Research Notes in Neural Computing - Home - Springer978-3-642-84545-1/1.pdf · Research Notes in...
Transcript of Research Notes in Neural Computing - Home - Springer978-3-642-84545-1/1.pdf · Research Notes in...
Research Notes in Neural Computing
Managing Editor Bart Kosko
Editorial Board S. Amari M.A. Arbib C. von der Malsburg
Advisory Board Y. Abu-Mostafa A.G. Barto E. Bienenstock J. D. Cowan
M. Cynader W. Freeman G. Gross U. an der Heiden M. Hirsch T. Kohonen 1.W. Moore
L. Optican A.I. Selverston R. Shapley B. Soffer P. Treleaven W. von Seelen B. Widrow S. Zucker
Michael A. Arbib Jorg-Peter Ewert Editors
Visual Structures and
Integrated Functions
With 174 Illustrations
Springer-Verlag Berlin Heidelberg New York
London Paris Tokyo Hong Kong Barcelona
Budapest
Michael A. Arbib Center for Neural Engineering University of Southern California Los Angeles. CA 90089-2520 USA
Managing Editor
Bart Kosko Engineering Image Processing Institute University of Southern California University Park Los Angeles. CA 90089-0782 USA
Jorg-Peter Ewert Universitat Kassel. GhK FB 19 - Abt. Neurobiologie Heinrich-Plett-Str. 40 W-3500 Kassel FRG
ISBN -13 :978-3-540-54241-4 e-ISBN-13 :978-3-642-84545-1
DOl: 10.1007/978-3-642-84545-1
Library of Congre;,;, Cataloging-in-PublIcation Data Visual structures and integrated functions I edited by Michael A. Arblb and lorg-Peter Ewert. -(Research note;, in neural computing: v. 3) "Papers presented at the Workshop on Vi;,ual Structure;, and Integrated Functions held at the University of Southern Callforllla 111 Lm. Angele;, on Augu;,t 8-10. 1990"--Pref.
1. Visual pathways--Congress. 2. Phy;,iology. Comparative-Congre;,;,. I. Arbib. Michael A. II. Ewert. Jorg-Peter. 1938 III. Workshop on Visual Structures and Integrated FunctlOm ([990: UllIver;,ity of Southern Ca[iforllla) IV. Serie;,. QP475.V6 [99[ 596'.OI823-dc20 9[-223[2
Thi;, work i;, subject to copyright. All nghts are re;,erved, whether the whole or part of the matenall;, concerned, specifically the rights of tramlatlOn, reprinting, re-use of illustrations, recitatIOn, broadcasting, reproductIOn on microfilm;, or in other ways, and storage in data bank;,. DuplicatIOn of thl;, publication or parts thereof IS only permitted under the provislOm of the German Copyright Law of September 9, 1985,111 its current ver;,lOn, and a copyright fee must always be paid. ViolatIOn;, fallunder the prosecution act of the German Copynght Law.
© Spnnger·Verlag Berlin Heidelberg [99[ Typesetting: camera-ready by author;,
The use of regi;,tered names, trademark;" etc. in tim publIcation does not imply, even in the ab;,ence of a specifIC statement, that such names are exempt from the relevant protective law;, and regulations and therefore free for general use.
33/3140 - 5432 [0 - Printed on aCid-free paper
Preface
This volume integrates theory and experiment to place the study of vision
within the context of the action systems which use visual information. This theme
is developed by stressing:
(a) The importance of situating anyone part of the brain in the context of its
interactions with other parts of the brain in subserving animal behavior. The title
of this volume emphasizes that visual function is to be be viewed in the context of
the integrated functions of the organism.
(b) Both the intrinsic interest of frog and toad as animals in which to study the
neural mechanisms of visuomotor coordination, and the importance of
comparative studies with other organisms so that we may learn from an analysis of
both similarities and differences. The present volume thus supplements our
studies of frog and toad with papers on salamander, bird and reptile, turtle, rat,
gerbil, rabbit, and monkey.
(c) Perhaps most distinctively, the interaction between theory and experiment.
Thus we offer a rich array of models in this volume. High-level schema models
show the basic functional interactions underlying visuomotor coordination, testable
by lesion experiments. Neural network models address data from neurophysiology
and neuroanatomy. We also offer fascinating new data with pointers towards the
way that models must develop to- address them. Although the stress in the
modeling is on Computational Neuroscience (the use of computational methods to
understand neurobiological phenomena), our results have implications for Neural Engineering (the use of ideas inspired, but not necessarily constrained, by the study
of the brain to design highly parallel, often adaptive, machines).
The present volume comprises the papers presented at the "Workshop on
Visual Structures and Integrated Functions" held at the University of Southern
California in Los Angeles on August 8-10, 1990. This Workshop is the fourth in a
series entitled "Visuomotor Coordination in Frog and Toad: Models and
Experiments." These workshops have centered on the evolving body of data about
frog and toad, and on the development of a set of models which together constitute
Rana computatrix, the "frog that computes." The study of Rana computatrix has
VI
implications that go far beyond the study of frogs and toads, per se, and past
workshops - and the present volume - have devoted much space/time to these
implications. It is not so many years ago that the study of invertebrates was regarded
as being, at best, peripheral by the majority of neuroscientists. Yet the increasing
attention to mechanisms of neural function has made Aplysia and other
invertebrates invaluable in the study of basic cellular mechanisms of facilitation,
rhythm generation, and habituation. However, drastic differences in organizational
principles separate the primate brain from the Aplysia nerve net. The study of
"creatures" which "evolve in the computer" can provide opportunities for
understanding organizational principles which are not to be sought solely in terms of cellular mechanisms but in terms of structural constructs (layers and modules),
functional constructs (schemas), and computational strategies (cooperative
computation in neural nets, adaptation, etc.).
As organizers of the meeting, we invited a group of scientists who could between
them address the issues (a) to (c) above of our continuing scientific enterprise. They
responded with lively talks, which generated much discussion to provide
"connective tissue" which greatly strengthened the meeting. To help the reader
gain an understanding of the connection between the papers and the broader
enterprise of which they are a part, we have provided two opening perspectives,
with Arbib reviewing the "Neural Mechanisms of Visuomotor Coordination: The
Evolution of Rana computatrix," while Ewert offers "A Prospectus for the Fruitful
Interaction Between Neuroethology and Neural Engineering." Following these, the
workshop papers have been grouped into sections, each with a unifying theme as set
forth in the following paragraphs:
From the Retina to the Brain: Jeffrey L. Teeters, Frank H. Eeckman, and Frank S.
Werblin build on Teeters' work in developing a leaky integrator model of cells in
frog retina to show how biophysical modeling can be coupled to experiments to help
understand change-sensitive inhibition in ganglion cells of the salamander retina.
Frederic Gaillard and Rene Garcia discuss properties of retinal cells which suggest
that classical ganglion cell types are best seen as representing peaks in a continuum
rather than discrete cell types, and extends this scheme to cells of the nucleus isthmi.
Thomas J. Anastasio provides further insights into distributed representations in
cell populations by showing how the distributed coding in the vestibular nuclei can
be understod in terms of an adaptive technique for identifying properties of cells
VII
mediating the vestibulo-oculomotor reflex in mammals; while Robert F. Waldeck and Edward Gruberg offer new findings about cells of the nucleus isthmi, studying the effects of optic chasm hemisection on the parsing of visual information.
Approach and Avoidance: Information from the two retinas must be combined and gated in an action-dependent way if it is to serve the needs of the organism. Paul Grobstein has shown a parcellation of tectal output which segregates the heading of a prey object from other data about its position. He further finds that these tectal outputs are coded by overall activity in a pathway, rather than by some variant of retinotopic coding. In their two papers, Michael A. Arbib and Alberto Cobas extend this work by providing models of prey-catching and predator avoidance which stress that, whereas stimulus and response direction are the same for prey, different maps are involved for predator location and escape direction. They then provide models of schema interactions which explain Grobstein's data on the medullary hemifield deficit and suggest new experiments on both approach and avoidance. Jim-Shih Liaw and Arbib then provide neural network models for a number of the schemas involved in predator-avoidance, while David J. Ingle provides exciting new data on "triggering" and "biasing" systems in avoidance behavior. The last two papers in this section introduce a comparative dimension. Paul Dean and Peter Redgrave discuss the involvement of the rat superior colliculus in approach and avoidance behaviors (most mammalian studies .note only its role in approach movements, as in visual saccades) while Colin G. Ellard and Melvyn A. Goodale offer data, on the computation of absolute distance in a visuomotor task by the mongolian gerbil, which is most suggestive for future modeling.
Generating Motor Trajectories: The previous sections show how visual data enters the brain, and give a high level view of the overall interactions involved in processing those data to commit the animal to some form of approach or avoidance behavior. This section looks briefly at the actual generation of motor behavior. Simon Giszter, Ferdinando A. Mussa-Ivaldi and Emilio Bizzi report experiments on the coding of motor space in the frog spinal cord which suggest that a wide variety of stimulation sites code for the movement of the frog's leg towards some (sitedependent) equilibrium situation. Reza Shadmehr's theoretical analysis of movement generation with kinematically redundant biological limbs provides a general limb control strategy consistent with this "equilibrium point hypothesis."
VIII
Finally, Ananda Weerasuriya focuses on non-limb movements by analyzing data on
motor pattern generators in anuran prey capture and providing pointers to neural
modeling of their interactions.
From Tectum to Forebrain: In many studies of visumotor coordination in frog and
toad, the central role is given to the tectum and its close associates the pretectum
and nucleus isthmi. Here we see that forebrain structures must be taken into
account as well, and we complement studies of frog and toad with comparative
studies of turtles, birds, and monkeys. Ingle explores the role of frog striatum in frog
spatial memory, and looks at mammalian homologies which are explored in greater
detail in the model by Peter F. Dominey and Arbib of the way in which many brain
regions interact in generation of delayed and multiple saccades in primates. Ewert,
N. Matsumoto, W.W. Schwippert and T.W. Beneke extend the data base on the
interactions of tectum and forebrain in toads by providing intracellular studies on
how striato-pretecto-tectal connections provide a substrate for arousing the toad's
response to prey. Philip S. Ulinski, Linda J. Larson-Prior and N. Traverse Slater
offer insights for comparative modeling with data on cortical circuitry underlying
visual motion analysis in turtles, while Toru Shimizu and Harvey Karten's
comparative study of evolutionary origins of the representation of visual space
show that structures brought together in laminated cortex in mammals may be
segregated in other species raises interesting questions about the utility of layered
structures in the brain. Finally, Edmund T. Rolls completes our comparative
analysis of forebrain mechanisms by with a study of information representation in
temporal lobe visual cortical areas of macaques.
Development, Modulation, Learning, and Habituation: The final section of this
volume considers how the nervous system changes on a variety of time scales.
Sarah Bottjer's study of hormonal regulation of birdsong development provides a
comparative dimension for the study by Albert Herrera and Michael Regnier of
hormonal regulation of behavior in male frogs, in which they present data on how
androgens control the neuromuscular substrate for amplexus (clasping) during the
mating season. Gabor Bartha, Richard F. Thompson and Mark A. Gluck provide a
study of sensorimotor learning and cerebellum which shows how modeling and
data are being integrated in the study of conditioning in rabbits. The volume closes
with three papers which exhibit the successful integration of theory and experiment
IX
in the study of basic learning behaviors in toads. Francisco Cervantes-Perez, Angel
D. Guevara-Pozas and Alberto A. Herrera-Becerra analyze data and modeling of the
modulation of prey-catching behavior. C. Merkel-Harff and Ewert expand the data
base with their study of learning-related modulation of toad's responses to prey by
neural loops involving the forebrain, while DeLiang Wang, Arbib and Ewert report
a dialog between modeling and experimentation in unravelling the dishabituation
hierarchy revealed in visual pattern discrimination in toads.
The meeting and this volume were made possible in part by funds from the
Industrial Affiliates Program of the Center for Neural Engineering (CNE) and from
the Program in Neural, Informational and Behavioral Sciences (NIBS) of the
University of Southern California, as well as from NIH grant lROl-NS24926 to
Michael Arbib. As such, they are part of the continuing dialogue between the study
of living brains and the study of neural network technology maintained by the
faculty and students of CNE and NIBS, as well as part of a continuing pattern of
international cooperation in seeking to integrate theory and experiment in our
efforts to understand the function of the brain in animal and human behavior, and
to probe the implications of that understanding for "perceptual robotics." We
would also like to record our warm thanks to Paulina Baligod-Tagle and her
assistant Hao Cao for all their aid in the organization and conduct of the Workshop.
Michael A. Arbib
J6rg-Peter Ewert
Table of Contents
Overview ..................................................................................................................................... 1
Michael A. Arbib: Neural Mechanisms of Visuomotor Coordination:
The Evolution of Rana computatrix ...................................................................................... 3
Jorg-Peter Ewert: A Prospectus for the Fruitful Interaction Between
Neuroethology and Neural Engineering ........................................................................... 31
From the Retina to the Brain ............................................................................................... 57
Jeffrey L. Teeters, Frank H. Eeckman, and Frank S. Werblin: A Computer Model
to Visualize Change Sensitive Responses in the Salamander Retina ......................... 59
Frederic Gaillard and Rene Garcia: Properties of Retinal and Retino-Tecto-
Isthmo-Tectal Units in Frogs ................................................................................................ 75
Thomas J. Anastasio: Distributed Processing in Vestibulo-Ocular and Other
Oculomotor Subsystems in Monkeys and Cats ................................................................. 95
Robert F. Waldeck and Edward R Gruberg: Optic Chiasm Hemisection and the
Parsing of Visual Information in Frogs ............................................................................. 111
Approach and Avoidance .................................................................................................... 123
Paul Grobstein: Directed Movement in the Frog: A Closer Look at a Central
Representation of Spatial Location .................................................................................... 125
Michael A. Arbib and Alberto Cobas: Prey-Catching and Predator Avoidance 1:
Maps and Schemas ................................................................................................................ 139
Alberto Cobas and Michael A. Arbib: Prey-Catching and Predator Avoidance 2:
Modeling the Medullary Hemifield Deficit ..................................................................... 153
Jim-Shih Liaw and Michael A. Arbib: A Neural Network Model for Response to
Looming Objects by Frog and Toad .................................................................................... 167
David J. Ingle: Control of Frog Evasive Direction: Triggering and Biasing
Systems ...................................................................................................................................... 181
Paul Dean and Peter Redgrave: Approach and Avoidance Systems in the Rat ....... 191
Colin G. Ellard and Melvyn A. Goodale: Computation of Absolute Distance
in the Mongolian Gerbil (Meriones unguiculatus): Depth Algorithms and
Neural Substrates ................................................................................................................... 205
XII
Generating Motor Trajectories ............................................................................................ 221
Simon Giszter, Ferdinando A. Mussa-Ivaldi and Emilio Bizzi: Equilibrium Point
Mechanisms in the Spinal Frog .......................................................................................... 223
Reza Shadmehr: Actuator and Kinematic Redundancy in Biological Motor
Control ..................................................................................................................................... 239
Ananda Weerasuriya: Motor Pattern Generators in Anuran Prey Capture ............. 255
From Tectum to Forebrain ................................................................................................... 271
David J. Ingle: The Striatum and Short-Term Spatial Memory: From Frog to
Man ........................................................................................................................................... 273
Peter F. Dominey and Michael A. Arbib: Multiple Brain Regions Cooperate in
Sequential Saccade Generation ........................................................................................... 281
J.-P. Ewert, N. Matsumoto, W. W. Schwippert and T. W. Beneke: Striato-Pretecto-
Tectal Connections: A Substrate for Arousing the Toad's Response to Prey ............ 297
Philip S. Ulinski, Linda J. Larson-Prior and N. Traverse Slater: Cortical Circuitry
Underlying Visual Motion Analysis in Turtles .............................................................. 307
Toru Shimizu and Harvey J. Karten: Computational Significance of Lamination
of the Telencephalon ............................................................................................................. 325
Edmund T. Rolls: Information Processing in the Temporal Lobe Visual Cortical
Areas of Macaques ................................................................................................................. 339
Development, Modulation, Learning, and Habituation ............................................... 353
Sarah W. Bottjer: Neural Mechanisms of Song Learning in a Passerine Bird ......... 355
Albert A. Herrera and Michael Regnier: Hormonal Regulation of Motor Systems:
How Androgens Control Amplexus (Clasping) in Male Frogs ................................... 369
Gabor T. Bartha, Richard F. Thompson and Mark A. Gluck: Sensorimotor
Learning and the Cerebellum ............................................................................................. 381
Francisco Cervantes-Perez, Angel D. Guevara-Pozas and Alberto A. Herrera-Becerra:
Modulation of Prey-Catching Behavior in Toads: Data and Modeling ...................... 397
C. Merkel-Harff and J.-P. Ewert: Learning-Related Modulation of Toad's
Responses to Prey by Neural Loops Involving the Forebrain ..................................... 417
De Liang Wang, Michael A. Arbib and Jorg-Peter Ewert: Dishabituation
Hierarchies for Visual Pattern Discrimination in Toads: A Dialog Between
Modeling and Experimentation .......................................................................................... 427