Electrophysiological Recording

Techniques 2nd Edition Robert P Vertes

Editor Timothy Allen Editor
Visit to download the full and correct content document:
https://ebookmeta.com/product/electrophysiological-recording-techniques-2nd-edition
-robert-p-vertes-editor-timothy-allen-editor/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

California Slavic Studies Volume VI Robert P. Hughes

(Editor)

https://ebookmeta.com/product/california-slavic-studies-volume-
vi-robert-p-hughes-editor/

Practical Recording Techniques The Step by Step

Approach to Professional Audio Recording 7th Edition
Bruce Bartlett

https://ebookmeta.com/product/practical-recording-techniques-the-
step-by-step-approach-to-professional-audio-recording-7th-
edition-bruce-bartlett/

Encyclopedia of modern optics 2nd Edition Duncan G

Steel Editor Robert D Guenther Editor

https://ebookmeta.com/product/encyclopedia-of-modern-optics-2nd-
edition-duncan-g-steel-editor-robert-d-guenther-editor/

NFPA 30 Flammable and Combustible Liquids Code Handbook

9th Edition Robert P Benedetti Editor

https://ebookmeta.com/product/nfpa-30-flammable-and-combustible-
liquids-code-handbook-9th-edition-robert-p-benedetti-editor/
Modern Techniques for Food Authentication 2nd Edition
Da-Wen Sun (Editor)

https://ebookmeta.com/product/modern-techniques-for-food-
authentication-2nd-edition-da-wen-sun-editor/

Social Selling Techniques to Influence Buyers and

Changemakers 2nd Edition Timothy Hughes

https://ebookmeta.com/product/social-selling-techniques-to-
influence-buyers-and-changemakers-2nd-edition-timothy-hughes/

Cardiology Procedures: A Clinical Primer 2nd Edition

Robert Hendel (Editor)

https://ebookmeta.com/product/cardiology-procedures-a-clinical-
primer-2nd-edition-robert-hendel-editor/

Lipoproteins in Diabetes Mellitus Contemporary Diabetes

2nd Edition Alicia J Jenkins Editor Peter P Toth Editor

https://ebookmeta.com/product/lipoproteins-in-diabetes-mellitus-
contemporary-diabetes-2nd-edition-alicia-j-jenkins-editor-peter-
p-toth-editor/

Woody Allen Interviews Revised and Updated 2nd Edition

Robert E. Kapsis

https://ebookmeta.com/product/woody-allen-interviews-revised-and-
updated-2nd-edition-robert-e-kapsis/
Neuromethods 192

Robert P. Vertes
Timothy A. Allen Editors

Electrophysiological
Recording
Techniques
Second Edition
NEUROMETHODS

Series Editor
Wolfgang Walz
University of Saskatchewan
Saskatoon, SK, Canada

For further volumes:

http://www.springer.com/series/7657
Neuromethods publishes cutting-edge methods and protocols in all areas of neuroscience as
well as translational neurological and mental research. Each volume in the series offers tested
laboratory protocols, step-by-step methods for reproducible lab experiments and addresses
methodological controversies and pitfalls in order to aid neuroscientists in experimentation.
Neuromethods focuses on traditional and emerging topics with wide-ranging implications to
brain function, such as electrophysiology, neuroimaging, behavioral analysis, genomics,
neurodegeneration, translational research and clinical trials. Neuromethods provides investi-
gators and trainees with highly useful compendiums of key strategies and approaches for
successful research in animal and human brain function including translational “bench to
bedside” approaches to mental and neurological diseases.
Electrophysiological Recording
Techniques
Second Edition

Edited by

Robert P. Vertes
Ctr for Complex Systems & Brain Sci, Florida Atlantic University, Boca Raton, FL, USA

Timothy A. Allen
Department of Psychology, Florida International University, Miami, FL, USA
Editors
Robert P. Vertes Timothy A. Allen
Ctr for Complex Systems & Brain Sci Department of Psychology
Florida Atlantic University Florida International University
Boca Raton, FL, USA Miami, FL, USA

ISSN 0893-2336 ISSN 1940-6045 (electronic)

Neuromethods
ISBN 978-1-0716-2630-6 ISBN 978-1-0716-2631-3 (eBook)
https://doi.org/10.1007/978-1-0716-2631-3

© Springer Science+Business Media, LLC, part of Springer Nature 2022

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations
and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to
be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This Humana imprint is published by the registered company Springer Science+Business Media, LLC, part of Springer
Nature.
The registered company address is: 1 New York Plaza, New York, NY 10004, U.S.A.
Preface to the Series

Experimental life sciences have two basic foundations: concepts and tools. The Neuro-
methods series focuses on the tools and techniques unique to the investigation of the
nervous system and excitable cells. It will not, however, shortchange the concept side of
things as care has been taken to integrate these tools within the context of the concepts and
questions under investigation. In this way, the series is unique in that it not only collects
protocols but also includes theoretical background information and critiques which led to
the methods and their development. Thus, it gives the reader a better understanding of the
origin of the techniques and their potential future development. The Neuromethods
publishing program strikes a balance between recent and exciting developments like those
concerning new animal models of disease, imaging, in vivo methods, and more established
techniques, including, for example, immunocytochemistry and electrophysiological tech-
nologies. New trainees in neurosciences still need a sound footing in these older methods in
order to apply a critical approach to their results.
Under the guidance of its founders, Alan Boulton and Glen Baker, the Neuromethods
series has been a success since its first volume published through Humana Press in 1985. The
series continues to flourish through many changes over the years. It is now published under
the umbrella of Springer Protocols. While methods involving brain research have changed a
lot since the series started, the publishing environment and technology have changed even
more radically. Neuromethods has the distinct layout and style of the Springer Protocols
program, designed specifically for readability and ease of reference in a laboratory setting.
The careful application of methods is potentially the most important step in the process
of scientific inquiry. In the past, new methodologies led the way in developing new dis-
ciplines in the biological and medical sciences. For example, physiology emerged out of
anatomy in the nineteenth century by harnessing new methods based on the newly discov-
ered phenomenon of electricity. Nowadays, the relationships between disciplines and meth-
ods are more complex. Methods are now widely shared between disciplines and research
areas. New developments in electronic publishing make it possible for scientists that
encounter new methods to quickly find sources of information electronically. The design
of individual volumes and chapters in this series takes this new access technology into
account. Springer Protocols makes it possible to download single protocols separately. In
addition, Springer makes its print-on-demand technology available globally. A print copy
can therefore be acquired quickly and for a competitive price anywhere in the world.

Saskatoon, SK, Canada Wolfgang Walz

v
Preface

The present volume is a continuation of the previous edition of Electrophysiological Record-

ing Techniques (2011) edited by Robert Vertes and Robert Stackman, Jr. As with the
previous edition, this volume explores various topics, incorporating state-of-the-art electro-
physiological and anatomical methods and their application to the study of several systems of
the brain involved in a range of functions. This edition includes several new chapters and
some chapters with extensive updates which complement and extend the content of the first
edition. For example, in this edition, new chapters emphasize the value and difficulty of
multi-site recordings using depth or surface electrodes. This latter point takes center stage
when considering the innumerable interactions between the many regions of the brain to
support behavior, and the many species needed to identify fundamental neurophysiological
mechanisms. The new addition also contains several chapters addressing the different
electrophysiological techniques used in non-human animals and humans, and discusses
the different advantages and the types of questions each can answer.
First, the new chapter by Choi and Hwang provides a comprehensive account of
electroencephalography (EEG) recordings applied to mouse research with an approach
that mimics the coverage from arrays more typical of a human psychiatric session. Although
EEG has been a feature of clinical neurology for almost 100 years, the authors introduced
several unresolved issues including the source of EEG signals and the biological basis of their
variation in behavior and disease that can be resolved in rodent research models. The authors
discuss sources of noise such as pink noise (1/f) and cable noise that can differentially affect
frequency specific parts of EEG recordings, and describe methods such as the type of
electrode, contacts, spacing, and surgical implant molds that are capable of reducing such
confounds. A procedural approach to EEG analysis is presented including data selection,
artifact removals, and the creation of spatiotemporal topologies.
The new chapter by Ito focuses on methods for high-density multisite thalamic and
cortical recordings allowing measures of interregional synchrony that lend themselves to
examining the neurophysiological mechanisms of complex brain-behavior relationships. As
an example, the chapter presents goal-directed spatial navigation in rodents as a model of
complex information processing in the brain. That is, this cognitive task requires sensation,
spatial representation, memory formation, goal representation, memory retrieval, decision-
making, and action selection and thus involves the interactions of several brain regions. The
author provides detailed and practical laboratory methods for building multisite hyperdrives
capable of targeting two structures in freely behaving rodents and simultaneously recording
the activity of high-density neuronal ensembles and local field potentials (LFPs). The author
further addresses several analytical challenges in multisite recordings with solutions includ-
ing spectral coherence analysis, spike-phase analysis, spike-field coherence, and the compar-
isons of measures across observed behavior conditions.
The chapter by Vivar and van Praag describes procedures for identifying inputs to adult-
born granule cells of the dentate gyrus and for characterizing the properties of these afferent
(projecting) neurons. Specifically, they provide a detailed description of techniques for
labeling adult-born neurons (starter cells) through dual viral injections and the retrograde
trans-synaptic transport of rabies virus from starter cells to monosynaptic inputs (traced

vii
viii Preface

cells) to them. By recoding from “tagged” traced cells (in slices), they could determine the
neurochemical identity and electrophysiological characteristics of neurons projecting to
new-born granule cells. Further, by adjusting the time interval between the first and second
viral injections into starter cells (between 3 and 90 days), they were able to define inputs to
immature versus mature granule cells. Finally, as was pointed out, the method allows for an
examination of the possible effects of various conditions (exercise, learning) or pathological
states (epilepsy) on the organization of afferents to new-born granule cells.
The chapter by Vertes, Linley, and Viena examines the nucleus reuniens (RE) of the
ventral midline thalamus, describing RE circuitry, functional properties, electrophysiological
characteristics, and involvement in CNS disorders. With respect to function, nucleus
reuniens has been shown to serve a critical role in cognitive, affective, and executive
functions. Specifically, alterations of RE significantly disrupt spatial working memory, atten-
tion/attentional set, reversal learning, and fear conditioning/fear extinction. Regarding
electrophysiological characteristics, subsets of RE neurons display spatial properties (head
direction cells, place cells), exhibit trajectory-dependent firing (predicting directional
choices), and discharge at significantly higher rates in waking and/or REM sleep compared
to slow wave sleep. The latter finding suggests that RE neurons may exert arousal or
attentional effects on target structures during waking or REM sleep. Finally, RE has been
linked to the CNS disorders of schizophrenia (SZ) and epilepsy. Disruptions of nucleus
reuniens produce abnormal oscillations between the hippocampus and medial prefrontal
cortex, and associated SZ-like cognitive deficits, whereas the over activation of reuniens
produces pronounced epileptic seizure activity.
In an extension of his chapter in the original volume, Bressler provides a detailed
overview of the neural mechanism responsible for event-related potentials (ERPs). As was
described, ERPs are triggered by external sensory and motor events or by “internal”
spontaneous or self-generated states, and represent dynamical changes in the brain to
internal/external events. Various methods were described for analyzing cortical ERPs, or
extracting their non-varying features from the general field activity. These include time
domain analysis, frequency domain analysis, spatial analysis, and interdependency analy-
sis—with each reflecting distinct aspects of the triggered ERPs. The ERP was compared to
its magnetic correlate, the event-related field (ERF), and as pointed out, for some analyses,
both methods are preferable to imaging techniques, as the time course of changes more
closely mimic the underlying neural events.
The chapter by Pittman-Polletta and Kocsis examines various methods used to assess the
cross-frequency coupling of electrophysiological signals (EEG and LFPs) such as amplitude-
amplitude coupling (AAC) and phase-phase coupling (PPC), but it describes in detail phase-
amplitude coupling (PAC). The PAC involves fluctuations in the amplitude of fast rhythms
dependent on the phase of slower rhythms. The authors provide a detailed description of the
signal processing techniques used to examine and quantify PACs—from the preprocessing
of data through to statistically determining the strength of PAC coupling. As was pointed
out, PAC analysis is a valuable tool for quantifying the relationship between simultaneously
occurring events, reflecting functional interactions between structures (or networks) exhi-
biting oscillations at different frequencies. The authors give examples of the application of
PAC analysis from their work showing interactions between delta/theta and high-frequency
oscillations (HFO) of the cortex and hippocampus as well as between high gamma activity in
the hippocampus (driver) and theta in the cortex (receiver).
The new chapter by Jayachandran and Allen looks at the neurophysiological mechanisms
that encode time and approaches to studying this fundamental aspect of cognition and
 Preface ix

behavior. Studying the neurobiological underpinnings of temporal content raises several

unique issues for rigorous science. For example, time cannot be generated in the lab or easily
isolated from other confounding time-varying covariates. Thus, the authors enumerate
fundamental design considerations to aid rigorous studies of time in a laboratory setting.
Jayachandran and Allen introduce a simple taxonomy for time, and emphasize how individ-
ual processing levels relate to actual versus reconstructed time. They detail methods for two
specific behavioral methods that can be used to study time in memory in conjunction with
in vivo electrophysiological recordings, and emphasize how these designs satisfy their
scientific recommendations. Next, the authors present results from several single-unit
activity (SUA) recording studies that have discovered ramping cells, time cells, and sequence
cells—each of which contribute to temporal processing—and they discuss how these cell
types might combine to support memory for time. Lastly, the authors present a
3D-printable dual-site silicon probe that can be chronically implanted in rats and used to
examine neural ensembles and LFPs across the large system of regions in the brain known to
be critical to encoding temporal content.
The new chapter by Griffin further focuses on multisite recordings emphasizing the
reason for using such difficult techniques, and the barriers to entry for new researchers
interested in large-scale brain networks. Methods for motivating rats are presented with
procedures presented for food restriction and the choice of food rewards, and advice is given
on the care needing to be taken when working with rats in motivated tasks that occur outside
their homecage and with an experimenter. Griffin provides detailed information related to
surgically implanting rats and how to go about their first on-head plugin event, and the time
needed to carefully drive tetrodes. Localizing neurophysiological signatures of hippocampal
layers, such as sharp wave ripples (SWRs), is discussed. The chapter tackles many practical
issues related to the construction and implementation of multisite recordings, and the
choices that can be made in their design related to the specific research question. Lastly,
the author discusses the need to integrate neural recordings with more interventionist
approaches such as lesions, inactivation, chemogenetics, and optogenetics—each providing
their own unique benefits to testing causal contributions to neurophysiological research.
The new chapter by Mattfeld addresses issues in behavior methodologies in neurosci-
ence, a critical topic for this book since the purpose of the brain is to advantage behavior.
The focus is on comparative approaches that have been, and should be, used in neurophysi-
ological studies in the field of behavioral and cognitive neuroscience. Mattfeld highlights the
fact that comparative approaches are frequently touted but rarely executed. The chapter
addresses a wide range of tasks but focuses on one with great comparative success called
conditional associative learning. The author details species-similar and species-specific con-
siderations in implementation and interpretation of this task. For example, the type of cue or
reward used may be different in each species but the underlying process held constant. The
author then addresses comparative loss of function approaches such as animal lesions and
human brain damage, and consideration needed in interpreting different approaches. The
chapter makes a comparison of the many electrophysiological approaches used in animals
and humans from single neuron recordings to LFPs and juxtaposes this with a detailed
account of functional magnetic resonance imaging (fMRI) during conditional associative
learning, in addition to EEG, magnetoencephalography (MEG), and positron emission
tomography (PET).
In the final chapter of this edition, Helfrich provides a comprehensive account of
intracranial electroencephalography (iEEG) performed in patients with drug-resistant epi-
lepsy. The author discusses the advantages from historical approaches and compares
x Preface

intracranial recording approaches in non-human primates (SUA, LFP and iEEG) and
humans (SUA, LFP, and iEEG) to non-invasive approaches used in humans (EEG, MEG,
fMRI, and PET) highlighting that each provides information about particular levels of
analysis, and that the transfer functions between levels are largely unknown. Unlike most
intracranial recordings, iEEGs tend to cover large areas of the brain, and have exceptional
spatial and temporal resolution. The author gets into the details of the physiological basis of
iEEG signal and addresses other approaches including adding arrays for unit recordings. The
author then provides practical methodological and equipment information for using iEEG
in the clinic. The chapter presents analytical approaches such as frequency decomposition
and event-related potentials, and the use of various statistical approaches for drawing
meaningful conclusions. Lastly, Helfrich provides several notes for consideration of issues
that arise in iEEG work in the clinic.
In conclusion, as a continuation of the previous volume, the present edition introduces
and details several recently developed techniques currently used to examine the electrophys-
iological properties of several CNS systems, which affect behavior across a range of species
from rodents to humans. It is our expectation that the present edition would have wide
applicability for basic researchers as well as for clinicians in utilizing some of the principles
and research designs described herein in their research programs.

Boca Raton, FL, USA Robert P. Vertes

Miami, FL, USA Timothy A. Allen
Contents

Preface to the Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
1 High-Density Electroencephalography in Freely Moving Mice. . . . . . . . . . . . . . . . 1
Jee Hyun Choi and Eunjin Hwang
2 Multisite Recording for the Analysis of Information Flow
Between Thalamocortical Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Hiroshi T. Ito
3 Rabies Virus Tracing of Monosynaptic Inputs
to Adult-Born Granule Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Carmen Vivar and Henriette van Praag
4 Nucleus Reuniens: Circuitry, Function, and Dysfunction . . . . . . . . . . . . . . . . . . . . 55
Robert P. Vertes, Stephanie B. Linley, and Tatiana D. Viena
5 Event-Related Potentials of the Cerebral Cortex . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Steven L. Bressler
6 Assessing Neural Circuit Interactions and Dynamics
with Phase-Amplitude Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Ben R. Pittman-Polletta and Bernat Kocsis
7 Candidate Neural Activity for the Encoding of Temporal
Content in Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Maanasa Jayachandran and Timothy A. Allen
8 Multisite Recording During Memory-Guided Behavior. . . . . . . . . . . . . . . . . . . . . . 183
Amy Lynn Griffin
9 Comparative Tasks for Comparative Neurophysiology . . . . . . . . . . . . . . . . . . . . . . . 193
Aaron T. Mattfeld
10 Human Intracranial Cognitive Neurophysiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Randolph F. Helfrich

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247

xi
Contributors

TIMOTHY A. ALLEN • Cognitive Neuroscience Program, Department of Psychology, Florida

International University, Miami, FL, USA; Department of Environmental Health
Sciences, Robert Stempel College of Public Health, Florida International University,
Miami, FL, USA
STEVEN L. BRESSLER • Center for Complex Systems & Brain Sciences, Florida Atlantic
University, Boca Raton, FL, USA
JEE HYUN CHOI • Center for Neural Science, Korea Institute of Science and Technology,
Seoul, South Korea
AMY LYNN GRIFFIN • Department of Psychological and Brain Sciences, University of
Delaware, Newark, DE, USA
RANDOLPH F. HELFRICH • Hertie Institute for Clinical Brain Research, Center for
Neurology, University of Tübingen, Tübingen, Germany
EUNJIN HWANG • Center for Neural Science, Korea Institute of Science and Technology, Seoul,
South Korea
HIROSHI T. ITO • Max Planck Institute for Brain Research, Frankfurt am Main, Germany
MAANASA JAYACHANDRAN • Cognitive Neuroscience Program, Department of Psychology,
Florida International University, Miami, FL, USA
BERNAT KOCSIS • Department of Psychiatry, Beth Israel Deaconess Medical Center/Harvard
Medical School, Boston, MA, USA
STEPHANIE B. LINLEY • Center for Complex Systems and Brain Sciences, Florida Atlantic
University, Boca Raton, FL, USA; Department of Psychology, Florida Atlantic University,
Boca Raton, FL, USA; Department of Psychological Science, University of North Georgia,
Dahlonega, GA, USA
AARON T. MATTFELD • Cognitive Neuroscience Program, Florida International University,
Miami, FL, USA
BEN R. PITTMAN-POLLETTA • Department of Mathematics & Statistics, Boston University,
Boston, MA, USA
HENRIETTE VAN PRAAG • Department of Biomedical Science, Charles E. Schmidt College of
Medicine, and Brain Institute, Florida Atlantic University, Jupiter, FL, USA
ROBERT P. VERTES • Center for Complex Systems and Brain Sciences, Florida Atlantic
University, Boca Raton, FL, USA; Department of Psychology, Florida Atlantic University,
Boca Raton, FL, USA
TATIANA D. VIENA • Department of Psychology, Florida International University, Miami,
FL, USA
CARMEN VIVAR • Laboratory of Neurogenesis and Neuroplasticity, Department of Physiology,
Biophysics and Neuroscience, Center for Research and Advanced Studies of the National
Polytechnic Institute, Mexico City, Mexico

xiii
 Chapter 1

High-Density Electroencephalography in Freely

Moving Mice
Jee Hyun Choi and Eunjin Hwang

Abstract
Electroencephalography (EEG), an electrophysiological monitoring method to record the electrical activity
of the brain, has been one of the most important noninvasive brain imaging tools in psychiatry, but
surprisingly little is known about how the neural correlates of various EEG features are linked to cognition.
Recent advances in neuroscience and related technologies make this an ideal time for new discoveries about
the origin and significance of the contents of EEG. In particular, understanding the molecular and cellular
mechanisms underlying diverse EEG features has been facilitated using mouse models under genetic,
pharmacological, or optogenetic manipulations. A core challenge in mouse EEG was to obtain topographi-
cal neuroimaging that can be compared to human EEG. To overcome this challenge, we have developed a
high-density EEG using a polyimide-based microarray that fits to the mouse skull and can be applied to
various studies with a high spatiotemporal accuracy in free behavioral states. The benefits of mouse high-
density EEG are not only that it provides cross-species neuroimaging data comparable to human EEG but
also that it helps in dissecting enigmatic brain activity by probing the neural substrates of cognition when
combined with optogenetics. The aim of this chapter is to introduce the methodological aspects of high-
density EEG in mice. We explain the electrodes, surgery, recording, and analysis procedures and present
applications in studying the origin of EEG signals. In addition, we point to potential areas where this
technique will provide mechanical insight into circuit dysfunction in major psychiatric conditions.

Key words Electroencephalography (EEG), Mouse, Microelectrode, Topography, Freely moving

mice, Preclinical studies

1 Introduction

Electroencephalography (EEG) and related measures have been

one of the most important neuroimaging tools in human neurosci-
ence and the clinic by providing a topographical mapping of
cortical activities. Topographical features of regional power spectral
density or networks of connectivity strength across different corti-
cal regions have been studied with EEG, and tremendous effort has
been made to understand where EEG signals come from and what
brain function they represent. Recent advances in noninvasive EEG
promoted the understanding of the neurological origin of

Robert P. Vertes and Timothy A. Allen (eds.), Electrophysiological Recording Techniques, Neuromethods, vol. 192,
https://doi.org/10.1007/978-1-0716-2631-3_1, © Springer Science+Business Media, LLC, part of Springer Nature 2022

1
2 Jee Hyun Choi and Eunjin Hwang

particular EEG features [1]. However, investigation of the molecu-

lar and cellular levels of perturbations in humans is greatly limited
due to its invasiveness and safety concerns.
In recent years, a variety of genetic mouse models have allowed
more comprehensive studies on the origin and meaning of neuronal
synchrony and oscillations featured in EEG signals. A thorough
investigation using optogenetic, electrophysiological, or neurolog-
ical intervention has subdued the controversy that EEG rhythms or
particular waveforms are not simple epiphenomena but have neu-
rological origins associated with cognitive or physiological func-
tions [2–5]. However, conventional EEG in mice using only one or
two electrodes is not ideally suited as a translational tool because of
its lack of spatial information. Most EEG phenotypes in human
patients are expected to be reproduced in mouse disease models. In
addition, recent studies found that a composite of neuronal oscilla-
tions at different frequency bands in different cortical regions is
crucial in orchestrating the massive parallel computing of neuronal
units [6]. Therefore, the regionally distinctive oscillatory patterns,
in conjunction with behaviors, remain to be described in mouse
models. The lack of spatial information is a hindrance for mouse
researchers in interpreting the data at the network level or associat-
ing it with the findings from human EEG.
A decade ago, we introduced high-density EEG for freely
moving mice using a polyimide-based microarray to the neurosci-
ence community [7, 8]. The primary purpose of this tool is to
signify EEG as a cross-species neuroimaging tool for comparative
study of neuronal activities in conjunction with cognition and
behaviors. The microarrays were specifically designed to fit the
extracranial surface of the mouse, accommodating 40–60 channels,
and had a thickness of 8–10 μm. The overall profile is shaped like a
pinnate leaf, allowing a secondary probe or optical fibers to intrude
into the deep brain, enabling simultaneous EEG/spike recording
or combination of EEG and optogenetics.
Despite the conceptual advance, there has been a limited
expansion in the research community in using high-density EEG
in mice. There are multiple reasons that researchers hesitate to
adapt high-density EEG in their studies. The major challenge is
the difficulty of analyzing and interpreting complex EEG data. The
montage of mouse high-density EEG is different from that of
humans, and mice have more movement artifacts than humans.
These technical issues require the researcher to put more effort
into EEG preprocessing before using conventional EEG neuroim-
aging platforms. The nonstationary nature of EEG requires an
experimental paradigm to produce equivalent epochs to be aver-
aged to extract only the test-relevant signals, which is more chal-
lenging in mice. In humans, most cognitive paradigms are designed
to make all trials equivalent; however, these repetitive test para-
digms are rare in mice. Most importantly, although mice and
 High-Density Electroencephalography in Freely Moving Mice 3

humans have core nervous systems and share central operational

mechanisms, evident differences exist, particularly in higher-level
cognition. Nonetheless, many fundamental mechanisms of neural
communication are universal across different species, and under-
standing these mechanisms from mice is likely to expand knowl-
edge of the human brain. The purpose of this chapter is to review
the methodology of high-density EEG and note several technical
considerations associated with it. We have been both developing
and recording high-density EEG for over 10 years, and we have
written this chapter based on our experience with the aim of
providing information to investigators who are interested in
obtaining high-density EEG in freely moving mice.

2 Materials

2.1 Technical Acquiring high-density EEG from mice in freely moving states
Challenges in EEG poses a significant technical challenge since the small size of the
Electrodes for Freely mouse brain (~0.5 cm3) limits the space for implantation of many
Moving Mice electrodes and the agile movements of the animal produce signifi-
cant movement artifacts. In the last decades, tremendous advances
in chronic neural recordings have been achieved, particularly in the
fields of implantable silicon probe arrays [9], accelerated by devel-
opment of wireless recording systems. Expanding these technolo-
gies to acquire high-density EEG can take several technical factors
into account but is not straightforward due to the following dis-
tinct features of high-density EEG. First, high-density EEG
requests to collect information from a broad region of cortex.
This implies that the electrodes should be distributed over the
surface of a semi-infinite medium, and the interface condition for
the contact should be equivalent for all EEG electrodes. The most
appropriate surface to apply the electrodes in mice is the skull
surface. This involves building a flexible, dry electrode that can fit
onto the curved skull surface. Second, EEG activities often appear
as oscillations whose frequencies are arranged from low (~1 Hz) to
high (~150 Hz). The serious issue in this broadband signal is the
abundance in the nonneural signal sources – some from physiologi-
cal origins such as heartbeat, respiration, vasomotion, etc., and
some from environmental noise, such as ventilation, acoustic reso-
nance of buildings, line noise, etc. Moreover, 1/f noise, which is
inversely proportional to the frequency, is common in biological
systems affecting the low-frequency regime of EEG. Therefore,
consistent contact/attachment to the skull with high mechanical
stability and effective electrical shielding is needed.
4 Jee Hyun Choi and Eunjin Hwang

2.2 Polyimide-Based To resolve the above issues, we applied a polyimide as the carrier
Microarray for Mouse substrate and fabricated a microarray as depicted in Fig. 1
High-Density EEG [7, 8]. Polyimide has been shown to have superior biocompatibility
and good mechanical properties [10] and has been widely used as a
substrate for flexible neural probes [11]. As the skull surface is
ellipsoidal, the overall structure has a backbone line with multiple
branches on both sides. The thickness is crucial for stable adher-
ence. There is a trade-off in the thickness of microarray: in the
thinner, the yielding rate of production drops, resulting in more
cost per array, whereas in the thicker, it is more easily detached from
the skull. We tested different thicknesses and found that the most
appropriate thickness was 7–8 μm with an acceptable level of up to
10 μm. Figure 1b shows different electrode configurations for
40 and 64 channels. In most cases, the 40-channel configuration
is enough to generate topographical mapping. The 64-channel
configuration is appropriate for somatosensory research, presenting
14 channels on top of the somatosensory cortex. For the electrical
contacts and interconnection lines, we used Pt [7], but later studies
showed that an Ag/AgCl electrode presents similar levels of signal
quality. A series of tests revealed that the most relevant features in
the electrode configuration are the size of electrical contact, the
width of the interconnection line, and the interline spacing. We
found that the adequate impedance for EEG measurements was
obtained with an electrode contact size of ~0.2 mm2. The width of
interconnection lines and the interline spacing are key factors in the
design because they determine the overall size. We found that there
is no significant generation of induced noise or cross-talk for widths
>15 μm and interline spacing >20 μm. With the best configura-
tion, the impedance ranges from 10 to 300 kΩ at 30 Hz when
measured on the skull.

Fig. 1 Polyimide-based microarray for mouse high-density EEG. (a) Picture of a 40-channel microarray. (b)
Configuration of 40-channel (left) and 60-channel (right) microarrays. Inset images are actual microarrays
positioned on a mouse skull. (c) Electrode locations overlaid on a cortical region map
 High-Density Electroencephalography in Freely Moving Mice 5

3 Methods

3.1 Fixation of For a favorable surgery outcome, we suggest using bone cement as
Microarray a fixation material. Different types of commercially available bone
cements are applicable, but an ideal cementing material should be
noncorrosive, tightly filling, and not affected by moisture and
should have long-term stability. A typical cause of failure is the
detachment of the cement from the skull. A couple of anchoring
microscrews on the skull helps in preventing this detachment. Also,
when one applies cement, any debris on the skull surface should be
cleaned. We use a tap-water-soaked cotton swab for the cleaning,
which is used for pressing the microarray as well. We make the
surface wet with tap water when we place and press the microarray
and wait until the surface is fully dry. Placing cement is extremely
important for preventing its detachment from the skull. We suggest
using low-viscosity cement and applying the cement with a brush.
One should check the cement is fully dried before placing the next
cement.

3.2 Recording The measurement principle and instrumentation for mouse EEG
are identical to those for human EEG. The basic principle of EEG
recording is to measure the potential difference between two points
as the direct consequence of the existing electric dipole created by a
synchronously generated postsynaptic potential in the cortex.
These potentials can be measured by a voltage amplifier. Typically,
the microarray electrode has sub-Ω-level resistance, but on the
skull, the impedance level is approximately 200 kΩ when measured
at 30 Hz. Determination of the interface impedance is not trivial
because the interface between the electrode and the skull constitu-
tes a significant resistance and capacitance. In human scalp EEG,
the interface impedance is at the level of ~50 kΩ, and in general, the
difference of impedance between channels does not affect the read-
ings. However, when variation of the interface impedances between
channels is high, we recommend normalizing the signal levels.
Typically, we perform normalization of EEG by dividing the spec-
tral power above 200 Hz in all channels using EEG acquired in the
quiescent moment of mouse. For comparisons between different
mice, normalization of EEG is required. Additionally, an electrode
with an impedance level above 500 kΩ should be excluded from
further analysis.

3.3 Artifacts and Recording extracranial EEG from freely moving mice is subject to
Noises more environmental noise than is present in other electrophysio-
logical modalities. The environmental noise may arise from a variety
of sources, such as the mouse’s own body, instrumentation, or the
experimenter. The waveforms of noises depend on their noise types,
and unfortunately, many noises have waveforms similar to the EEG
6 Jee Hyun Choi and Eunjin Hwang

such as sleep waves, epilepsy, oscillations, etc. Therefore, it is of

utmost importance to create perfect conditions to minimize the
noise. First, it is critical to keep the animal within a closed equipo-
tential surface so that the net electric flux outside the brain is close
to zero. This is performed by shielding the recoding place with a
conducting shell (Faraday cage) and connecting all the electrical
devices, experimenters, and animals to this cage. In chronic record-
ing, a metal floor is preferred to a glass or plastic floor. In acute or
head-fixed recording, the earbars or fixers should be wired to the
Faraday cage to maintain the body and cage equipotential. To
discharge any electric surge, grounding the Faraday cage to the
Earth is helpful, but in a noise-rich environment (e.g., near subway
or in the hospital), a floating ground could be better than earth
ground. Another serious contamination in chronic recording of
mouse EEG is the swing noise generated when the wire bundle
swings as the head moves. We found that the noise comes mostly
from the connection between the wires and microarray connector,
and hence it is suggested to this part rigid to avoid any artifact due
to connector movement. Despite all the above efforts, an ideal
recording without artifacts is pragmatically unachievable. There-
fore, EEG researchers normally perform offline postprocessing to
remove the contamination.

3.4 Analyzing EEG The surface potential produced by a mouse brain has an order of a
few μV, which is a comparable level to the human brain. Basically,
EEG signals are nonstationary data and the operations are done in a
time domain or frequency domain. In the case of averaging multi-
ple events in the time domain, an ergodic property of the signals is
assumed, i.e., all events are equivalent, which may be violated in the
real world. Hence, prior to the data analysis, several experimental
considerations are critical to obtain consistent signals across trials.
For example, cognitively or behaviorally equivalent trials, precise
timing, and uncorrelated noise across different trials are all impor-
tant conditions for a good EEG. In the case of a study of spontane-
ous activities, spectral power or phase-amplitude coupling
properties are analyzed in the frequency domain. In either case,
preprocessing is critical to uncorrelate the noise from the signal.

3.4.1 Preprocessing One of the first steps in EEG analysis is to identify and remove
potentially contaminated signals or artifacts. Mice produce signifi-
cant levels of artifacts, especially in a freely mobile condition, and in
many situations, these artifacts are not stereotypical and are con-
fused with real brain activity, such as epileptic activity. During the
experiment, an accelerometer recording the movement [12] or
videotaping helps in distinguishing the movement artifacts from
the signals. In our experience, laboratory mice, even the wild type,
produce epileptic events more often than humans. In most cases,
the epileptic events occur during a quiescent state and do not occur
in all channels like movement artifacts do, and this characteristic
 High-Density Electroencephalography in Freely Moving Mice 7

allows the investigators to distinguish the events. It is recom-

mended to separate the epileptic periods prior to further analysis.
The other commonly monitored artifact is the electrocardiog-
raphy (ECG), i.e., the electrical activity of the heart. ECG is often
observed in acute recording and has a stereotypic waveform. The
artifact with a well-defined temporal signature like ECG could be
easily identified with an independent component analysis (ICA),
which is a blind source separation method, decomposing a signal
into statistically independent components from each other
[13]. ICA could also be used to separate neural signals which are
independent from other signals in the time and space domain. An
open-source algorithm of ICA is available from EEGLAB, a
MATLAB-based toolbox released at the Swartz Center for Compu-
tational Neuroscience [14] (downloadable at https://sccn.ucsd.
edu/eeglab/).
In the case of mouse EEG, muscle activities are not detected,
probably due to cementing. Instead, a concern arose in comparing
rhythmical activities across a long period of time because the mouse
brain is more vulnerable to variation of pink noise. Pink noise
appears as a low-frequency component with an inverse function of
frequency in the frequency domain. It is also known to be ubiqui-
tous. The strength of pink noise fluctuates over time to a significant
degree in the mobile condition. Hence, investigators should be
cautious when they compare spectral power across time. It is
recommended to eliminate the pink noise or study the power in a
ratiometric manner.

3.4.2 EEG Topography One of the advantages of high-density EEG in mice is the ability to
investigate the spatiotemporal change of cortical activation with
topographical analysis. EEG topography is a neuroimaging tech-
nique for mapping the spatial distributions of cortical activity.
Basically, EEG topography is a contour plot on the top surface of
the mouse brain. First, we create a mesh (set of imaginary points) to
represent a cortical surface. Typically, the stereotaxic resolution is
used for the mesh size so that any EEG channel can be correspon-
dent to a mesh. Second, we limit the surface to an ellipsoidal
boundary. Third, the potential value of the mesh is calculated
from an interpolation method. According to our tests, different
types of interpolation do not affect the results to a noticeable
degree. We normally use cubic spline interpolation. Figure 2
shows a cine-mode presentation of EEG topographies, while the
ventral posterior medial nucleus of the thalamus (VPM) of a Thy1-
ChR2-EYFP mouse was optogenetically stimulated. The outer sur-
face of the mouse neocortex was rendered by a spm_surf function in
the SPM8 open-source toolbox (Wellcome Trust Centre for Neu-
roimaging UCL, London, UK) based on the mouse MRI down-
loaded from the open database of the magnetic resonance
microimaging neurological atlas group [15].
8 Jee Hyun Choi and Eunjin Hwang

Fig. 2 Topographical mapping of cortical signal propagation in response to optogenetic stimulation of VPM. (a)
Schematics of the signal propagation pathway (upper) and site of stimulation (lower). (b) Cine-mode
presentation of EEG topography

4 Note on Applications

4.1 Simultaneous EEG is a widely used method to monitor human brain activities due
Recordings of High- to its noninvasiveness and high temporal resolution. However,
Density EEG and since human EEG measures the voltage changes from the scalp,
Subcortical Brain an investigation of subcortical origins or their interplay with corti-
Activities cal activity is highly limited in human EEG. In this section, we
introduce a method to combine high-density EEG with a conven-
tional local field potential (LFP) to simultaneously monitor brain
activities in the subcortical region and cortex in mice. By combin-
ing EEG and LFP, we are able to investigate the functional connec-
tivity between the subcortical region and cortex. Here, we show an
example of combined EEG and LFP in a study of spike-and-wave
discharge (SWD) in the thalamus and cortex, simultaneously. The
SWDs are known to be the EEG hallmark of absence seizure [16]
and are often associated with loss of consciousness [17]. We moni-
tored the thalamic activation with cortical activation and the func-
tional connectivity between the two regions when the mouse
experienced absence seizure. The origin of such connectivity is
still under investigation, but an emergent phenomenon of syn-
chrony between the thalamus and cortex has been observed previ-
ously. The synchrony analysis of LFP and EEG quantitatively
assesses the functional connectivity of the thalamus and cortex,
and a calculation of time lag of thalamic and cortical SWDs leads
us to infer the causation of the signal.
 High-Density Electroencephalography in Freely Moving Mice 9

4.1.1 Methodological Most of the procedures of implementing LFP together with EEG
Considerations of are the same as implementation of EEG only. The only difference is
Simultaneous Recording of that making a hole for insertion of the LFP electrode and the
High-Density EEG and LFP fixation of the LFP electrode takes place after the EEG microarray
has been attached on the skull. In the case where the subcortical
region of interest is along the midline, where the interconnection
lines of EEG channels are located, the hole can be moved laterally
by x in the same coronal plane, and the depth of electrode insertion,
d, is calculated by the Pythagorean theorem (x2 + z2 ¼ d2), in which
z is the dorsoventral coordinate. The angle of the electrode inser-
tion equals the tangent of z/x. The position of the LFP electrode
should be confirmed by brain histology after all experiments are
conducted. Typically, it is recommended to match the impedances
of the reference and active electrode at similar levels [18], but
sharing the same reference and ground electrodes for the LFP
electrode delivers similar signal quality. Moreover, it is important
to reduce the number of electrodes for the well-being of the
animals in long-term chronic recordings.

4.1.2 Comparison of EEG One of the most interesting questions to explore about seizures is
and LFP Signals During the seizure initiation factors. Here, we exemplify the application of
SWD high-density EEG to pharmacologically induced absence seizure.
To produce SWDs in the mouse, we injected gamma-butyrolactone
(GBL, 100 mg/kg, i.p.) in awake mice and monitored them during
the freely moving state. A previous study of simultaneous record-
ings of intracellular potentials and EEG showed that the spike of
SWD in EEG corresponds to synchronous burst firings activated by
low-voltage activated channels [19], which consequently recruit
the neurons to activate synchronously. By comparing the time
lags between spike peaks, the initiation or propagation of seizure
can be studied. Figure 3a shows an exemplary EEG and LFP during
SWD. In each channel, the first positive spikes were identified and
marked with inverted triangles. The polarity of the depth electrode
is the opposite to that of the surface electrode; thus, it was inverted
in the plot. In the particular event, the first spike was monitored in
the frontal cortex including the prefrontal and motor cortices. The
level of seizure was calculated by total power in the period of SWD
(Fig. 3c), and the functional connectivity between different cortical
regions (Fig. 3d) was represented as thickness of the lines. In this
case, the functional connectivity was calculated by cross-correlation
coefficient, which assesses the similarity of two signals. Other vari-
ables, such as the phase synchronization index, could also be used
to gauge the synchronization level of the two regions [20]. Most of
the connections were between the two sites located symmetrically
with respect to the midline. Likewise, the functional connectivity
between the thalamus and cortex (Fig. 3e) was calculated by the
cross-correlation coefficient and mapped as a color map. The tem-
poral relationship between spikes was assessed with peak latency
and mapped for the first and second peaks, respectively (Fig. 3f, g).
10 Jee Hyun Choi and Eunjin Hwang

Fig. 3 Simultaneous recording of EEGs and LFPs during GBL-induced SWD. (a) Exemplary traces of SWD. (b)
Cortical high-density EEG montage and relative location of LFP electrode in VB (green dot). (c) Power map of
SWD. Total power from 1 Hz to 60 Hz was used for topographical representation. (d) Cortico-cortical functional
connectivity during SWD. Functional connectivity was measured by computing the cross-correlation between
channels. Connections stronger than 0.95 were shown as links between channels. (e) Functional connectivity
between the cortex and thalamus. (f) Latency map of first spikes with respect to the first spike of the FP1
channel. (g) Latency map of second spikes with respect to the second spike of the FP1 channel

4.2 Use of High- Despite much investigations by pharmaceutical companies, neuro-

Density EEG in psychiatry or neurological diseases notoriously have a low clinical
Preclinical Studies trial success rate (6.2% without biomarker and 8.3% with bio-
marker, data from 2005 to 2015 [21]). One compelling explana-
tion is a failure in selecting the right animal models and relevant
biomarkers for the disease. Among major diseases, there are chal-
lenges and limitations in the validated and predictive animal models
for neuropsychiatry or neurological diseases because most cognitive
dysfunctions in human patients are not directly assessable in animal
models. Therefore, it is critical to develop cross-species assays for
comparing equivalent test scores in human patients and animal
 High-Density Electroencephalography in Freely Moving Mice 11

models. In this respect, high-density EEG could be pivotal to

bridge the translational gap to the clinic by presenting the compar-
ative neuromarkers of cognitive dysfunctions.
One suggestion is the auditory steady-state response (ASSR)
test battery. The ASSR is an electrophysiological response entrained
to periodic sound pulses and, in particular, the ASSR appears as a
resonant response to stimuli in the gamma frequency band
(30–50 Hz) in humans [22] and rodents [23]. Gamma-band
ASSR has been suggested as a neuromarker for the detection of
psychosis. For instance, abnormal gamma-band ASSR has been
robustly observed in patients with first-episode or chronic schizo-
phrenia [24–26]. The ASSR test could be effective and useful in
screening the schizophrenia mouse model. Although it is known
that more than 100 genes are associated with schizophrenia [27],
schizophrenia mouse models are not validated by the genetic fac-
tors alone since the gene-environment interactions also play a
crucial role for the manifestation of schizophrenia symptoms
[28]. Therefore, most validations of schizophrenic mouse models
have been based on the behavioral tests such as the prepulse inhibi-
tion evaluating sensorimotor gating [29], locomotor activity
reflecting psychomotor agitation [30], social withdrawal [31], cog-
nitive deficits in working memory, learning, and attention [30],
which deliver neither direct evidence nor comparative results. Con-
trary to the behavioral assays above, the gamma-band ASSR is
observed in both humans and animals [32]. Additionally, it is a
cross-species comparative marker because the generator for
gamma-band oscillations has a cellular origin rather than a circuitry
origin [33, 34]. An example of an impaired ASSR and auditory
evoked potential was reported for a schizophrenia mouse model
[35]. In many cases, knockout mouse models might carry more
phenotypes than patients. We believe that a combination of fine
tuning of genetic modification and cross-species comparative high-
density EEG could be used to effectively identify the patient-like
mouse model.
In sum, we believe that more precise selection of a mouse
model of disease could be achieved with high-density EEG valida-
tion, which will enhance the translational value of the mouse model
and maximize the synergy of preclinical and clinical studies.

References

1. Cohen MX (2017) Where does EEG come
from and what does it mean? Trends Neurosci
40(4):208–218. https://doi.org/10.1016/j.
tins.2017.02.004
2. Bragin A, Jando G, Nadasdy Z, Hetke J,
Wise K, Buzsaki G (1995) Gamma (40-100
Hz) oscillation in the hippocampus of the
behaving rat. J Neurosci 15(1 Pt 1):47–60
3. Buzsaki G, Wang XJ (2012) Mechanisms of
gamma oscillations. Annu Rev Neurosci 35:
203–225. https://doi.org/10.1146/annurev-
neuro-062111-150444
4. O’Keefe J, Recce ML (1993) Phase relation-
ship between hippocampal place units and the
EEG theta rhythm. Hippocampus 3(3):
12 Jee Hyun Choi and Eunjin Hwang
317–330. https://doi.org/10.1002/hipo.
450030307
5. Tort AB, Kramer MA, Thorn C, Gibson DJ,
Kubota Y, Graybiel AM, Kopell NJ (2008)
Dynamic cross-frequency couplings of local
field potential oscillations in rat striatum and
hippocampus during performance of a T-maze
task. Proc Natl Acad Sci U S A 105(51):
20517–20522. https://doi.org/10.1073/
pnas.0810524105
6. Singer W (2013) Cortical dynamics revisited.
Trends Cogn Sci 17(12):616–626. https://
doi.org/10.1016/j.tics.2013.09.006
7. Choi JH, Koch KP, Poppendieck W, Lee M,
Shin HS (2010) High resolution electroen-
cephalography in freely moving mice. J Neuro-
physiol 104(3):1825–1834. https://doi.org/
10.1152/jn.00188.2010
8. Lee M, Kim D, Shin HS, Sung HG, Choi JH
(2011) High-density EEG recordings of the
freely moving mice using polyimide-based
microelectrode. J Vis Exp (47):2562. https://
doi.org/10.3791/2562
9. Buzsaki G, Stark E, Berenyi A, Khodagholy D,
Kipke DR, Yoon E, Wise KD (2015) Tools for
probing local circuits: high-density silicon
probes combined with optogenetics. Neuron
86(1):92–105. https://doi.org/10.1016/j.
neuron.2015.01.028
10. Richardson RR Jr, Miller JA, Reichert WM
(1993) Polyimides as biomaterials: preliminary
biocompatibility testing. Biomaterials 14(8):
627–635
11. Viventi J, Kim DH, Vigeland L, Frechette ES,
Blanco JA, Kim YS, Avrin AE, Tiruvadi VR,
Hwang SW, Vanleer AC, Wulsin DF, Davis K,
Gelber CE, Palmer L, Van der Spiegel J, Wu J,
Xiao J, Huang Y, Contreras D, Rogers JA, Litt
B (2011) Flexible, foldable, actively multi-
plexed, high-density electrode array for
mapping brain activity in vivo. Nat Neurosci
14(12):1599–1605. https://doi.org/10.
1038/nn.2973
12. Hwang E, Kim S, Han K, Choi JH (2012)
Characterization of phase transition in the tha-
lamocortical system during anesthesia-induced
loss of consciousness. PLoS One 7(12):
e50580. https://doi.org/10.1371/journal.
pone.0050580
13. Bell AJ, Sejnowski TJ (1995) An information-
maximization approach to blind separation and
blind deconvolution. Neural Comput 7(6):
1129–1159
14. Delorme A, Makeig S (2004) EEGLAB: an
open source toolbox for analysis of single-trial
EEG dynamics including independent compo-
nent analysis. J Neurosci Methods 134(1):
9–21. https://doi.org/10.1016/j.jneumeth.
2003.10.009
15. Ma Y, Smith D, Hof PR, Foerster B,
Hamilton S, Blackband SJ, Yu M, Benveniste
H (2008) In vivo 3D digital atlas database of
the adult C57BL/6J mouse brain by magnetic
resonance microscopy. Front Neuroanat 2:1.
https://doi.org/10.3389/neuro.05.001.
2008
16. Pinault D, O’Brien TJ (2005) Cellular and
network mechanisms of genetically-
determined absence seizures. Thalamus Relat
Syst 3(3):181–203. https://doi.org/10.
1017/S1472928807000209
17. Blumenfeld H (2005) Consciousness and epi-
lepsy: why are patients with absence seizures
absent? Prog Brain Res 150:271–286.
https://doi.org/10.1016/S0079-6123(05)
50020-7
18. Schomer DL, da Silva FL (2011) Nieder-
meyer’s electroencephalography: basic princi-
ples, clinical applications, and related fields,
6th edn. Lippincott Williams & Wilkins,
Philadelphia
19. Pinault D, Leresche N, Charpier S, Deniau JM,
Marescaux C, Vergnes M, Crunelli V (1998)
Intracellular recordings in thalamic neurones
during spontaneous spike and wave discharges
in rats with absence epilepsy. J Physiol 509
(Pt 2):449–456
20. Stam CJ (2005) Nonlinear dynamical analysis
of EEG and MEG: review of an emerging field.
Clin Neurophysiol 116(10):2266–2301.
https://doi.org/10.1016/j.clinph.2005.
06.011
21. Wong CH, Siah KW, Lo AW (2018) Estima-
tion of clinical trial success rates and related
parameters. Biostatistics. https://doi.org/10.
1093/biostatistics/kxx069
22. Galambos R, Makeig S, Talmachoff PJ (1981)
A 40-Hz auditory potential recorded from the
human scalp. Proc Natl Acad Sci U S A 78(4):
2643–2647
23. Conti G, Santarelli R, Grassi C, Ottaviani F,
Azzena GB (1999) Auditory steady-state
responses to click trains from the rat temporal
cortex. Clin Neurophysiol 110(1):62–70
24. Kwon JS, O’Donnell BF, Wallenstein GV,
Greene RW, Hirayasu Y, Nestor PG, Hasselmo
ME, Potts GF, Shenton ME, McCarley RW
(1999) Gamma frequency-range abnormalities
to auditory stimulation in schizophrenia. Arch
Gen Psychiatry 56(11):1001–1005
25. Spencer KM, Salisbury DF, Shenton ME,
McCarley RW (2008) Gamma-band auditory
steady-state responses are impaired in first epi-
sode psychosis. Biol Psychiatry 64(5):
 High-Density Electroencephalography in Freely Moving Mice
369–375. https://doi.org/10.1016/j.
biopsych.2008.02.021
26. Thune H, Recasens M, Uhlhaas PJ (2016) The
40-Hz auditory steady-state response in
patients with schizophrenia: a meta-analysis.
JAMA Psychiat 73(11):1145–1153. https://
doi.org/10.1001/jamapsychiatry.2016.2619
27. Schizophrenia Working Group of the Psychiat-
ric Genomics C (2014) Biological insights
from 108 schizophrenia-associated genetic
loci. Nature 511(7510):421–427. https://
doi.org/10.1038/nature13595
28. van Os J, Rutten BP, Poulton R (2008) Gene-
environment interactions in schizophrenia:
review of epidemiological findings and future
directions. Schizophr Bull 34(6):1066–1082.
https://doi.org/10.1093/schbul/sbn117
29. Geyer MA, McIlwain KL, Paylor R (2002)
Mouse genetic models for prepulse inhibition:
an early review. Mol Psychiatry 7(10):
1039–1053. https://doi.org/10.1038/sj.mp.
4001159
30. Powell CM, Miyakawa T (2006)
Schizophrenia-relevant behavioral testing in
rodent models: a uniquely human disorder?
Biol Psychiatry 59(12):1198–1207. https://
doi.org/10.1016/j.biopsych.2006.05.008
31. Wilson CA, Koenig JI (2014) Social interaction
and social withdrawal in rodents as readouts for
13
investigating the negative symptoms of schizo-
phrenia. Eur Neuropsychopharmacol 24(5):
759–773. https://doi.org/10.1016/j.
euroneuro.2013.11.008
32. Brenner CA, Krishnan GP, Vohs JL, Ahn WY,
Hetrick WP, Morzorati SL, O’Donnell BF
(2009) Steady state responses: electrophysio-
logical assessment of sensory function in
schizophrenia. Schizophr Bull 35(6):
1065–1077. https://doi.org/10.1093/
schbul/sbp091
33. Sohal VS, Zhang F, Yizhar O, Deisseroth K
(2009) Parvalbumin neurons and gamma
rhythms enhance cortical circuit performance.
Nature 459(7247):698–702. https://doi.org/
10.1038/nature07991
34. Cardin JA, Carlen M, Meletis K, Knoblich U,
Zhang F, Deisseroth K, Tsai LH, Moore CI
(2009) Driving fast-spiking cells induces
gamma rhythm and controls sensory responses.
Nature 459(7247):663–667. https://doi.org/
10.1038/nature08002
35. Shahriari Y, Krusienski D, Dadi YS, Seo M,
Shin HS, Choi JH (2016) Impaired auditory
evoked potentials and oscillations in frontal and
auditory cortex of a schizophrenia mouse
model. World J Biol Psychiatry 17(6):
4 3 9 – 4 4 8 . h t t p s : // d o i . o r g / 1 0 . 3 1 0 9 /
15622975.2015.1112036
 Chapter 2

Multisite Recording for the Analysis of Information Flow

Between Thalamocortical Regions
Hiroshi T. Ito

Abstract
High-density multichannel recordings are a powerful technique to investigate the information represented
in the activity of neuronal ensembles. Recent technological advancement has achieved the development of
small and low-weight recording devices, allowing for the use of hundreds of channels to monitor the neural
activity from freely behaving animals. Because of the substantial increase in the channel number of
recording devices, it is more common to record the neural activity from multiple brain regions simulta-
neously, allowing for researchers to investigate functional dependency and interactions5 between regions.
However, multiregional recordings often require different techniques from the ones used for single-region
studies. This chapter will review the methodological details of simultaneous recordings from multiple brain
regions as well as analysis techniques used for such datasets.

Key words Spatial navigation, Multiregional interactions, Ensemble analysis, Freely behaving record-
ings, Tetrodes

1 Introduction

Animals need to adapt their behaviors to dynamically changing

environments, and thus functional demands of the brain are differ-
ent from moment to moment. A large body of experimental evi-
dence suggests that the brain can cope with various behavioral
demands by changing functional interactions between regions.
Synchrony of oscillatory activity in the brain has been considered
a key mechanism to modulate multiregional interactions [1–3].
While many studies have demonstrated the difference in the degree
of synchrony depending on behavioral demands, direct demonstra-
tion of modulation of information flow by this mechanism is still
experimentally challenging. Here, large-scale recordings of spike
activity together with local field potentials from multiple brain
regions are a key technique to investigate this question.
Goal-directed spatial navigation is considered a behavior that
requires multiregional interactions in the brain. When animals

Robert P. Vertes and Timothy A. Allen (eds.), Electrophysiological Recording Techniques, Neuromethods, vol. 192,
https://doi.org/10.1007/978-1-0716-2631-3_2, © Springer Science+Business Media, LLC, part of Springer Nature 2022

15
16 Hiroshi T. Ito

intend to visit a particular location in space, they first need to

identify their current position in the environment. Then, based
on previous experiences or memories about the environment,
they are required to estimate the direction and route to the desti-
nation. This entire process requires multistep computations,
including the process of incoming sensory signals, retrieval of
memory about the environment, and decision and planning of the
destination and next actions. Each of these processes requires
cooperation of multiple brain regions, and in support of this idea,
previous studies have reported temporal coordination of the activ-
ity between distant brain regions during navigation [4–8]. For
example, when rats make decisions about the next route, spike
times of neurons in the prefrontal cortex, a brain area crucial for
action planning or decision making [9, 10], are phase-locked to the
theta rhythm in local field potentials in the hippocampus, a part of
the brain’s spatial representation system [11–13]. This temporal
coordination or synchrony is considered a mechanism of transfer-
ring information necessary for route decisions between the prefron-
tal cortex and the hippocampus [4, 14, 15]. Multiregional
recording further helped identify the exact information content
that is transferred between regions [16–18]. These studies revealed
that the midline thalamic nucleus reuniens (RE) [18–20] (Fig. 1a)
is an anatomical hub to transfer the animal’s next movement direc-
tion from the prefrontal cortex to the hippocampus during route
decisions. In the following sections, I will review the techniques
used in these studies as an example, but the same techniques should
also be applicable to other brain regions.

2 Construction of Microdrive for Multisite Recordings

2.1 Tetrode and Each channel of recording electrodes picks up voltage signals
Microdrive derived from hundreds of neurons nearby. Voltage signals gener-
ated by neurons are usually associated with either spikes or synaptic
potentials. A high-pass filter is used to isolate signals derived from
spikes, because spike waveforms are mostly confined in a high-
frequency spectral range compared to those of synaptic potentials.
A typical range of passbands used for spike detection is between
300 and 6000 Hz. To investigate the activity of individual neurons,
it also requires to classify and assign detected spikes to individual
neurons. While spike waveforms are often used as a criterion for
spike classification, the waveforms of the same neurons are not
necessarily always the same, for example, during spike bursting,
leading to a possibility of misclassification. Here, a high-density
channel configuration of electrodes provides another criterion for
spike classification [21, 22]. When the electrode has multiple chan-
nels in small space, signals from the same neuron can be detected
across multiple channels that are placed at different positions
Multisite Recording for the Analysis of Information Flow Between. . . 17

Fig. 1 Microdrive construction: (a) A scheme showing anatomical locations of the

nucleus reuniens (RE) and the hippocampus CA1. (b) Magnetic stirrer to make
tetrodes. The bottom white bar rotates to make twists of wires. (c) Left. A Harlan
28 microdrive with split bundles. Right. The bottom side view of tetrode bundles
showing 28 holes for tetrodes and the inset showing an extended tetrode from
the bundle. d. Tetrode bundle is constructed by binding 14 metal tubings
together by heat-shrink tubings and applying solder in the middle. The bundle
is then cut in the middle, resulting in two bundles, each of which can hold
14 tetrodes

relative to the neuron. Due to the decay of electrical signals along

the distance from the source, the ratio of spike amplitudes between
channels gives a distinct signature of the relative position of indi-
vidual neurons, improving the accuracy of spike classification.
Tetrodes are a type of electrode with four independent channels
made by twisted wires. A tetrode is constructed manually from a
single 17 μm wire made of 90% platinum and 10% iridium (Cali-
fornia Fine Wire). First, a wire loop is formed by connecting both
ends with heat-resistant copper tape. The wire loop is then twisted
18 Hiroshi T. Ito

and hanged over a magnetic stirrer using a magnetic bar to hold the
bottom twist of the wire (Fig. 1b). After rotating the magnetic bar
for a few minutes, multiple twists of four wires will be formed. This
tetrode is then heated up to 210
C with a heat gun for 2 min for
sealing.
A microdrive of tetrodes can be obtained commercially, for
example, from Neuralynx or Axona, but it is also possible to con-
struct with your own design using a 3D printer. A Harlan 28 drive
from Neuralynx, for example, can hold 28 independently movable
tetrodes (Fig. 1c), which is a good size for the recordings from the
hippocampus and the nucleus reuniens simultaneously in behaving
rats. The microdrive will be fixed on the animal’s skull using several
anchor screws (e.g., M1.6, 3 mm) during the surgery. The ground
screws of the drive are typically placed on the skull above the
cerebellum for recordings from the hippocampus, the thalamus,
or other neocortical regions, as the cerebellar signals are presum-
ably largely uncorrelated with ones in these recording areas. The
use of reference tetrode is useful to reduce noise from muscle
activity (e.g., chewing noise), which helps obtain better spike wave-
forms for classification.

2.2 Design of Split For simultaneous multisite recordings, it is important to configure

Bundle the insertion angles and positions of individual tetrodes desirably so
that they can precisely target different brain structures. For this
purpose, we construct cannula bundles using 30 gauge hypodermic
tubings to hold individual tetrodes at precise positions. First, a
desired number of tubings are held together by using a pair of
heat-shrink tubings (Fig. 1d). Then a thin layer of solder is applied
in the middle of the tubing to fix them firmly. Finally, the soldered
region in the middle should be cut into two pieces, resulting in two
tetrode bundles. The separation distance and angle of each bundle
should be carefully adjusted for targeted brain regions and fixed
with dental cement (Fig. 1c).
Usually, each hypodermic tubing of the bundle is dedicated for
one tetrode. However, for targeting deep brain structures (>5 mm
depth), the strength of tetrodes is sometimes not sufficient to
penetrate the brain tissue. We thus sometimes make a bundle of
two tetrodes that are glued together and come out from one
hypodermic tubing. This strategy works well for recordings from
the thalamic nucleus reuniens that is located approximately 7 mm
below from the cortical surface [16].

2.3 Recording of the Several companies offer a complete set of recording systems with
Neural Activity software integrated with video tracking of the animal’s positions
Together with the and head directions, which is necessary for the investigation of
Animal’s Behaviors neural correlates to behaviors. If one wants to use an open-source
system (e.g., Open Ephys), it is necessary to combine the recording
system with a position tracking software to monitor the animal’s
 Multisite Recording for the Analysis of Information Flow Between. . . 19

behaviors. In such a case, it is important to make sure to use the

same time clock for position tracking and recording systems. Usu-
ally, all timestamps should be aligned to the clock signals of the
recording system. For example, in our system, whenever a new
image frame from CCD camera is acquired by the image grabber
(PC2-Comp Express, DALSA), a TTL pulse is sent to the recording
system so that the times of individual frame acquisitions are
registered as the timestamps of the recording data. Positions and
head directions of the animal can be extracted by monitoring two
LEDs with different colors (e.g., red and green) mounted on the
headstage.

3 Analysis of Multisite Recording Data

Simultaneous recordings from multiple brain regions expand the

possibility for analyses, because one can use not only typical meth-
ods used for single brain regions but also additional methods to
investigate interactions between the regions. This section describes
several useful techniques, using the analysis of thalamo-
hippocampal circuit as an example.

3.1 Spectral Spectral coherence analysis is a widely used technique to assess the
Coherence Analysis degree of synchrony of oscillatory activity between regions. This
technique is mathematically related to spectral power analysis; while
the spectral power analysis is based on an autocovariance function
of local field potentials (LFPs) from a single brain region, the
spectral coherence analysis is on a cross-covariance function of
LFPs between regions.
Spectral power density:
X
τ¼1
S xx ðωÞ ¼ γ xx ðτÞe 2πiωτ
τ¼1

γ xx ðτÞ ¼ E ½ðx t  μx Þðx tþτ  μx Þ : ðAutocovarianceÞ

x t : signal at time t, μ : signal average
Cross-spectrum:
X
τ¼1
S xy ðωÞ ¼ γ xy ðτÞe 2πiωτ
τ¼1
h  i
γ xy ðτÞ ¼ E ðx t  μx Þ y tþτ  μy : ðCross‐covarianceÞ

y : signal from the other brain region

20 Hiroshi T. Ito

Spectral coherence:
 
S xy ðωÞ2
C xy ðωÞ ¼
S xx ðωÞS yy ðωÞ
Therefore, spectral coherence is considered “normalized”
cross-spectrum. The calculation of spectral coherence does not
require spike data, and thus it can be performed if at least one
channel is located in individual brain regions.
While spectral coherence analysis provides the degree of coher-
ence across different frequencies, it does not give time-dependent
changes. When one wants to investigate behavior-dependent
changes of coherence, for example, during a goal-directed naviga-
tion task, a time-resolved strategy should be applied by using a
sliding time window. There are several choices of windows (e.g.,
Hanning or Hamming), which reduces the edge effects due to the
handling of short-duration datasets. It is also known that the multi-
taper method helps reduce estimation bias [23–25], and a useful
MATLAB toolbox, Chronux, is available for its
implementation [25].
In Fig. 2a, the spectral coherence analysis was applied when the
animal performed a continuous alternation task, in which the ani-
mal was required to choose either right or left turn at the
T-junction of the maze alternately to obtain a reward. The analysis
reveals a peak of coherence at the theta range of frequency
(6–12 Hz) between LFPs in CA1 and the nucleus reuniens (RE).
The RE has mutual anatomical connections with the prefrontal
cortex and also gives rise to inputs in the area CA1 and the sub-
iculum regions of the hippocampus [19, 26–28]. The RE has been
considered a key anatomical hub for the communication between
the prefrontal cortex and the hippocampus because virtually no
direct projection exists from the prefrontal cortex to the hippocam-
pus [27]. Therefore, synchrony at the theta rhythm implies func-
tional coupling between RE and CA1, mediating prefrontal-
hippocampal interactions. To investigate if coherence is modulated
during the task phase, time-resolved spectral coherence analysis was
implemented between RE and CA1 using a sliding time window of
512 ms. The analysis shows a prominent coherence in the theta
frequency range that changes from time to time during the behav-
ioral task (Fig. 2b). The degrees of theta coherence are further
mapped on spatial positions of the maze using a color code,
which revealed a clear spatial bias of theta coherence (Fig. 2c).
The theta coherence between CA1 and RE is higher at the stem
part of the maze in which the animals are required to make a
decision of the next trajectory. The enhanced synchrony between
CA1 and RE implies a change of information flow, suggesting a key
role of their functional coupling during trajectory decisions.
 Multisite Recording for the Analysis of Information Flow Between. . . 21

Fig. 2 Spectral coherence analysis between CA1 and RE LFPs. (a) Top: LFPs from the hippocampus CA1 and
the nucleus reuniens (RE). Bottom: the left two plots showing spectral power of LFPs from CA1 and RE. The
right plot shows spectral coherence. Note a peak of coherence at the theta range of frequency (6–12 Hz). (b)
Time-resolved spectral coherence of LFPs from CA1 and RE. The degree of coherence is color-coded. A
prominent coherence peak at the theta rhythm is observed, but it varies over time. (c) The degree of spectral
coherence is plotted on spatial positions of the maze. Left. A rat performed a continuous alternation task in an
8-shaped maze. At the stem part of the maze, the animal was required to choose either right or left trajectory
based on the alternation rule (i.e., choosing the different direction from the previous trial). The right top panel
shows the spectral coherence between CA1 and RE at the theta rhythm on the maze. The right bottom panel
shows the phase coherence estimated by ISPC. In both plots, a prominent increase of theta coherence is
observed at the stem part of the maze

Caution for the interpretation of spectral coherence analysis is

that the degree of coherence reflects not only phase synchrony but
also the amplitude coordination of oscillatory activity. When one
wants to restrict the focus on phase synchrony, an alternative met-
ric, intersite phase clustering (ISPC) [29], can be used:

1 X iðθx ðt Þθy ðt ÞÞ
N
ISPC ¼j e j
N t¼1
22 Hiroshi T. Ito

where θx and θy are instantaneous phases at a specific frequency

band, which can be estimated by a Hilbert transform described in
the following section. The data is sampled at t ¼ 1, 2,. . ., N, and the
phase consistency over time is calculated. A time-resolved version of
this method was used to assess the phase coordination between the
theta rhythms in CA1 and RE by using a sliding time window of
512 ms, which confirmed the enhanced ISPC at the stem part of
the maze (Fig. 2c). The results together suggest the enhanced
phase synchrony of theta rhythm between CA1 and RE at the
stem part of the maze.

3.2 Spike-Phase While coherence analysis based on local field potentials (LFPs) is a
Analysis Across Brain useful method to assess the degree of synchrony between brain
Regions regions, it is important to note that LFPs are not necessarily local
signals. A study found that LFPs can spread more than a centimeter
range in the brain of nonhuman primates [30]. It is thus difficult to
exclude a possibility that the measured spectral coherence is influ-
enced by other brain regions near the recording electrodes. This
problem can be solved by using the correlation analysis based on
spike activity, because spikes can be detected only within a small
distance range from the electrode (~50 μm) [31]. Therefore, spike-
based analysis is an important addition to confirm synchrony
between regions.

3.3 Extraction of One strategy to investigate temporal coordination based on spike

Instantaneous Phase activity is to measure the correlation of spike times of neurons in
by the Hilbert one brain region to the phase of oscillatory activity in another brain
Transform region. This analysis requires the estimates of instantaneous phases
of the oscillatory activity in LFPs, and they can typically be esti-
mated by using a Hilbert transform. The Hilbert transform is a
method for signal processing to obtain an analytic form of signals.
To implement this approach, the LFP signal should first be filtered
by a band-pass filter for the given frequency range. For example, if
one wants to estimate instantaneous phases of theta oscillations, the
passband of the filter should typically be in the range of 6–12 Hz. A
commonly used filter type for this purpose is either a Butterworth
infinite-impulse-response (IIR) filter or a finite-impulse-response
(FIR) equiripple filter. The Butterworth filter is known to have a
flat magnitude in the passband, but have a slow roll-off toward the
stopband. In contrast, the equiripple filter has small magnitude
fluctuations (ripples) in the passband, but it is a linear phase filter
which guarantees the maintenance of waveform shapes, and fur-
thermore, its design algorithm, Parks-McClellan method, allows
for the flexible specification of the ripple size and roll-off speed of
the stopbands. After filtering of the signal, the application of a
Hilbert transform will help separate amplitude and phase compo-
nents of the oscillatory activity as in the following procedure.
 Multisite Recording for the Analysis of Information Flow Between. . . 23

The LFP signal after the band-pass filtering can roughly be

expressed with an amplitude-modulated trigonometric function:
uðt Þ ffi A ðt Þ cos ðωt þ θÞ
Here, our aim is to identify the functions of instantaneous
amplitudes A(t) and phases wt + θ. The application of the Hilbert
transform will give the following signal, exchanging cosign with
sign function:
Hilbertðuðt ÞÞ ¼ A ðt Þ sin ðωt þ θÞ
This result is then used to construct the analytic form of the
signal:
ua ðt Þ ¼ uðt Þ þ i  Hilbertðuðt ÞÞ
¼ A ðt Þ½ cos ðωt þ θÞ þ i  sin ðωt þ θÞ
¼ A ðt Þ e iðωtþθÞ
This analytic form enables us to extract its amplitude A(t) or
phase wt + θ, by simply calculating the absolute value |ua(t)| or
angular component arg[ua(t)] of the signal, respectively.
The extracted phases by this method are then used to measure
the instantaneous phases at spike times, revealing the relationships
between spike times and oscillatory phases. For a quantitative anal-
ysis of spike phases, a circular histogram of spike phases can be
constructed to assess the degree of spike-phase modulation. The
phase bias of circular distribution is summarized by a mean vector
length that ranges from 0 to 1 (low to high phase bias). A useful
toolbox for circular statistics is available in MATLAB [32].

3.4 Spike-Field As described in the previous section, spike-phase analysis based on a

Coherence Analysis Hilbert transform requires the band-pass filtering of the data at a
Revealed Phase- specific frequency range. A possible ambiguity here is the choice of
Locking of RE Spikes frequency range of the filter. For example, the passbands for gamma
to CA1 Theta oscillations used in previous studies are 30–80 Hz, 40–100 Hz, or
25–140 Hz, and it is not clear which range should be used for a
particular dataset. Furthermore, when one wants to have an over-
view of spike-phase relationships across a wide range of frequency,
the repetitive application of spike-phase analysis using different sets
of band-pass filters may result in a risk of multiple comparisons of
statistical tests.
Here, spike-field coherence is another spike-phase estimation
method that does not require filtering of the datasets. This method
is based on the analysis of a spike-triggered local field potential
(LFP) average, and first, short LFP segments around spike times
(e.g., 300 ms) are aligned so that spike times of a particular neuron
are at the center of the segments (Fig. 3a). These LFP segments
around spike times are then averaged, giving a spike-triggered LFP
average. The key idea of this analysis is that while this averaging
24 Hiroshi T. Ito

Fig. 3 Spike-field coherence analysis between RE spikes and CA1 LFP. (a) Procedure of spike-field coherence
method. For each spike of a given RE cell, segments are extracted from CA1 LFP, so that the spike times are
aligned at the center of the segments. After averaging, a clear theta rhythm is observed in the real data, which
results in a prominent peak at the theta rhythm in the spectral power (left). By contrast, with simulated random
spikes, the LFP average as well as its spectral power are almost flat. The spectral power of the averaged LFP
 Multisite Recording for the Analysis of Information Flow Between. . . 25

procedure will reduce spike-uncorrelated signals, the components

of LFP that are temporally consistent relative to the spike times will
become prominent. While the spike-triggered LFP average itself
gives an idea of spike-phase relationship, the shape of averaged
waveform also depends on the baseline spectral power of LFP
segments. This dependency can be canceled out by taking a nor-
malization procedure, calculating the ratio of spectral power
between the spike-triggered LFP average and the baseline LFP
segments across frequencies. The final result is called spike-field
coherence, a metric of spike-phase correlation in a wide range of
frequency [33]. Spike-field coherence takes values from 0 to
1 (or 0–100%).
As described in the procedure, the spike-field coherence analy-
sis requires calculations of spectral power of short LFP segments.
One can use a conventional spectral power method based on Four-
ier transform. However, as this Fourier-based method is designed
for stationary signals, it is not necessarily ideal for detecting tran-
sient oscillatory events, such as bursts of gamma oscillations. It has
been suggested that such transient events can better be detected by
using a wavelet-based method. The wavelet-based method can be
used for spectral power estimation, by measuring the amplitudes of
the LFP signals that were convolved with complex Morlet wavelets
ranging from 4 to 140 Hz at 2 Hz intervals [7, 18].

3.5 Comparison of The analysis of spike-phase modulation is a useful technique to

Spike-Phase Locking understand the process of the brain’s signal between regions, and
Between Stem and it is often necessary to compare these values between groups, for
Nonstem Regions of example, between different animals (e.g., normal vs. knockout),
the Maze experimental conditions (e.g., with or without optogenetic excita-
tion/inhibition), or behavioral states (e.g., task phases). Here, it is
important to note that spike-phase estimation methods described
in the previous sections are dependent not only on the degree of
spike-phase coupling but also on the number of spikes used to
calculate them [34–36]. This is because the estimation of these
metrics is biased; even if spikes are not correlated to LFPs at all,
the calculated coherence value will never become zero with a finite
number of spikes. As the table of Fig. 3c shows, the degree of spike-
field coherence is largely affected by the number of spikes used for
ä

Fig. 3 (continued) segment is then normalized, giving a spike-field coherence. A prominent peak of coherence
is observed at the theta range, indicating that spikes of this particular RE cell are phase-locked to the theta
rhythm in CA1 LFP, but not in other frequency ranges (e.g., beta, gamma). (b) Comparison of spike-field
coherence between the stem and the nonstem parts of the maze. The coherence in the theta range is largely
increased at the stem part of the maze, indicating enhanced spike-time modulation by the CA1 theta. (c)
Influence of the spike number used to compute the spike-field coherence. A small number of spikes generally
give a larger coherence value, and therefore, it is important to use the same number of spikes for the group
comparison
26 Hiroshi T. Ito

calculation. Therefore, for the comparison of spike-phase modula-

tion between groups, one should be careful to equalize the number
of spikes for each group. A bootstrap resampling procedure can be
used for this purpose. For example, we set the sampling number of
spikes to be 4/5 of the minimum number of spikes across the
groups. Then, a specified number of spikes are randomly chosen
from each group to estimate spike-field coherence. This procedure
is repeated 100 times, and their averages are considered as a repre-
sentative value of the group (e.g., [18]).
By applying this procedure, we compared the spike-field coher-
ence between RE spikes and CA1 LFP between two task states,
either when animals are on the stem or nonstem parts of the maze
in the alternation task. On the stem part, animals are required to
make a trajectory decision at the upcoming T-junction, but on the
nonstem part, they can simply run along the track. First, we
observed a prominent peak at the theta range of frequency in the
spike-field coherence plot, indicating that RE spikes are phase-
locked to the CA1 theta. This coherence was further enhanced
when the animals were on the stem part of the maze, suggesting
that spike times of RE neurons are temporally coordinated at the
phases of CA1 theta (Fig. 3b). Together with the spectral coher-
ence analysis of LFPs, the results together confirmed the idea that
CA1 and RE are functionally coupled at the theta rhythm, particu-
larly during trajectory decisions.

3.6 Signal Although spike-field coherence analysis reveals temporal coordina-

Directionality Analysis tion between regions, the analysis does not provide directionality of
the impact. Several analyses are available to estimate the direction-
ality of signal flow, such as Granger causality or transfer entropy
[37–40]. One can also examine the information flow at a specific
frequency range, for example, by examining the microstructure of
temporal coupling between the CA1 theta and the spike activity in
another brain region [14, 18, 41]. This strategy is based on the
assumption of the signal time delay between regions due to the
finite traveling speed of spike activity along the axons.
To implement the signal directionality analysis between RE and
CA1, the mean vector lengths of spike phases of RE neurons are
calculated relative to a series of CA1 LFP traces that were band-pass
filtered at 6–12 Hz and temporally shifted from 200 ms to 200 ms
at 10 ms intervals. We can then identify the temporal shift that gives
the maximum mean vector length [14]. If the maximum is achieved
by shifting CA1 LFP backward, there must be a delay of the impact
of RE spikes on CA1 LFPs, indicating the directionality from RE to
CA1. By contrast, if the maximum can be achieved by shifting CA1
LFP forward, the results indicate the directionality from CA1 to RE
(Fig. 4a). The averaged mean vector length across all RE cells is
plotted in Fig. 4b. The plot indicates that the peak of the mean
vector length is achieved when CA1 LFP is shifted ~50 ms
 Multisite Recording for the Analysis of Information Flow Between. . . 27

Fig. 4 Signal directionality analysis between RE spikes and CA1 LFP. (a) Illustrations showing an idea behind
the analysis. If CA1 influences RE spikes, the correlation will become stronger by temporally shifting CA1 LFP
forward, because it will cancel out the axonal conduction delay. The opposite is the case if RE influences CA1.
(b) The plot shows the mean vector lengths for all recorded RE cells relative to CA1 theta at different amounts
of time shifts (thick line, means; shaded, standard errors). The peak was observed by shifting CA1 LFP
backward, indicating the signal directionality from RE to CA1

backward relative to RE spikes, indicating that RE spikes give the

maximum impact on CA1 theta after a small delay of axonal
conduction.

4 Decoding Analysis to Investigate Populational Representations of Trajectories

Place cells in the hippocampus change their firing rates depending

on the animal’s choice of trajectory on the T-junction of the maze
[42, 43]. While this rate modulation is thought to reflect either
previous or next behavioral states of the animal, place cells also
change their firing rates depending on the animal’s running
speed, head direction, or position [11, 44, 45], and therefore, the
analysis should account for the contributions of each of these
compounds separately. Here, ANOVA and ANCOVA can be used
to assess the significance of trajectory dependency by considering
the contribution of each of these behavioral variables (e.g., [43]).
As ANOVA or ANCOVA requires the segmentation of the data, the
stem position is first divided into six equally sized bins (15 cm), and
the following behavioral parameters are measured for each bin and
each trial: (1) firing rate, the number of spikes divided by the
amount of time spent in the bin; (2) running speed, the averaged
position shift per time in the bin; (3) head direction, the averaged
angle of two colored LEDs on the headstage; and (4) lateral posi-
tion, the averaged position perpendicular to the long axis of the
central stem. Then a two-way ANOVA is performed with trial type
(either left- or right-turn trials) and six spatial bins as independent
variables and firing rate as the dependent measure. The cells with a
significant main effect of the trial type are identified as potential
28 Hiroshi T. Ito

trajectory-dependent cells. For these cells, additional analysis is

performed to examine whether variations in speed, heading, or
lateral position might account for the differences in firing rate
between trial types. This can be examined with a two-way
ANCOVA with trial type and bins as the independent variables,
firing rate as the dependent measure, and speed, head direction,
and lateral position as covariates. Cells that continued to show a
significant difference in firing rate between left- and right-turn trials
(i.e., the change of firing rate cannot be explained by the animal’s
running speed, head direction, or lateral position) are classified as
trajectory dependent. This analysis revealed that 55.1% of CA1 cells
and 42.2% of RE cells are classified as trajectory dependent [16].

4.1 Population While the analysis based on ANOVA and ANCOVA can classify
Vector-Based each cell as either trajectory dependent or independent, this analysis
Decoding does not necessarily give the idea of information represented in a
brain area as a neural ensemble. A key advantage of decoding
analysis is the ability to evaluate the predictability of a particular
behavioral parameter from ensemble neural activity, and therefore,
it serves as a powerful strategy to assess its behavioral relevance.
Decoding analysis is usually performed in two steps, training
and testing phases. The dataset should be divided into two parts,
training and test datasets, where the training dataset is used to
optimize the decoding parameters and the test dataset is used
only for the evaluation of the decoder’s performance (Fig. 5b).
This is because the decoding performance should always be evalu-
ated with samples that are not used for parameter optimization, and
especially in a machine-learning algorithm, one needs to be cau-
tious about the decoder’s overfitting to the training dataset, which
will result in poor performance in new datasets even if it has almost
perfect performance on the training dataset.
A relatively simple form of decoding can be performed by
constructing vectors of firing rates of a neuronal ensemble (i.e.,
population vector). First, firing rates of each neuron were normal-
ized to z scores, so that a neuronal bias due to the difference of
baseline firing rates should be reduced. Then the averaged rate
vectors on the stem were calculated separately for either right- or
left-trajectory trials from the training dataset that is constructed by
removing one trial as a test dataset (Fig. 5a). Finally, the difference
between dot products for a test rate vector and the individual
trajectory vectors gives the decoded trajectory as in the following
equation:
 
T ¼ sign v  c right  v  c left
in which cright and cleft are the population vectors from the training
dataset for right- or left-trajectory trials and v is a vector of firing
rates of the same population of neurons on the test dataset. T is the
output value of the classifier; 1 or 1 represents right or left
Fig. 5 Decoding of the animal’s next trajectory from a population of neurons. (a) Illustration showing a
construction of a population vector of firing rates of neurons on the stem part of the maze. Each neuron shows
different levels of modulation in firing rates between right- and left-oriented trajectories, which is summarized
as two population rate vectors for each of the two trajectories. The decoder works based on the similarity of a
given vector to these two population vectors. (b) Cross-validation to test the decoder’s performance. The
dataset is divided into training and test datasets, and only the training dataset is used to construct the average
PVs for the two trajectories. The decoder’s performance is then tested with a test dataset by taking dot
products of a given vector with the PVs for the two trajectories. (c) Linear classifier in the machine-learning-
based approach. The firing rates of individual neurons are weighted and summated to give an output. The
weight adjustment allows for increasing or decreasing the contribution of particular cells for the best
classification performance. (d) Performance of the decoders predicting the animal’s next trajectory based
on neurons in RE and CA1. The chance level is 50%. The performance was in general better for the SVM
decoder compared with the PV decoder. The high decoding performance of CA1 cells was significantly
impaired by the lesions of RE, indicating the importance of RE for trajectory representation in CA1
30 Hiroshi T. Ito

trajectory, respectively. This procedure was repeated for all trials to

be tested (i.e., leave-one-out cross-validation), and the classification
accuracy on the test datasets was considered an estimate of the
decoding performance.

4.2 Support Vector While a population vector decoder can achieve a good performance
Machine for Decoding when an ensemble of neurons represents one common feature such
of the Next Trajectory as head direction or spatial position, neurons often exhibit mixed
Choice representations of multiple modalities [46–50], and it is thus pref-
erable for a decoder to have a mechanism to adjust the contribution
of individual neurons. This can be achieved by a machine-learning-
based decoding strategy, such as a support vector machine. Unlike a
population vector method in which the decoding efficacy can be
deteriorated by the inclusion of neurons that does not contribute to
the decoding feature, a machine-learning algorithm is able to dis-
regard such irrelevant information.
A support vector machine is one of the supervised learning
algorithms, in which new data are classified based on the class labels
of the training datasets. The performance of decoder should always
be tested with a dataset different from the one used in training (i.e.,
cross-validation), because even a decoder with perfect classification
performance in the training dataset does not necessarily give good
performance in another dataset, due to overfitting of the decoder
that impairs the generalization ability. The support vector machine
is considered an algorithm with excellent generalization ability [51]
and is widely used for machine-learning-based classification
problems.
While the support vector machine can be implemented with
nonlinear function, for simplicity, we use a linear decoder,
expressed as in the following equation, to predict the next trajec-
tory based on the spike rates on the stem (Fig. 5c):
T ¼ b þ w1  F 1 þ w2  F 2 þ w3  F 3 . . . : ¼ b þ w T F

F ¼ ½ F 1 , F 2 , F 3 , . . .T , w ¼ ½w 1 , w 2 , w 3 , . . .T
where F is a vector of firing rates for each cell, w is a vector of
respective weights, b is a scalar offset, and T is an output value of the
classifier, either 1 for right turn or 1 for left turn. The optimal
weights of the decoder were determined by a support vector
machine algorithm to maximize the separation margin for better
generalization performance. Briefly, for a given number N of trials
of rate-trajectory pairs, (Fi, Ti), i ¼ 1,2,3,. . .N, we searched for
w that satisfies the following condition:

1 T X N
min w wþC ξi
w, b, ξ 2
i¼1
 
Subject to y i b þ w T F i  1  ξi , ξi ,  0:
where C is a penalty parameter for misclassification, which we
usually set at 1, but changing the C value may improve the
 Multisite Recording for the Analysis of Information Flow Between. . . 31

performance depending on the dataset. A useful toolbox to imple-

ment support vector machine classification with MATLAB is
available [52].

4.3 Neurons in CA1 To assess the trajectory information represented in a neural ensem-
and RE Represent the ble in CA1 and RE, we constructed a decoder to understand the
Next Trajectory in the information about the animal’s next trajectory represented in these
Alternation Task neurons. The mean firing rates on each spatial bin on the stem were
used as the inputs of the classifier that predicts the animal’s next
trajectory. We used a one-leave-out cross-validation approach to
assess the performance of two decoders based on either population
vector or support vector machine. Both classifiers are able to predict
the animal’s next trajectory choice significantly better than a chance
level of 50%, suggesting that neurons in both CA1 and RE repre-
sent the information about the next trajectory as a population
(Fig. 5d). We further confirmed that the support vector machine
classifier performed better than a population vector decoder, indi-
cating excellent performance of the machine-learning-based
approach. To further confirm the signal directionality of the trajec-
tory information, we recorded CA1 neurons from the animals with
lesions in RE. We then implemented the same decoding approach
based on a support vector machine and found that the decoding
performance of the animal’s next trajectory was significantly
impaired in these animals (Fig. 5d), suggesting that RE is necessary
for CA1 neurons to express the trajectory information.

5 Optogenetic Approach to Confirm the Causality of Information Flow

A frequently asked question in multiregional studies is whether we

can tell the causality of impact between regions. While several
methods, including signal directionality analysis in the previous
section, are available to infer the causality, the direct method to
answer this question would be to manipulate the neural activity in
an anatomically and temporally specific manner. Historically, an
excitotoxic reagent, such as ibotenic acid or NMDA, has been
used to make permanent lesions in a particular brain region. For
example, as shown in the previous section, the trajectory-
dependent firing of CA1 cells was largely abolished by lesions of
RE, indicating that RE plays a key role for the hippocampus to
express the trajectory information. However, the major caveat of
such lesion studies is that it is often difficult to assess the role of the
brain region due to functional compensation by other brain areas
over time after damage. It has been demonstrated that different
behavioral deficits are observed between acute and chronic inacti-
vation of a particular brain region [53]. Furthermore, in the case of
the trajectory-dependent firing of place cells, the result of the lesion
experiment does not provide a clear answer to the question of
32 Hiroshi T. Ito

whether RE sends the trajectory information of ongoing trials

dynamically or is necessary for long-term circuit organization for
the hippocampus to express the trajectory information. In order to
make this distinction, temporally selective manipulation is neces-
sary. Here, the optogenetic manipulation technique [54, 55] is a
great tool for spatiotemporal and cell-type-specific manipulations.
A light-application system for optogenetic experiments can
easily be coupled with the recording system, as several companies,
such as Neuralynx or Plexon, offer an integrated system for both
recordings and optogenetics. It is also possible to build a separate
laser system that can be synchronized with the recording system
through TTL-mediated timestamps. In our system, a diode-
pumped solid-state laser or diode laser (Shanghai Laser & Optics
Century) is used as a coherent light source that is coupled with an
optic-fiber patch cable through a collimator (Thorlabs). The other
end of the patch cable is then connected through fiber ferrules with
the optic fiber implanted in the animal’s brain. To achieve the
widespread impact in the region, a high-NA large-diameter fiber
is often preferable (400 μm, 0.50 NA; FP400URT, Thorlabs).
We took an approach of virus-mediated expression of opsins in
the RE. The adeno-associated virus encoding opsins was injected
into RE, using a 10 μl NanoFil syringe with a UMP3 pump con-
troller (World Precision Instruments). The injection rate was
100 nl/min and ~0.5 μl in total for each injection site. The expres-
sion of opsins usually takes 3–4 weeks. Although the impact of
optogenetics can be validated by the fiber location as well as virus
expression pattern by histological examination after experiments, it
is ideal to assess the modulation of neuronal firing in behaving
animals directly. For this aim, two tetrode wires were glued to the
optic fiber such that the tips of the tetrodes are extended ~0.75 mm
from the end of the optic fiber (Fig. 6a). Individual wires of
tetrodes are then connected with a Mill-Max connector
(ED85100-ND, DigiKey) for the attachment of a recording device.
This configuration allows for the measurement of the activity of
neurons near the optic fiber to assess the impact of the light
application (Fig. 6b).
The laser light can be delivered whenever the silencing is nec-
essary by using a TTL-mediated switch of the laser source or by
using a mechanical shutter. Depending on experiments, the silenc-
ing may be necessary for a long duration, for example, in the range
of several minutes. However, such a long-time laser application is
known to generate heat that may influence neural activity nearby
[56]. Therefore, it is desirable to design experiments so that the
duration of continuous laser application should be as minimal as
possible.
As shown in Fig. 6b, when a 635 nm laser (RLM635TA;
Shanghai Laser & Optics Century) was applied to RE neurons
expressing Jaws [57], firing of a representative RE neuron was
 Multisite Recording for the Analysis of Information Flow Between. . . 33

Fig. 6 Optogenetic manipulation of the activity of RE neurons. (a) Top: Schema showing the design of optrode.
Two tetrodes are glued to the optic fiber so that the tetrodes should extend ~0.75 mm from the tip of the optic
fiber. Bottom. After the injection of AAV-Jaws, the optrode was inserted so that the tip of the optic fibers is
0.25–0.5 mm above the RE. (b) Silencing of a representative RE cell during laser application. The top plot
shows the spike raster and the bottom shows the spike-count histograms. The 635 nm laser was applied
between 0 and 5 sec in the time axis. The spike activity was almost completely silenced during the laser
application. (c) Trajectory-dependent firing of a CA1 cell during the silencing of RE. A representative CA1 place
cell showing the trajectory-dependent rate change between right- and left-oriented trajectories. When RE
neurons were silenced by the laser application, the same cell still exhibited position-selective firing at the
stem, but its trajectory-dependent rate change was largely diminished, indicating that RE cells dynamically
modulate the firing rates of CA1 cells depending on the animal’s trajectory plan of ongoing trials

strongly suppressed. The activity of a CA1 neuron recorded from

the same animal showed trajectory-dependent firing when the laser
was off (Fig. 6c). In the next behavioral session, the laser was
applied during the behaviors, resulting in silencing of RE neurons.
In order to minimize the duration of continuous laser application,
the laser was stopped during food consumption and restarted when
34 Hiroshi T. Ito

the animal initiated the next trials. Under the silencing of RE

neurons, while the same CA1 neuron still exhibited position-
selective firing at the stem part of the maze, the neuron lost its
trajectory dependency and the firing rates between right- and left-
oriented trials became almost the same. These data thus suggest
that RE neurons dynamically modulate firing rates of CA1 cells
during behaviors, sending the information about the next trajec-
tories to the hippocampus.

References

1. Singer W (1993) Synchronization of cortical
activity and its putative role in information
processing and learning. Annu Rev Physiol
55:349–374
2. Gregoriou GG, Gotts SJ, Zhou H, Desimone
R (2009) Long-range neural coupling through
synchronization with attention. Prog Brain Res
176:35–45
3. Fries P (2015) Rhythms for cognition: com-
munication through coherence. Neuron 88:
220–235
4. Jones MW, Wilson MA (2005) Theta rhythms
coordinate hippocampal-prefrontal interac-
tions in a spatial memory task. PLoS Biol 3:
e402
5. Ji D, Wilson MA (2007) Coordinated memory
replay in the visual cortex and hippocampus
during sleep. Nat Neurosci 10:100–107
6. Tort AB, Kramer MA, Thorn C, Gibson DJ,
Kubota Y, Graybiel AM, Kopell NJ (2008)
Dynamic cross-frequency couplings of local
field potential oscillations in rat striatum and
hippocampus during performance of a T-maze
task. Proc Natl Acad Sci U S A 105:
20517–20522
7. Colgin LL, Denninger T, Fyhn M, Hafting T,
Bonnevie T, Jensen O, Moser MB, Moser EI
(2009) Frequency of gamma oscillations routes
flow of information in the hippocampus.
Nature 462:353–357
8. Ito HT (2017) Prefrontal-hippocampal inter-
actions for spatial navigation. Neurosci Res
129:2–7
9. Rainer G, Rao SC, Miller EK (1999) Prospec-
tive coding for objects in primate prefrontal
cortex. J Neurosci 19:5493–5505
10. Gold JI, Shadlen MN (2007) The neural basis
of decision making. Annu Rev Neurosci 30:
535–574
11. O’Keefe J, Dostrovsky J (1971) The hippo-
campus as a spatial map. Preliminary evidence
from unit activity in the freely-moving rat.
Brain Res 34:171–175
12. O’Keefe J, Nadel L (1978) The hippocampus
as a cognitive map. Oxford University Press,
Oxford, UK
13. Moser EI, Kropff E, Moser MB (2008) Place
cells, grid cells, and the brain’s spatial represen-
tation system. Annu Rev Neurosci 31:69–89
14. Siapas AG, Lubenov EV, Wilson MA (2005)
Prefrontal phase locking to hippocampal theta
oscillations. Neuron 46:141–151
15. Benchenane K, Peyrache A, Khamassi M, Tier-
ney PL, Gioanni Y, Battaglia FP, Wiener SI
(2010) Coherent theta oscillations and reorga-
nization of spike timing in the hippocampal-
prefrontal network upon learning. Neuron 66:
921–936
16. Ito HT, Zhang SJ, Witter MP, Moser EI,
Moser MB (2015) A prefrontal-thalamo-hip-
pocampal circuit for goal-directed spatial navi-
gation. Nature 522:50–55
17. Hallock HL, Wang A, Griffin AL (2016) Ven-
tral midline thalamus is critical for
hippocampal-prefrontal synchrony and spatial
working memory. J Neurosci 36:8372–8389
18. Ito HT, Moser EI, Moser MB (2018) Supra-
mammillary nucleus modulates spike-time
coordination in the prefrontal-Thalamo-hippo-
campal circuit during navigation. Neuron 99:
576–587.e5
19. Herkenham M (1978) The connections of the
nucleus reuniens thalami: evidence for a direct
thalamo-hippocampal pathway in the rat. J
Comp Neurol 177:589–610
20. Vertes RP, Linley SB, Groenewegen HJ, Witter
MP (2014) Thalamus. In: Paxinos G (ed) The
rat nervous system. Elsevier Academic Press,
San Diego
21. McNaughton BL, O’Keefe J, Barnes CA
(1983) The stereotrode: a new technique for
simultaneous isolation of several single units in
the central nervous system from multiple unit
records. J Neurosci Methods 8:391–397
22. Gray CM, Maldonado PE, Wilson M,
McNaughton B (1995) Tetrodes markedly
improve the reliability and yield of multiple
Another random document with
no related content on Scribd:
large party of Miaki’s men, all armed, and watching as outposts.
Some were for shooting us, but others hesitated. Every musket was,
however, raised and levelled at me. Faimungo poised his great spear
and said, “No, you shall not kill Missi to-day. He is with me.” Having
made this flourish, he strode off after his own men, and my
Aneityumese followed, leaving me face to face with a ring of levelled
muskets. Sirawia, who was in command of this party, and who once
like Nowar had been my friend, said to me, Judas like, “My love to
you, Missi.” But he also shouted after Faimungo, “Your conduct is
bad in taking the Missi away; leave him to us to be killed!”
I then turned upon him, saying, “Sirawia, I love you all. You must
know that I sought only your good. I gave you medicine and food
when you and your people were sick and dying under measles; I gave
you the very clothing you wear. Am I not your friend? Have we not
often drunk tea and eaten together in my house? Can you stand there
and see your friend shot? If you do, my God will punish you
severely.”
He then whispered something to his company which I did not
hear; and, though their muskets were still raised, I saw in their eyes
that he had restrained them. I therefore began gradually to move
backwards, still keeping my eyes fixed on them, till the bush hid
them from my view, whereon I turned and ran after my party, and
God kept the enemy from following. I would like to think that
Sirawia only uttered the cruel words which I heard as a blind to save
his own life; for at this time he was joined to Miaki’s party, his own
people having risen against him, and had to dissemble his friendly
feelings towards me. Poor Sirawia! Well I knew that Miaki would
only use him as a tool for selfish interests, and sacrifice him at last.
All this showed how dangers grew around our path. Still we trusted
in Jehovah Jesus, and pressed on in flight. A second hostile party
encountered us, and with great difficulty we also got away from
them. Soon thereafter a friendly company crossed our path. We
learned from them that the enemies had slaughtered other two of
Manuman’s men, and burned several villages with fire. Another
party of the enemy encountered us, and were eager for our lives. But
this time Faimungo withstood them firmly, his men encircled us, and
he said, “I am not afraid now, Missi; I am feeling stronger near my
own land!”
 Hurrying still onwards, we came to that village on their high
ground called Aneai, i.e., Heaven. The sun was oppressively hot, the
path almost unshaded, and our whole party very exhausted,
especially Faimungo, carrying his load of stolen goods. So here he sat
down on the village dancing ground for a smoke, saying,—
“Missi, I am near my own land now. We can rest with safety.”
In a few minutes, however, he started up, he and his men, in wild
excitement. Over a mountain, behind the village and above it, there
came the shoutings, and anon the tramp, tramp of a multitude
making rapidly towards us. Faimungo got up and planted his back
against a tree. I stood beside him, and the Aneityumese woman and
the two men stood near me, while his men seemed prepared to flee.
At full speed a large body of the tallest and most powerful men that I
had seen on Tanna came rushing on and filled the dancing ground.
They were all armed, and flushed with their success in war. A
messenger had informed them of our escape, probably from Miaki,
and they had crossed the country to intercept us. Faimungo was
much afraid, and said,—
“Missi, go on in that path, you and your Aneityumese; and I will
follow when I have had a smoke and a talk with these men.”
I replied, “No, I will stand by your side till you go; and if I am
killed, it will be by your side. I will not leave you.”
He implored us to go on, but that I knew would be certain death.
They began urging one another to kill us, but I looked round them as
calmly as possible, saying, “My Jehovah God will punish you here
and hereafter, if you kill me or any of His servants.”
A killing-stone, thrown by one of the savages, grazed poor old
Abraham’s cheek, and the dear soul gave such a look at me, and then
upwards, as if to say, “Missi, I was nearly away to Jesus.” A club was
also raised to follow the blow of the killing-stone, but God baffled the
aim. They encircled us in a deadly ring, and one kept urging another
to strike the first blow or fire the first shot. My heart rose up to the
Lord Jesus; I saw Him watching all the scene. My peace came back to
me like a wave from God. I realized that I was immortal till my
Master’s work with me was done. The assurance came to me, as if a
voice out of Heaven had spoken, that not a musket would be fired to
wound us, not a club prevail to strike us, not a spear leave the hand
in which it was held vibrating to be thrown, not an arrow leave the
bow, or a killing-stone the fingers, without the permission of Jesus
Christ, whose is all power in Heaven and on Earth. He rules all
Nature, animate and inanimate, and restrains even the savage of the
South Seas. In that awful hour I saw His own words, as if carved in
letters of fire upon the clouds of Heaven: “Seek, and ye shall find.
Whatsoever ye shall ask in My name, that will I do, that the Father
may be glorified in the Son.” I could understand how Stephen and
John saw the glorified Saviour as they gazed up through suffering
and persecution to the Heavenly Throne! Yet I never could say that
on such occasions I was entirely without fear. Nay, I have felt my
reason reeling, my sight coming and going, and my knees smiting
together when thus brought close to a violent death, but mostly
under the solemn thought of being ushered into Eternity and
appearing before God. Still, I was never left without hearing that
promise in all its consoling and supporting power coming up through
the darkness and the anguish, “Lo, I am with you alway.” And with
Paul I could say, even in this dread moment and crisis of being, “I am
persuaded that neither death nor life ... nor any other creature, shall
be able to separate us from the love of God which is in Christ Jesus
our Lord.”
Faimungo and others now urged us to go on in the path. I said,
“Faimungo, why are we to leave you? My God heard your promise
not to betray me. He knows now what is in your heart and in mine. I
will not leave you; and if I am to die, I will die by your side.”
He replied, “Now, I go on before; Missi, keep close to me.”
His men had gone, and I persuaded my Aneityumese to follow
them. At last, with a bound, Faimungo started after them. I followed,
keeping as near him as I could, pleading with Jesus to protect me or
to take me home to Glory. The host of armed men also ran along on
each side with their weapons ready; but leaving everything to Jesus, I
ran on as if they were my escort, or as if I saw them not. If any reader
wonders how they were restrained, much more would I, unless I
believed that the same Hand that restrained the lions from touching
Daniel held back these savages from hurting me! We came to a
stream crossing our path. With a bound all my party cleared it, ran
up the bank opposite, and disappeared in the bush. “Faint yet
pursuing,” I also tried the leap, but I struck the bank and slid back on
my hands and knees towards the stream. At this moment I heard a
crash above my head amongst the branches of an overhanging tree,
and I knew that a killing-stone had been thrown, and that that
branch had saved me. Praising my God, I scrambled up on the other
side, and followed the track of my party into the bush. The savages
gazed after me for a little in silence, but no one crossed the stream;
and I saw them separate into two, one portion returning to the
village and another pressing inland. With what gratitude did I
recognise the Invisible One who brought their counsels to confusion!
I found my party resting in the bush, and amazed to see me
escaped alive from men who were thirsting for my blood. Faimungo
and his men received me with demonstrations of joy, perhaps feeling
a little ashamed of their own cowardice. He now ascended the
mountain and kept away from the common path to avoid other
Native bands. At every village enemies to the Worship were ready to
shoot us. But I kept close to our guide, knowing that the fear of
shooting him would prevent their shooting at me, as he was the most
influential Chief in all that section of the island.
One party said, “Miaki and Karewick said that Missi made the
sickness and the hurricanes, and we ought to kill him.”
Faimungo replied, “They lie about Missi! It is our own bad conduct
that makes us sick.”
They answered, “We don’t know who makes the sickness; but our
fathers have taught us to kill all foreign men.”
Faimungo, clutching club and spear, exclaimed, standing betwixt
them and us, “You won’t kill Missi to-day!”
In the flight we passed springs and streamlets, but though parched
with sickening thirst, not one of us durst stoop down to drink, as we
should have been almost certainly killed in the act. Faimungo now
sent his own men home by a near path, and guided us himself till we
were close upon the shore. There, sitting down he said,—
“Missi, I have now fulfilled my promise. I am so tired, I am so
afraid, I dare not go farther. My love to you all. Now go on quickly!
Three of my men will go with you to the next rocks. Go quickly!
Farewell.”
These men went on a little, and then said, “Missi, we dare not go!
Faimungo is at war with the people of the next land. You must keep
straight along this path.”
So they turned and ran back to their own village.
To us this district was especially perilous. Many years ago the
Aneityumese had joined in a war against the Tannese of this tribe,
and the thirst for revenge yet existed in their hearts, handed down
from sire to son. Besides, Miaki had incited the people here to
murder the Teachers and me if we attempted to escape this way.
Most providentially the men were absent on a war expedition, and
we saw only three lads and a great number of women and children,
who ran off to the bush in terror. In the evening the enraged savages
of another district assaulted the people of the shore villages for
allowing us to pass, and, though sparing their lives, broke in pieces
their weapons of war—a very grievous penalty. In the next district, as
we hasted along the shore, two young men came running after us,
poising their quivering spears. I took the useless revolver out of my
little native basket, and raising it cried,—
“Beware! Lay down your spears at once on the sand, and carry my
basket to the next landing at the black rocks.”
They threw their spears on the sand, lifted the bag, and ran on
before us to the rocks which formed the march betwixt them and
their enemies. Laying it down, they said appealingly, “Missi, let us
return to our home!” And how they did run, fearing the pursuit of
their foes.
In the next land we saw none. After that we saw crowds all along,
some friendly, others unfriendly, but they let us pass on, and with the
blessing of Almighty God we drew near to Mr. Mathieson’s Station in
safety. Here a man gave me a cocoa-nut for each of our party, which
we greatly required, having tasted nothing all that day, and very little
for several days before. We were so weak that only the struggle for
life enabled us to keep our feet; yet my poor Aneityumese never
complained and never halted, not even the woman. The danger and
excitement kept us up in the race for life, and by the blessing of God
we were now approaching the Mission House, praising God for His
wonderful deliverances.
Hearing of our coming, Mr. Mathieson came running to meet me.
They had heard of my leaving my own Station, and they thought I
was dead! They were themselves both very weak; their only child had
just been laid in the grave, and they were in great grief and in greater
peril. We praised the Lord for permitting us to meet; we prayed for
support, guidance, and protection; and resolved now, in all events, to
stand by each other till the last.
Before I left the Harbour I wrote and left with Nowar letters to be
given to the Captains of any vessels which called, for the first, and the
next, and the next, telling them of our great danger, that Mr.
Mathieson was almost without food, and that I would reward them
handsomely if they would call at the Station and remove any of us
who might be spared thence to Aneityum. Two or three vessels
called, and, as I afterwards learned, got my letters; but, while buying
my stolen property from the Natives for tobacco, powder, and balls,
they took no further notice of my appeals, and sailed past Mr.
Mathieson’s straight on to Aneityum. “The tender mercies of the
wicked are cruel!”
Let me now cull the leading events from my Journal, that
intervened betwixt this date and the final break-up of the Mission on
Tanna—at least for a season—though, blessed be God! I have lived to
see the light rekindled by my dear friends Mr. and Mrs. Watt, and
shining more brightly and hopefully than ever. The candle was
quenched, but the candle-stick was not removed!
On Wednesday, 22nd January, 1862, we heard that other three of
Manuman’s people had been killed and a district burned with fire.
Though this poor man was one of Nowar’s chief friends, yet I heard
him say before my flight, “When so many children are being killed,
why do they not send one for food to me and my family? They are as
tender and good as the young fowls!” A remark like this lets you see
deep into the heart of a Cannibal, and he a sort of half-converted one,
if I may use such an expression; certainly not one of the worst type
by any means.
On the 23rd January, Mr. Mathieson sent for Taura, Kati, and
Kapuku, his three principal Chiefs, to induce them to promise
protection till a vessel called to take us away. They appeared friendly,
and promised to do their best. Alas! the promises of the Tannese
Chiefs had too often proved to be vain.
On Friday, 24th January, report reached our Station that Miaki
and his party, hearing that a friendly Chief had concealed two of
Manuman’s young men, compelled him to produce them and club
them to death before their eyes. Also, that they surrounded
Manuman’s party on a mountain, and hemmed them in there, dying
of starvation and trying to survive on the carcases of the dead and on
bark and roots. Also, that Miaki had united all the Chiefs, friends and
foes alike, in a bond of blood, to kill every one pertaining to the
whole Mission on Tanna. Jesus rules.

SPRINGING FORWARD HE
CAUGHT THE CLUB.

On Sunday, the 26th January, thirty persons came to worship at

the Mission House. Thereafter, at great risk, we had Worship at three
of the nearest and most friendly villages. Amidst all our perils and
trials, we preached the Gospel to about one hundred and sixteen
persons. It was verily a sowing time of tears; but, despite all that
followed, who shall say that it was vain! Twenty years have passed,
and now when I am writing this, there is a Church of God singing the
praises of Jesus in that very district of Tanna. On leaving the second
village, a young lad affectionately took my hand to lead me to the
next village; but a sulky, down-browed savage, carrying a ponderous
club, also insisted upon accompanying us. I led the way, guided by
the lad. Mr. Mathieson got the man to go before him, while he
himself followed, constantly watching. Coming to a place where
another path branched off from ours, I asked which path we took,
and, on turning to the left as instructed by the lad, the savage getting
close behind me, swung his huge club over his shoulder to strike me
on the head. Mr. Mathieson, springing forward, caught the club from
behind with a great cry to me; and I, wheeling instantly, had hold of
the club also, and betwixt us we wrested it out of his hands. The poor
creature, craven at heart however blood-thirsty, implored us not to
kill him. I raised the club threateningly, and caused him to march in
front of us till we reached the next village fence. In terror lest these
Villagers should kill him, he gladly received back his club, as well as
the boy his bow and arrows, and they were lost in the bush in a
moment. At the village from which this man and boy had come, one
savage brought his musket while we were conducting worship, and
sat sullen and scowling at us all the time. Mocking questions were
also shouted at us, such as—“Who made the rains, winds, and
hurricanes? Who caused all the disease? Who killed Mr. Mathieson’s
child?” They sneered and scoffed at our answers, and in this Taura
the Chief joined the rest. They retorted that trading vessels had
called at the Harbour, and that all my clothes and property had been
sold for muskets, powder, caps, and balls, so that Miaki and his men
had plenty of ammunition for fighting purposes now! After this,
feeling that no one could be trusted, we ceased visiting these villages,
and refrained from exposing ourselves at any distance from the
Mission House.
On the 27th, at daylight, a vessel was seen in the offing, as if to
tantalize us. The Captain had been at the Harbour, and had received
my letter from Nowar. I hoisted a flag to induce him to send or come
on shore, but he sailed off for Aneityum, bearing the plunder of my
poor Mission House, purchased for ammunition and tobacco from
the Natives. He left the news at Aneityum that I had been driven
from my Station some time ago, and was believed to have been
murdered.
On the 29th January, the young Chief Kapuku came and handed to
Mr. Mathieson his own and his father’s war-gods and household
idols. They consisted chiefly of a basket of small and peculiar stones,
much worn and shining with use. He said,—
“While many are trying to kill you and drive the worship of
Jehovah from this island, I give up my gods, and will send away all
Heathen idols from my land.”
 On the 31st, we learned that a party of Miaki’s men were going
about Mr. Mathieson’s district inciting the people to kill us.
Faimungo also came to inform us that Miaki was exerting all his
artifice to get us and the Worship destroyed. Manuman even sent,
from inland, Raki, his adopted son, to tell me of the fearful sufferings
that he and his people were now passing through, and that some
were killed almost every day. Raki’s wife was a Chief’s daughter, who,
when the war began, returned to her father’s care. The savages of
Miaki went to her own father’s house and compelled him to give her
up as an enemy. She was clubbed and feasted on.
On Sabbath, 2nd February, thirty-two people attended the
morning service. I addressed them on the Deluge, its causes and
lessons. I showed them a doll, explaining that such carved and
painted images could not hear our prayers or help us in our need,
that the living Jehovah God only could hear and help. They were
much interested, and after Worship carefully examined the doll. Mr.
Mathieson and I, committing ourselves to Jesus, went inland and
conducted worship at seven villages, listened to by in all about one
hundred people. Nearly all appeared friendly. The people of one
village had been incited to kill us on our return; but God guided us to
return by another way, and so we escaped.
During the day, on 3rd February, a company of Miaki’s men came
to the Mission House, and forced Mrs. Mathieson to show them
through the premises. Providentially, I had bolted myself that
morning into a closet room, and was engrossed with writing. They
went through every room in the house and did not see me,
concluding I had gone inland. They discharged a musket into our
Teacher’s house, but afterwards left quietly, greatly disappointed at
not finding me. My heart still rose in praise to God for another such
deliverance, neither by man nor of man’s planning!
Worn out with long watching and many fatigues, I lay down that
night early, and fell into a deep sleep. About ten o’clock the savages
again surrounded the Mission House. My faithful dog Clutha,
clinging still to me amid the wreck of all else on Earth, sprang quietly
upon me, pulled at my clothes, and awoke me, showing danger in her
eye glancing on me through the shadows. I silently awoke Mr. and
Mrs. Mathieson, who had also fallen asleep. We committed ourselves
in hushed prayer to God and watched them, knowing that they could
not see us. Immediately a glare of light fell into the room! Men
passed with flaming torches; and first they set fire to the Church all
round, and then to a reed fence connecting the Church and the
dwelling-house. In a few minutes the house, too, would be in flames,
and armed savages waiting to kill us on attempting an escape! Taking
my harmless revolver in the left hand and a little American
tomahawk in the right, I pled with Mr. Mathieson to let me out and
instantly again to lock the door on himself and wife. He very
reluctantly did so, holding me back and saying,—
“Stop here and let us die together! You will never return!”
I said, “Be quick! Leave that to God. In a few minutes our house
will be in flames, and then nothing can save us.”
He did let me out, and locked the door again quickly from the
inside; and, while his wife and he prayed and watched for me from
within, I ran to the burning reed fence, cut it from top to bottom, and
tore it up and threw it back into the flames, so that the fire could not
by it be carried to our dwelling-house. I saw on the ground shadows,
as if something were falling around me, and started back. Seven or
eight savages had surrounded me, and raised their great clubs in air.
I heard a shout—“Kill him! kill him!” One savage tried to seize hold
of me, but, leaping from his clutch, I drew the revolver from my
pocket and levelled it as for use, my heart going up in prayer to my
God. I said,—
“Dare to strike me, and my Jehovah God will punish you! He
protects us, and will punish you for burning His Church, for hatred
to His Worship and people, and for all your bad conduct. We love
you all; and for doing you good only you want to kill us. But our God
is here now to protect us and to punish you.”
They yelled in rage, and urged each other to strike the first blow,
but the Invisible One restrained them. I stood invulnerable beneath
His invisible shield, and succeeded in rolling back the tide of flame
from our dwelling-house.
At this dread moment occurred an incident, which my readers may
explain as they like, but which I trace directly to the interposition of
my God. A rushing and roaring sound came from the South, like the
noise of a mighty engine or of muttering thunder. Every head was
instinctively turned in that direction, and they knew, from previous
hard experience, that it was one of their awful tornadoes of wind and
rain. Now, mark, the wind bore the flames away from our dwelling-
house, and had it come in the opposite direction, no power on Earth
could have saved us from being all consumed! It made the work of
destroying the Church only that of a few minutes; but it brought with
it a heavy and murky cloud, which poured out a perfect torrent of
tropical rain. Now, mark again, the flames of the burning Church
were thereby cut off from extending to and seizing upon the reeds
and the bush; and, besides, it had become almost impossible now to
set fire to our dwelling-house. The stars in their courses were
fighting against Sisera! The mighty roaring of the wind, the black
cloud pouring down unceasing torrents, and the whole surroundings,
awed those savages into silence. Some began to withdraw from the
scene, all lowered their weapons of war, and several, terror-struck,
exclaimed,—
“That is Jehovah’s rain! Truly their Jehovah God is fighting for
them and helping them. Let us away!”
A panic seized upon them; they threw away their remaining
torches; in a few moments they had all disappeared in the bush; and
I was left alone, praising God for His marvellous works. “O taste and
see that God is good! Blessed is the man that trusteth in Him!”
Returning to the door of the Mission House, I cried,—
“Open and let me in. I am now all alone.”
Mr. Mathieson let me in, and exclaimed,—
“If ever, in time of need, God sent help and protection to His
servants in answer to prayer, He has done so to-night! Blessed be His
holy name!”
In fear and in joy we united our praises. Truly our Jesus has all
power, not less in the elements of Nature than in the savage hearts of
the Tannese. Precious Jesus! Does He not chide us, saying,
—“Hitherto ye have asked nothing in My Name. Ask and ye shall
receive, that your joy may be full!”? How much help, blessing, and
joy we lose every day, because we do not take all to Jesus as we
ought! Often since have I wept over His love and mercy in that
deliverance, and prayed that every moment of my remaining life may
be consecrated to the service of my precious Friend and Saviour!
 All through the remainder of that night I lay wide awake keeping
watch, my noble little dog lying near me with ears alert. Early in the
morning friends came weeping around us. Our enemies were loudly
rejoicing. It had been finally resolved to kill us at once, to plunder
our house and then to burn it. The noise of the shouting was
distinctly heard as they neared the Mission premises, and our
weeping, friendly Natives looked terror-struck, and seemed anxious
to flee for the bush. But just when the excitement rose to the highest
pitch, we heard, or dreamed that we heard, a cry higher still, “Sail
O!” We were by this time beginning to distrust almost our very
senses; but again and again that cry came rolling up from the shore,
and was repeated from crowd to crowd all along the beach, “Sail O!
Sail O!” The shouts of those approaching us gradually ceased, and
the whole multitude seemed to have melted away from our view. I
feared some cruel deception, and at first peered out very cautiously
to spy the land. But yonder in very truth a vessel had sailed into the
bay. It was the Blue Bell, Captain Hastings. I set fire to the reeds on
the side of the hill to attract his attention. I put a black shawl as a flag
on one end of the Mission House and a white sheet on the other.
This was one of the vessels that had been to Port Resolution, and
had sailed past to Aneityum some time ago. I afterwards saw the
mate and some of the men wearing my shirts, which they had bought
from the Tannese on their former visit. At the earnest request of
Doctors Geddie and Inglis, Mr. Underwood, the owner, had sent
Captain Hastings to Tanna to rescue us if yet alive. For this purpose
he had brought twenty armed men from Aneityum, who came on
shore in two boats in charge of the mate, the notorious Ross Lewin.
He returned to the ship with a boat-load of Mr. Mathieson’s things,
leaving ten of the Natives to help us to pack more and carry them
down to the beach, especially what the Missionary thought most
valuable.
The two boats were now loaded and ready to start. It was about
two o’clock in the afternoon, when a strange and painful trial befell
us. Poor dear Mr. Mathieson, apparently unhinged, locked himself
all alone into what had been his study, telling Mrs. Mathieson and
me to go, for he had resolved to remain and die on Tanna. We tried
to show him the inconsistency of praying to God to protect us or
grant us means of escape, and then refuse to accept a rescue sent to
us in our last extremity. We argued that it was surely better to live
and work for Jesus than to die as a self-made martyr, who, in God’s
sight, was guilty of self-murder. His wife wept aloud and pled with
him, but all in vain! He refused to leave or to unlock his door. I then
said,—
“It is now getting dark. Your wife must go with the vessel, but I will
not leave you alone. I shall send a note explaining why I am forced to
remain; and as it is certain that we shall be murdered whenever the
vessel leaves, I tell you God will charge you with the guilt of our
murder.”
At this he relented, unlocked the door, and accompanied us to the
boats, in which we all immediately left.
Meantime, having lost several hours, the vessel had drifted
leeward; darkness suddenly settled upon us, and when we were out
at sea we lost sight of her and she of us. After drifting about for some
hours in a heavy sea and unable to find her, those in charge of the
boats came near for consultation, and, if possible, to save the lives of
all. We advised that they should steer for Port Resolution by the
flame of the Volcano—a never-failing light-house, seen fifty miles
away—and there await the vessel. The boats were to keep within
hearing of each other by constant calling; but this was soon lost to
the ear, though on arriving in the bay we found they had got to
anchor before us. There we sat in the boats and waited for the
coming day. As the light appeared, we anchored as far out as
possible, beyond the reach of musket shots; and there without water
or food we sat under a tropical sun till mid-day came, and still there
was no sign of the vessel. The mate at last put all the passengers and
the poorest seamen into one boat and left her to swing at anchor;
while, with a strong crew in the other, he started off in search of the
vessel.
In the afternoon, Nowar and Miaki came off in a canoe to visit us.
Nowar had on a shirt, but Miaki was naked and frowning. He urged
me to go and see the Mission House, but as we had seen a body of
men near it I refused to go. Miaki declared that everything remained
as I had left it, but we knew that he lied. Old Abraham and a party
had slipped on shore in a canoe, and had found the windows
smashed and everything gone except my books, which were scattered
about and torn in pieces. The armed men there wanted to kill the
Aneityumese, but others said, “Not till we get Missi killed too!” They
learned that Miaki had sold everything that he could sell to the
Traders. The mate and men of the Blue Bell had on my very clothes.
They boasted that they had bought them for a few figs of tobacco and
for powder, caps, and balls. But they would not return a single shirt
to me, though I was without a change! We had all been without food
in the boat since the morning before, so Nowar brought us off a
cocoa-nut each, and two very small roasted yams for the ladies.
Those, however, only seemed to make our thirst the more severe, and
we spent a trying day in that boat under a burning sun. Miaki said,—
“As our fathers did not destroy Missi Turner’s house, we will not
destroy yours.”
But after a time, failing to persuade me to accompany him and fall
into a trap, he muttered,—
“We have taken everything your house contained, and would take
you too if we could; for we hate the Worship, it causes all our
diseases and deaths; it goes against our customs, and it condemns
the things we delight in.”
Nowar informed me that only a few nights before this, Miaki and
his followers went inland to a village where last year they had killed
ten men. Having secretly placed a savage at the door of every house,
at a given signal they yelled, and when the terrified inmates tried to
escape they killed almost every man, woman, and child. Some fled
into the bush, others rushed to the shore. A number of men got into a
canoe to escape, but hearing women and children crying after them
they returned, and taking those they could with them they killed the
rest lest they should fall alive into Miaki’s hands. These are surely
“they who through fear of death are all their lifetime subject to
bondage.” The Chief and nearly his whole village were cut off in one
night! Not an uncommon thing in those Islands, where war becomes
chronic, and the thirst for blood becomes insatiable. The dark places
of the Earth are “full of the habitations of horrid cruelty.” To have
actually lived amongst the Heathen and seen their life gives a man a
new appreciation of the power and blessings of the Gospel, even
where its influence is only very imperfectly allowed to guide and
restrain the passions of men. Oh, what will it be when all men in all
nations love and serve the glorious Redeemer!
 This Miaki and his followers were a scourge and terror to the
whole island of Tanna. They intensely hated Nowar, because he
would not join in their cruelties. Yet he and Manuman and Sirawia
and Faimungo continued to survive long after war and death had
swept all the others away. The first three lived to be very old men,
and to the last they made a profession of being Christians, though
their knowledge was very limited and their inconsistencies very grave
and very numerous. Happy is it for us that we are not the judges, for
souls either of the white or the dark skin, as to how many and
grievous things may be forgiven, and whether there be or be not that
spark of love, that grain of faith which the Lord the Pitiful will
graciously accept and increase![1]
1. See Appendix A. “The Prayer of the Tannese,” etc.
About five o’clock in the evening the vessel hove in sight. Before
dark we were all safely on board, and were sailing for Aneityum.
Though both Mr. and Mrs. Mathieson had become very weak, they
stood the voyage wonderfully. Next day we were all safely landed. We
had offered Captain Hastings £20 to take us to Aneityum, but he
declined any fare. However, we divided it amongst the mate and
crew, for they had every one shown great kindness to us on the
voyage. After arriving on Aneityum, Mrs. Mathieson gradually sank
under consumption, and fell asleep in Jesus on 11th March, 1862, in
the full assurance of a glorious resurrection, and was interred there.
Mr. Mathieson, becoming more and more depressed after her death,
went over to Mr. Creagh’s Station, on Maré, and there died on 14th
June, 1862, still trusting in Jesus, and assured that he would soon be
with Him in Glory. Never more earnest or more faithful souls
entered the Mission field, but they both suffered from weakness and
ill-health during all their time on Tanna, and had frequently to seek
change by removal for a short period from the island. Their memory
is very fragrant to me as fellow-labourers in the Gospel of Jesus.
After their death, I was the only one left alive in all the New
Hebrides Mission north of Aneityum to tell the story of those pioneer
years, during which were sown the seeds of what is now fast
becoming a glorious harvest. Twenty-five years ago, all these dear
brethren and sisters who were associated with me in the work of the
Mission were called home to Glory, to cast their crowns at the feet of
Jesus and enjoy the bliss of the redeemed, while I am privileged still
to toil and pray for the salvation of the poor Islanders, and plead the
cause of the Mission both in the Colonies and at home, in which work
the Lord has graciously given me undreamt-of success. My constant
desire and prayer are that I may be spared to see at least one
Missionary on every island of the group, to unfold the riches of
redeeming love and to lead the poor Islanders to Jesus for salvation.
What could be taken in three boats was saved out of the wreck of
Mr. Mathieson’s property; but my earthly all perished, except the
Bible and the translations into Tannese. Along with the goods
pertaining to the Mission, the property which I had to leave behind
would be undervalued at £600, besides the value of the Mission
House, etc. Often since have I thought that the Lord stripped me thus
bare of all these interests, that I might with undistracted mind
devote my entire energy to the special work soon to be carved out for
me, and of which at this moment neither I nor any one had ever
dreamed. At any rate, the loss of my little earthly all, though
doubtless costing me several pangs, was not an abiding sorrow like
that which sprang from the thought that the Lord’s work was broken
up at both Stations, and that the Gospel was for the time driven from
Tanna.
In the darkest moment, I never doubted that ultimately the victory
there, as elsewhere, would be on the side of Jesus, believing that the
whole Earth would yet be filled with the glory of the Lord. But I
sometimes sorely feared that I might never live to see or hear of that
happy day! By the goodness of the Ever-merciful One I have lived to
see and hear of a Gospel Church on Tanna, and to read about my
dear fellow-Missionaries, Mr. and Mrs. Watt, celebrating the Holy
Supper to a Native Congregation of Tannese, amid the very scenes
and people where the seeds of faith and hope were planted not only
in tears, but tears of blood,—“in deaths oft.”
My own intention was to remain on Aneityum, go on with my work
of translating the Gospels, and watch the earliest opportunity, as God
opened up my way, to return to Tanna. I had, however, got very weak
and thin; my health was undoubtedly much shaken by the continued
trials and dangers through which we had passed; and therefore, as
Dr. and Mrs. Inglis were at home carrying the New Testament
through the press in the language of Aneityum, and as Tanna was
closed for a season, Dr. Geddie, the Rev. Joseph Copeland, and Mr.
Mathieson all urged me to go to Australia by a vessel then in the
Harbour and leaving in a few days. My commission was to awaken an
interest among the Presbyterian Churches of our Colonies in this
New Hebrides Mission which lay at their doors, up till this time
sustained by Scotland and Nova Scotia alone. And further, and very
specially, to raise money there, if possible, to purchase a new Mission
Ship for the work of God in the New Hebrides,—a clamant necessity,
which would save all future Missionaries some of the more terrible of
the privations and risks of which a few examples have in these pages
already been recorded.
After much prayerful deliberation with my brethren and with my
own heart before God, I somewhat reluctantly felt constrained to
undertake the task. If my story was to be the means of providing
more Missionaries for the Islands, and of providing a commodious
Ship for the service of the Mission alone, to keep open their
communications with the outer world and with Christian influences,
not to speak of carrying their provisions at fixed periods, or rescuing
them when in troubles and perils from the jaws of death, I was not
unwilling to tell it again and again, if the Lord would open up my
path. God knows my heart, and any one who really knows me will
easily admit, that no selfish or egotistical motive has influenced me
in reciting through all the Australasian Colonies, New Zealand,
Scotland, and latterly in many parts of England and Ireland, the
incidents of my career and experience, first of all on Tanna, and
thereafter for nearly twenty years—as the Second Part of my
biography will relate—on the neighbouring island of Aniwa; an island
entirely given to me by the Lord, the whole population of which
became Christian; and they and their race will be my crown of joy
and rejoicing in the day of the Lord Jesus.
With regrets, and yet with unquenchable hope for these Islands, I
embarked for Australia, having received the solemn promise of my
brethren, that in entering upon this great effort I was to be left
absolutely free of all control, and empowered to carry out the work as
God might seem to guide me, and open up my way. I had only
spoken to one man in Sydney; all the doors to influence had
therefore to be unlocked, and I had no helper, no leader, but the
Spirit of my Lord. The Second Part of this Autobiography, should
God spare me to write it, will record His marvellous goodness in
using my humble voice and pen and the story of my life for
interesting thousands and tens of thousands in the work of Missions,
and especially for binding together the children of the Sabbath
Schools of Australasia in a Holy League of help to the New Hebrides,
which has already borne precious fruit to His glory, and will continue
to do so for ages to come.
Oftentimes, while passing through the perils and defeats of my
first years in the Mission field on Tanna, I wondered, and perhaps
the reader hereof has wondered, why God permitted such things. But
on looking back now, I already clearly perceive, and the reader of my
future pages will, I think, perceive, that the Lord was thereby
preparing me for doing, and providing me materials wherewith to
accomplish the best work of all my life—the kindling of the heart of
Australian Presbyterianism with a living affection for these Islanders
of their own Southern Seas—the binding of all their children into a
happy league of shareholders, first in one Mission Ship, and finally in
a larger and more commodious Steam-Auxiliary, and, last of all, in
being the instrument under God of sending out Missionary after
Missionary to the New Hebrides, to claim another island and still
another for Jesus. That work, and all that may spring from it in time
and Eternity, never could have been accomplished by me, but for
first the sufferings and then the story of my Tanna enterprise!
Some unsophisticated souls who read these pages will be
astonished to learn, but others who know more of the heartless
selfishness of human creatures, will be quite prepared to hear, that
my leaving Tanna was not a little criticized, and a great deal of
nonsense was written, even in Church Magazines, about the breaking
up of the Mission. All such criticism came, of course, from men who
were themselves destitute of sympathy, and who, probably, never
endured one pang for Jesus in all their comfortable lives. Conscious
that I had, to the last inch of life, tried to do my duty, I left all results
in the hands of my only Lord, and all criticisms to His unerring
judgment. Hard things also were occasionally spoken to my face. One
dear friend, for instance, said,—
“You should not have left. You should have stood at the post of
duty till you fell. It would have been to your honour, and better for
the cause of the Mission, had you been killed at the post of duty like
the Gordons and others.”
 I replied,—“I regard it as a greater honour to live and to work for
Jesus, than to be a self-made martyr. God knows that I did not refuse
to die; for I stood at the post of duty, amid difficulty and danger, till
all hope had fled, till everything I had was lost, and till God, in
answer to prayer, sent a means of escape. I left with a clear
conscience, knowing that in doing so I was following God’s leading,
and serving the Mission too. To have remained longer would have
been to incur the guilt of self-murder in the sight of God.”
Never for one moment have I had occasion to regret the step then
taken. The Lord has so used me, during the five-and-twenty years
that have passed over me since my farewell to Tanna, as to stamp the
event with His own most gracious approval. Oh, to see a Missionary,
and Christian Teachers, planted on every island of the New
Hebrides! For this I labour, and wait, and pray. To help on the
fulfilment thereof is the sacred work of my life, under God. When I
see it accomplished, or in a fair way of being so, through the
organization that will provide the money and call forth the men, I
can lay down my head as peacefully and gratefully as ever warrior
did, with the shout of victory in his ears,—“Lord, now lettest Thou
Thy servant depart in peace!”

For the present, my pen is here laid aside. I shall wait to see what
use the Lord makes of Part First of my autobiography, before I
prosecute the theme. If the Christian public seems not to find in it
the help and quickening that some friends think it likely to bestow on
those who read, the remainder need not be written. Part Second, if
called for, will contain a record, in many respects, an utter contrast
to all that has gone before, and yet directly springing therefrom, as
will be seen by all who look beneath the surface. I am penning these
words in 1887, and five-and-twenty years lie betwixt this date and
my farewell to Tanna. These years, if ever published, will tell the
story of my visiting all the Colonial Churches, and collecting the
purchase money of our white-winged Mission Ship, the Dayspring;
my return to Scotland, visiting all the home congregations in 1864,
and securing several new Missionaries to follow me to the New
Hebrides; my second marriage, and settlement on Aniwa, with her
whom the good Lord still spares to me, the mother of our happy
family, and my God-given helpmeet in all the work of the Gospel; the
conversion of that whole island of Aniwa from idolatry, and the
planting there of a Church and Congregation of Christ, from which
have since gone forth many Native Evangelists and Teachers. Then
there will fall to be recorded my call from the Islands in recent years
to revisit all the Colonial Presbyterian Congregations once again,
telling them the story of the Conversion of Aniwa—the sinking of the
well, and other incidents, which turned an entire people from idols
and from cannibalism to the service of the living and true God—
whereby the Churches, and especially the children, were led more
and more to make the New Hebrides their own very harvest field in
the Heathen world. And finally, I will have to tell how I was again
sent home to Scotland in 1884 to raise money for the purchase or
building of a steam-auxiliary Mission Ship, now urgently required in
the interests of the Mission, both because of the great increase in the
number of the Missionaries and the necessities of so many families;
and also and chiefly to avert the dreadful disappointments and loss
of time, and thereby sometimes of life itself, caused by the frequent
becalming of our little Dayspring in these thickly-islanded seas. That
part of the story will show the fruits of the education and perils and
experiences of a lifetime, in the marvellous impression produced by
the simple and unadorned recital of the story of Tanna and Aniwa,
amongst the Christian people of Scotland, Ireland, and England.
Multitudes were blessed in almost every town where a meeting was
granted me. Three Missionaries devoted themselves to the New
Hebrides, and are already labouring there; while others consecrated
themselves to several of the great seats of Foreign Mission enterprise
in Africa and Asia. I returned to my own Church of Victoria with a
sum of nearly £9,000, of which £6,000 was for the new Missionary
Steam-Auxiliary, and the remainder for the outfit and support of
more Missionaries for the Islands; and that money I handed over to
the Australian Church, where it awaits, at interest in the bank, the
arrangements being made by all the Colonies to take each their due
share in the future up-keep of the Ship. For this—for everything—for
all, praise be to the Lord! I never asked one subscription, except in
prayer and in my public appeals. The Lord sent in all freely to me
through the hands of His people; to Him be all the glory. I went back
to Aniwa, and found the work of the Lord going forward there as if in
a regularly settled Congregation at home, fostered and guided by an
occasional visit of my ever dear and genuine friends, Mr. and Mrs.