to several critical reviews, one of the milestones in the research area of oscillatory dynamics considering the EEG as one of most important functional signals from the brain. As one of the essential ideas of this 1980 book, the EEG was considered as a cognitive and memory-related signal par excellence. The Brain Dynamics Research Program I proposed in 1976 (see also Chapter 1) was applied in experiments on animal and human brains. Since then, the fundamental association of EEG oscillations with memory and the role of memory in brain dynamics have become central themes. Accordingly, memory replaced EEG in the title of this book. In the past 20 years, the concepts and methods of this research program were applied partially or as an ensemble by several research groups. Some of them underlined the same views and applied similar methods independently or simply ignored the scope of EEG–Brain Dynamics. Herculean efforts worldwide were necessary to develop this new trend of neuroscience. Before my 1980 predictions were almost completely realized and extended, a quiet revolution took place in the functional interpretation of EEG–brain oscillations. This revolution was described extensively in my 1998 and 1999 volumes titled Brain Function and Oscillations. In the concluding chapter of my 1980 monograph, I stated: The newer, more sophisticated theory is better than its predecessor because it gives a good description of more expended domains of science, or a more accurate description of the same domain, or both. The correspondence between the newer theory and its predecessor (a) gives one the power to recover the older theory from the newer; (b) can be exhibited by straightforward mathematics; and (c) according to the historical records, often guided the development of the newer theory.

The understanding of EEG–Brain Dynamics presented in the 1980 volume was mainly based on the dynamics of neural populations under normal conditions of health and behavior, in the human and animal brain. Accordingly, my next comment was that, “The theories presented herein will certainly be recovered from studies of smaller neural populations, and future theories based on (a) neurodynamics, in which the dynamic features of single neurons [and] (b) EEG–ERP dynamics, in which the dynamic features of EEG and ERPs under normal and pathological conditions of behavior are comparatively studied. Twenty-five years have passed since the writing of these statements and the EEG renaissance (Basar ¸ , 1997) and the time is ripe to develop a new construct or devise a new theory to rethink the predecessors. New scientific developments are evolving very quickly and for this reason an attempt to establish a new framework seems necessary. It is my hope that this book will launch a new study framework that links oscillatory brain activity with the concept of dynamic memory leading to a dynamic APLR alliance and all brain functions in the whole brain. The states of a dynamic system change continuously. The brain is a dynamic system and evolving memory is an essential and inseparable element of whole-brain work. Further, the new framework incorporates a model of the hierarchy of memories as a continuum in which the transitions and recurrent work of EEG oscillations constitute the major novelties. The framework launched in this book intends to recover the neurons–brain theory outlined in 1999 and open further studies related to memory and whole-brain work (Chapter 9 and Chapter 11). Such new steps can

© 2004 by CRC Press, LLC

TF1663__Epilogue.fm Page 218 Wednesday, May 12, 2004 8:22 PM

be undertaken only by applying the analysis of neural populations, not with studies of single neurons or fMRI. I hope that the new draft of a theory on whole-brain work and memory and related predictions will also be as constructive as its predecessors, EEG–Brain Dynamics (1980) and Brain Function and Oscillations (E. Basar ¸ , 1998; 1999) Can we soon expect a quiet revolution in memory dynamics? Time will tell.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 237 Friday, May 7, 2004 8:47 AM

Abbreviations and Glossary 1.1 ANATOMICAL ABBREVIATIONS CA3 CE GEA HI IC LG MG OC RF SC

Ca3 layer in hippocampus Cerebellum Gyrus ectosylvian anterior, auditory cortex Hippocampus Inferior colliculus Lateral geniculate nucleus Medial geniculate nucleus Occipital cortex, area 17 Reticular formation Superior colliculus

1.2 OTHER ABBREVIATIONS AEP AFC CAP EEG EHF EP ERO ERP FFT fMRI LTM MEF MEG MMN MOR OB RMS STM TRFC VEP WMS 3.ATT

Auditory-evoked potential Amplitude frequency characteristics Combined analysis procedure for determining EEGs and EPs Electroencephalogram Enhancement factor Evoked potential Event-related oscillation; includes ERPs and induced rhythms Event-related potential Fast Fourier transform Functional magnetic resonance imaging Long-term memory Magnetic-evoked field Magnetoencephalography Mismatch negativity Major operating rhythm Oddball Root mean square Short-term memory Transient response–frequency characteristics method Visual-evoked potential Working memory system Third attended signal in the omitted signal paradigm.(last auditory stimulation before omitted one)

1.3 GLOSSARY Active memory We modify the definition of active memory stated by Fuster (1995) as follows: At any given time in the awake organism, a widely distributed and changing representational network is active in the whole brain, that is, the active memory manifested by EEG oscillations.

© 2004 by CRC Press, LLC

TF1663_book.fm Page 238 Friday, May 7, 2004 8:47 AM

Alphas Ensemble of diverse 10-Hz oscillations in the brain. Alpha response Oscillatory component of an evoked potential in the 8- to 13-Hz frequency range (see Chapter 6). Alpha system (selectively distributed) See selectively distributed oscillatory systems (Chapter 6). Amplitude frequency characteristics Spectra of evoked responses in frequency domain potentials (see Appendix 2). Delta response Oscillatory component of an evoked potential in the 0.5- to 3.5-Hz frequency range (see Chapter 6). Enhancement factor In an experimental record of EEG–EP epochs, the enhancement factor c is the ratio of the maximal time-locked response amplitude (max) to the rms value of the spontaneous activity just prior to the stimulus; both signals (spontaneous and evoked activities) are filtered within the same frequency pass bands: c=

max 2 2rms

Episodic representations Neural firing patterns that encode sequences of events that compose a unique, personal experience (Eichenbaum, 2000). Evoked frequency response Dominant maximum in AFC. Evolving memory The process of memory formation; also denoted as evolving memory; probably the most important process during transition from one memory state to another. Feature detectors The primary features of stimuli as heat, force, light, sound, and chemical substances are transduced selectively at the peripheral ends of sets of sensory (afferent) nerve fibers. Different groups of the sensory fibers respond selectively to different forms of impinging energy at lower thresholds than others. This tuning is often called feature detection and is accomplished during evolution of species by the development of specific transducer mechanisms for different forms of energy, either in nerve endings or in complex sensory organs in which afferent fibers terminate. Examples are the mammalian retina and cochlea and the pressure transducers of the primate hand (see Chapter 7; statements of Sokolov, 2001 and Mountcastle, 1998). Gamma response Oscillatory component of an evoked potential in the 30- to 60-Hz frequency range (see Chapter 6). Gamma system (selectively distributed) See selectively distributed oscillatory systems (Chapter 6). Habituation Decrease in behavioral response to a repeated, benign stimulus (Bailey et al., 2000). Internal EPs The rule of excitability is formulated as follows: If a brain structure has spontaneous rhythmic activity in a given frequency channel, then this structure is tuned to the same frequency, and produces internal evoked potentials to internal afferent impulses originating in the CNS or responds in the form of evoked potentials to external sensory stimuli with patterns similar to those of internal evoked potentials. Limbic structures A collection of subcortical structures including the prominent hippocampus and amygdala that are important for processing memory and emotional information. Longer-acting memory An expression introduced in this book as a replacement for longterm memory and intended to differentiate working memory from persistent memory; a new proposition in memory categorization (Figure 9.7 and Figure 9.12); fresh memory traces acquired in everyday experiences are temporarily stored in longer-acting memory before they reach persistent memory level; based on descriptions of memory levels introduced in Chapter 9, persistent memory combines built-in memory with physiological © 2004 by CRC Press, LLC

TF1663_book.fm Page 239 Friday, May 7, 2004 8:47 AM

memory (an ensemble of submemories as echoic memory, iconic memory, olfactory memory, etc.) and stabilized parts of longer-acting memory acquired during a lifetime. Major operating rhythms Some brain rhythms are more distinguished and dominant in comparision to others, for example, the posterior alpha and frontal theta. Multimodal response Neural activity elicited by more than one sensory modality. Phase-locked and nonphase-locked activities Nonphase-locked activities contain evoked oscillations that are not rigidly time-locked to the moment of stimulus delivery, for example, induced alpha, beta, gamma, and other oscillations that may relate to specific aspects of information processing. In the framework of the additive model of evoked potentials, nonphase-locked activities include background EEGs. For analysis of only nonphase-locked or both phase-locked and nonlocked EEG responses, specific approaches have been used. Phase-locked activity is suggested to include all types of event-related potentials. For quantification of phase-locked activity, the averaging procedure is usually applied; phase-locked responses are enhanced and the nonphase-locked ones are attenuated. Place cells Principal hippocampal cells that fire selectively when an animal is in a particular location in its environment. Prepotent responses Reflexive actions, either innate or well established through a great deal of experience (Miller, 2000). Priming Facilitation of recognition, reproduction, or bases in selection of stimuli that have recently been perceived. Procedural memory Represantation of a series of actions or perceptual processing functions that occur unconsciously and typically produce increased speed or accuracy with repetitions (Eichenbaum, 2000). Resonance Response that may be expected of an underdamped system when a periodic signal of a characteristic frequency is applied. The response is characterized by a surprisingly large output amplitude for relatively small input amplitude, (i.e., the gain is large). An illustration is the annoying vibration that occurs in a house when certain periodic stimuli are applied. Selectively distributed oscillatory systems Via combined analyses of EEGs and EPs, we emphasized the functional importance of oscillatory responses in the framework of brain dynamics related to association and long distance communication in the brain. We assumed that alpha, theta, and gamma networks (or systems) are selectively distributed in the brain (for ranges, see Chapters 24 through 26 in Basar ¸ , 1999). We have tentatively assigned functional properties, namely sensory-cognitive functions, to alpha, theta, delta, and gamma resonant responses. A sensory stimulation evokes 10-Hz enhancements in several cortical (primary auditory cortex, primary visual cortex) and subcortical (hippocampus) structures of the brain. The synchronous occurrences of such responses in multiple areas hint at the existence of distributed oscillatory systems and parallel processing. Such diffuse networks would facilitate information transfer according to the general theory of resonance phenomena. Although alpha responses are observable in multiple brain areas, they are markedly dependent on the site of recording. The dependence of the alpha response on the adequacy of the stimulus for the brain area under study hints at a special functional role of alpha responses in primary sensory processing. The term diffuse was used to describe the distributed nature of the frequency responses in the brain. It is not yet not possible to define connections between the elements of these systems by neuron tracking or define the directions of signal flow and exact boundaries of neuronal populations involved. However, rhythmic phenomena in these frequency ranges are not unique features of the observed single subsystem of the brain. Their simultaneous existence in distant brain structures may be a relevant and important point in the description of an integrative neurophysiology. © 2004 by CRC Press, LLC

TF1663_book.fm Page 240 Friday, May 7, 2004 8:47 AM

Semantic knowledge Organization of factual information independent of the specific episodes in which the information was acquired. Synaptic plasticity Change in functional properties of a synapse as a result of use (Bailey et al., 2000). Theta response Oscillatory component of an evoked potential in the 4- to 8-Hz frequency range. Theta system (selectively distributed) See selectively distributed systems (Chapter 6). Top-down Brain signals that convey knowledge derived from prior experience rather than sensory stimulation. TRFC method A Fourier method that determines frequency characteristics from transient response. Wavelet analysis Method of time–frequency analysis; can be used to search and find repeatable and phase-locked signals in a given frequency window. Working memory Representation of items held in consciousness during experiences or after retrieval of memories.

© 2004 by CRC Press, LLC