https://t.

me/MBS_MedicalBooksStore
 i

The Interoceptive Mind
ii
 iii

The Interoceptive Mind
From Homeostasis to
Awareness
Edited by

Manos Tsakiris
Helena De Preester

1
iv

1
Great Clarendon Street, Oxford, OX2 6DP
United Kingdom
Oxford University Press is a department of the University of Oxford.
It furthers the University’s objective of excellence in research, scholarship,
and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries
© Oxford University Press 2019
The moral rights of the authors‌have been asserted
Impression: 1
First Edition published in 2019
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system, or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press, or as expressly permitted
by law, by licence or under terms agreed with the appropriate reprographics
rights organization. Enquiries concerning reproduction outside the scope of the
above should be sent to the Rights Department, Oxford University Press, at the
address above
You must not circulate this work in any other form
and you must impose this same condition on any acquirer
Published in the United States of America by Oxford University Press
198 Madison Avenue, New York, NY 10016, United States of America
British Library Cataloguing in Publication Data
Data available
Library of Congress Control Number: 2018945622
ISBN 978–​0–​19–​881193–​0
Printed and bound by
CPI Group (UK) Ltd, Croydon, CR0 4YY
Oxford University Press makes no representation, express or implied, that the
drug dosages in this book are correct. Readers must therefore always check
the product information and clinical procedures with the most up-​to-​date
published product information and data sheets provided by the manufacturers
and the most recent codes of conduct and safety regulations. The authors and
the publishers do not accept responsibility or legal liability for any errors in the
text or for the misuse or misapplication of material in this work. Except where
otherwise stated, drug dosages and recommendations are for the non-​pregnant
adult who is not breast-​feeding
Links to third party websites are provided by Oxford in good faith and
for information only. Oxford disclaims any responsibility for the materials
contained in any third party website referenced in this work.
 v

Preface

Interoception is the body-​to-​brain axis of signals originating from the internal body and
visceral organs (such as gastrointestinal, respiratory, hormonal, and circulatory systems).
It plays a unique role in ensuring homeostasis. Interoception therefore refers to the sensing
of the state of the inner body and its homeostatic needs, to the ever-​fluctuating state of
the body beneath its sensory (exteroceptive) and musculoskeletal sheath. By bringing
together the perspectives of experimental psychology and cognitive neuroscience, psy-
chophysiology, psychiatry, clinical psychology, and philosophy, this volume aims to go
beyond the known role of interoception for homeostasis in order to ask, and hopefully
provide, important insights on the role that interoception plays for our mental life and
lived experience, for awareness, affect, and cognition.
The perspectives in the ensuing 17 chapters largely fall within the embodied cognition
approach that attempted to ground cognition and the self in the body. Over the last three
decades, modern psychology and cognitive neurosciences have focused on the impor-
tance of the body as the starting point for a science of the self and the subject. However,
this focus concerned the body as perceived from the outside, as, for example, when we
recognize ourselves in the mirror, or when the brain integrates sensorimotor information
to create our sense of body ownership and agency, or even when we perceive other people’s
bodies and thereby their mental states via processes of embodied simulation. A first step
toward that direction was to consider the role that sensorimotor signals play for the expe-
rience and the awareness of one’s self, which goes beyond their well-​known role in motor
control and sensory perception. For example, research on the sense of agency over one’s
actions and the sense of ownership of one’s body demonstrates how these fundamental
experiences rely on specific processes of sensorimotor interaction and multisensory inte-
gration, respectively (Haggard, 2005; Tsakiris, 2010). Another, closely related step in that
direction was the recognition of the proprioceptive dimension of the body for agency,
ownership, and the embodied self (Gallager & Cole, 1995; Bermúdez, Marcel, & Eilan,
1995). In the prolific field of social cognitive neuroscience, similar considerations re-
garding the role of observable sensorimotor events have been extended to social cogni-
tion and our understanding of other minds (Gallese, Keysers, & Rizzolatti, 2004). Such
theoretical advances have been instrumental in explaining key aspects of the bodily self
and its social cognition.
Notwithstanding the influential research that accumulated in this area, it is clear that
our fields have neglected another important dimension of the body, namely the inter-
oceptive body, which is the body as perceived from the inside. This visceral dimension
of embodiment has, of course, a long-​standing history of predominantly physiological
investigations during the last century, but it was only relatively recent that interoception
vi

vi Preface

has gained a rapidly expanding interest in the study of human mind. The seminal work of
Bud Craig (2009) and Antonio Damasio (1999), despite their differences, placed the vis-
ceral milieu, its homeostatic functioning, and our interoceptive awareness thereof at the
center stage of self-​awareness. In parallel, the development of new methods to measure,
not simply interoceptive signaling but our awareness thereof opened up the field of in-
teroceptive research to psychologists and cognitive neuroscientists. In parallel with the
exponential increase in the number of publications on the topic, researchers and scholars
across physiology, cognitive neuroscience, psychology, philosophy, and medicine are
working on interoception without necessarily sharing the same conceptual base or nec-
essarily realizing how their investigations link with the findings and insights of other
disciplines and methodologies.
Despite recent advances (Garfinkel et  al., 2015; Kleckner et  al., 2015), there is con-
sensus that interoception research must develop psychologically relevant and philosoph-
ically sound theoretical foundations, a wider and more grounded measurement model
and a fuller characterization of the links between different interoceptive dimensions
and systems if it is to achieve its appropriate place within the life and mind sciences.
This ambitious aim necessitates wide-​ranging, dedicated, and systematic theoretical and
methodological enquiries into the hierarchical relations in interoceptive processing, the
horizontal relations across interoceptive modalities, and the causal relations between
interoception and awareness. For example, psychological research into interoceptive
awareness has mainly used tasks that quantify our accuracy in detecting single heartbeats.
However, as the influential work of Garfinkel, Seth, Barrett, Suzuki, and Critchley (2015)
shows, we must be aware of the hierarchical levels of interoception, from interoceptive
sensibility, to accuracy and eventual awareness, and how these may impact cognition in
health and in illness. Similarly, in relation to horizontal relations, classic approaches to
interoception focus on four systems—​the cardiovascular, respiratory, gastrointestinal,
and urogenital (Adam, 1998; Cameron, 2002), but a fundamental question concerns
the interrelation of awareness across different interoceptive systems and the potentially
distinctive role that each system plays for cognition. Finally, unlike exteroception, it is
particularly difficult to have experimental control over the inputs to the interoceptive
system and/​or to interfere causally with it, and therefore developing paradigms and the-
oretical approaches that can probe the causal links between interoception and cognition
will accelerate our knowledge.
The contributions collected in this multidisciplinary volume represent an attempt to
provide a reference for the conceptualization of this excitingly deep connection between
our body and mind. As such it offers an overview of the state-​of-​art in psychological
and neuroscience research, of recent developments in clinical-​psychological models for
normal and pathological functioning, and of new theories that frame interoception at the
intersection between philosophy of mind and the broader context of embodied cognition.
To that end, its scope ranges from the psychology and neuroscience of interoception (Part
I), to clinical implications of recent research taking into account interoception (Part II),
and to theoretical-​philosophical frameworks and models of interoception (Part III).
 vi

Preface vii

The introductory chapter by Berntson, Gianaros, and Tsakiris goes straight into the
heart of the matter by providing the historical context that led to the development of a
science of interoception and a state-​of-​the-​art overview of the organizational principles
of the interoceptive system. First, it is explained that the autonomic nervous system is
not only or even primarily an efferent motor system (historically the predominant view)
but constitutes an elaborate afferent system. Second, the authors explain that the intero-
ceptive system does not operate independently of higher brain functions. By addressing
the top-​down and bottom-​up organizational principles that underpin interoception, this
introductory chapter sets the stage for the subsequent chapters in the volume.
Part I, Mentalizing interoception: Advances and challenges focuses on recent advances
in experimental psychology and cognitive neuroscience that highlight the role that inter-
oceptive signals and awareness thereof play for our mental life, beyond their known role
for homeostasis. Allen and Tsakiris consider the implications of the embodied predictive
processing account for the conceptualization of interoceptive signals as “first priors” and
their role in providing the mind with a biologically plausible model of one’s body and
self. Babo-​Rebelo and Tallon-​Baudry review recent electrophysiological studies that ex-
tend the role of interoception to show how the cortical processing of cardiac signals may
generate a subject-​centered reference frame that may underlie different and perhaps dis-
tinct facets of the self, such as thought generation and visual consciousness. Wittmann
and Meissner discuss the embodiment of subjective time and present evidence to show
how the accumulation of physiological signals forms the basis for the subjective impres-
sion of time. Aziz and Ruffle delve deeper into the viscera to describe the bi-​directional
brain−gut axis whose function underpins the generation of “gut-​feelings.” Such feeling
states are important for sensations but also the experience of distinct and often salient
experiences such as pain, nausea, and appetite. The last chapter of this first part by Von
Morh and Fotopoulou focuses on the homeostatically relevant experiences of pain and
pleasure, in particular affective touch, and discuss their peripheral neurophysiological
specificity (i.e. bottom-​up) and their top-​down social modulations within a predictive
coding framework.
Part II: From health to disease: Interoception in physical and mental health considers
the role of interoceptive processes and their corresponding psychological concepts across
a range of clinical conditions, from aberrant emotional processing to anxiety, eating
disorders, symptom perception and overall well-being. Recent experimental findings are
presented and reviewed in the context of hypotheses about the integration of peripheral
afferent signals with central cognitive operations and their role in shaping subjective ex-
perience in health and disease.
Quadt, Critchley, and Garfinkel provide an overview of the influence of internal bodily
states on emotion. By presenting a predictive coding account of interoceptive predictions
errors, they highlight the distinct ways in which deficits in interoceptive abilities may un-
derpin aberrant emotional processes characteristic of several clinical conditions. Khalsa
and Feinstein focus on the regulatory battle for control that ensues in the central nervous
system when there is a discrepancy between predicted and current bodily states (i.e. when
vi

viii Preface

somatic error signals are present). They argue that anxiety disorders are driven by so-
matic errors that chronically fail to be adaptively regulated. Herbert and Pollatos use a
predictive coding framework analogous to the one introduced by Quadt, Critchley, and
Garfinkel, and apply it to our understanding of eating behavior. They characterize an-
orexia and bulimia nervosa as a profound impairment of the “self,” with dysfunctional
interoception at its core. Yoris and colleagues explore the interoceptive dysfunctions
following neurological damage or neurodegeneration and emphasize the significance of
interoception to promote a hitherto missing synthesis of simultaneous autonomic, emo-
tional, and social cognition deficits in neurology. Van den Bergh, Zacharioudakis, and
Petersen focus on interoception and the perception of bodily sensations in the context
of symptom perception, and spell out the consequences of the highly variable relation-
ship between symptoms and physiological dysfunction for the disease model. Farb and
Logie focus on the appraisal of interoceptive signals and its consequences for subjective
well-​being. The modifying role of attention for habitual appraisals of interoceptive sig-
nals is presented, along with a novel breathing-​focused task for measuring interoceptive
awareness.
Part III Toward a philosophy of interoception:  subjectivity and experience approaches
interoception from a theoretical and philosophical perspective. Because until now the
field of interoception has been driven mainly by scientists rather than philosophers, this
part represents a highly novel departure for philosophy of mind and subjectivity, often
starting from a phenomenological point of view. The relation between subjective expe-
rience and physiological processes is an intricate and complicated one. Both the notion
of “arousal” in emotion and the experiential dimension of interoception more generally
stand in need of descriptive analysis and theoretical framing.
Colombetti and Harrison disentangle the physiology and the experience of “arousal” in
emotion and argue for the recognition of the multiple systems and pathways involved in
emotional arousal, including not only the autonomic nervous system but also pathways
of the endocrine and immune system, somatic sensations, and “background” bodily
feelings. Empirical studies in interoception have profound consequences for the way
the self or subjectivity is conceived of and conceptualized. The interoceptive dimension
of the embodied subject forces us to rethink phenomena such as body ownership and
the self, the relation between interoception and exteroception, and the coming about of
subjectivity itself. De Vignemont discusses the many ways the notion of interoception is
understood, and considers the contribution of interoception for the awareness of one’s
body as one’s own. She approaches body ownership in affective terms and as rooted in
self-​regulatory interoceptive feelings such as hunger and thirst. Corcoran and Hohwy
consider the limitations of homeostasis and favor a reconciliatory position in which ho-
meostasis and allostasis are conceived as equally vital but functionally distinct modes of
physiological control to account for the sophisticated regulatory dynamics observed in
complex organisms. De Preester focuses on a basis form of subjectivity and its origin in
interoceptive processes. She argues that the topographic representation of interoceptive
body states in the brain is unfit for explaining the coming about of subjectivity, and offers
 ix

Preface ix

directions for another model that takes into account the inherent characteristics of sub-
jectivity. Leder closes the volume, offering a phenomenology of inner-​body experience
and explaining how this experience is influenced by models drawn from the outer world.
In line with insights gained in the first and the second part, he points out the importance
of inner-​body experience for health and well-​being.
The different chapters included across the three parts are interrelated in various
ways, and the synergy between the chapters crosses the boundaries of the disciplines,
opening up opportunities for fruitful dialog between fields that otherwise remain too
often separate. For example, attention to subjectivity and subjective experience, to self-​
awareness and the experience of self, are common threads throughout the volume, to-
gether with the intricate role of emotions and their relation to interoception. Similarly,
several chapters are motivated by recent predictive coding accounts (Clark, 2013;
Friston, 2010) that have been extended from cognition and attention to affect (Critchley
& Garfinkel, 2017), bodily self-​awareness (Apps & Tsakiris, 2014; Seth, 2013), and
mental disorders (Feldman Barrett & Simmons, 2015). A  concern for human well-​
being and human health and suffering is also noticeable in all the chapters, and the
capacity to contribute to one’s own well-​being by paying attention to in-​depth bodily
signals is a recurrent theme. The chapters thus share a common concern for what it
means to experience oneself, for the crucial role of emotions, and for issues of health
and well-​being, discussed on the joint basis of our bodily existence and interoception,
resulting in a more than usual attention for the phenomenology of subjective experi-
ence in disciplines outside philosophy. Together, the chapters show that disciplinary
specialization is not a hindrance for dialogue but can result into mutual enrichment.
We hope that the scholarly research presented in this volume will further motivate
the much-​anticipated coming of age of interoceptive research in psychology, cognitive
neurosciences, and philosophy.
Helena De Preester
Manos Tsakiris

References
Adam, G. (1998). Visceral Perception, Understanding Internal Cognition. New York, NY: Springer.
Apps, M. A. J. and Tsakiris, M. (2014). The free-​energy self: A predictive coding account of self-​
recognition. Neuroscience & Biobehavioral Reviews, 41, 85–​97. <https://​doi.org/​10.1016/​
j.neubiorev.2013.01.029>
Bermúdez, J. L., Marcel, A. J., and Eilan, N. (eds) (1995). The Body and the Self. Cambridge,
MA: MIT Press.
Cameron, O. (2002). Visceral Sensory Neuroscience: Interoception. Oxford: Oxford University Press.
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science.
Behavioral and Brain Sciences, 36, 181–​204. <https://​doi.org/​10.1017/​S0140525X12000477>
Craig, A. D. B. (2009). How do you feel—​now? The anterior insula and human awareness. Nature
Reviews. Neuroscience, 10(1), 59–​70. <https://​doi.org/​10.1038/​nrn2555>
Critchley, H. D. and Garfinkel, S. N. (2017). Interoception and emotion. Current Opinion in Psychology,
17, 7–​14. doi: 10.1016/​j.copsyc.2017.
x

x Preface

Damasio, A. (1999). The Feeling of What Happens: Body, Emotion and the Making of Consciousness.
New York, NY: Harcourt.
Feldman Barrett, L. and Kyle Simmons, W. (2015). Interoceptive predictions in the brain. Nature
Publishing Group, 16, 419–​29. <https://​doi.org/​10.1038/​nrn3950>
Friston, K. (2010). The free-​energy principle: a unified brain theory? Nature Reviews. Neuroscience,
11(2), 127–​38. <https://​doi.org/​10.1038/​nrn2787>
Gallagher, S. and Cole, J. (1995). Body image and body schema in a deafferented subject. Journal of
Mind and Behavior, 16(4), 369–​89.
Gallese, V., Keysers, C., and Rizzolatti, G. (2004). A unifying view of the basis of social cognition.
Trends in Cognitive Sciences, 8, 396–​403.
Garfinkel, S. N., Seth, A. K., Barrett, A. B., Suzuki, K., and Critchley, H. D. (2015). Knowing your own
heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology,
104, 65–​74. <https://​doi.org/​10.1016/​j.biopsycho.2014.11.004>
Haggard, P. (2005). Conscious intention and motor cognition. Trends in Cognitive Sciences, 9(6), 290–​5.
<https://​doi.org/​10.1016/​j.tics.2005.04.012>
Kleckner, I. R., Wormwood, J. B., Simmons, W. K., Barrett, L. F., and Quigley, K. S. (2015).
Methodological recommendations for a heartbeat detection-​based measure of interoceptive
sensitivity. Psychophysiology, 52, 1432–​40. <https://​doi.org/​10.1111/​psyp.12503>
Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive
Sciences, 17(11), 565–​73. <https://​doi.org/​10.1016/​j.tics.2013.09.007>
Tsakiris, M. (2010). My body in the brain: A neurocognitive model of body-​ownership.
Neuropsychologia, 48(3), 703–​12. <https://​doi.org/​10.1016/​j.neuropsychologia.2009.09.034>
 xi

Acknowledgments

We would like to thank our colleague and friend Ophelia Deroy who introduced us to
Martin Baum, Senior Commissioning Editor for Psychology and Neuroscience at Oxford
University Press. We are grateful to Martin for supporting and selecting our proposal
for this volume among the many high-​quality proposals that OUP receives. We are also
grateful to April Peake for her assistance during the early stages of preparation and to
Charlotte Holloway for her assistance during the final production stages. We would also
like to acknowledge the NOMIS Foundation Distinguished Scientist Award to Manos
Tsakiris and a research grant from the University College Ghent to Helena De Preester
that provided the editors with the time and space of mind needed for the timely develop-
ment of this volume.
Of course, this volume wouldn’t exist without the excellent and timely contributions
made by all the authors who fully engaged with our vision for this volume and its multi-​
and cross-​disciplinary emphasis. We also thank the authors for acting as referees for
each other’s chapters along with Vivien Ainley, Laura Crucianelli, Chris Dijkerman,
Karl Friston, Philip Gerains, Rebekka Hufendiek, Peter Reynaert, Roy Salomon, Stefan
Sütterlin, and Dan Zahavi who generously devoted their precious time in reviewing sev-
eral chapters.
Last, but not least, the co-​editors would like to thank each other for a seamless mutually
enriching and supportive collaboration over the last two years.
xi
 xi

Contents

List of Contributors  xv

Part I Introduction
1 Interoception and the autonomic nervous system: Bottom-​up
meets top-​down  3
Gary G. Berntson, Peter J. Gianaros, and Manos Tsakiris

Part II  Mentalizing interoception: Advances and challenges

2 The body as first prior: Interoceptive predictive processing and the
primacy of self-​models  27
Micah Allen and Manos Tsakiris
3 Interoceptive signals, brain dynamics, and subjectivity  46
Mariana Babo-​Rebelo and Catherine Tallon-​Baudry
4 The embodiment of time: How interoception shapes the
perception of time  63
Marc Wittmann and Karin Meissner
5 The neurobiology of gut feelings  80
Qasim Aziz and James K. Ruffle
6 The cutaneous borders of interoception: Active and social inference of
pain and pleasure on the skin  102
Mariana von Mohr and Aikaterini Fotopoulou

Part III  From health to disease: Interoception

in physical and mental health
7 Interoception and emotion: Shared mechanisms and
clinical implications  123
Lisa Quadt, Hugo D. Critchley, and Sarah N. Garfinkel
8 The somatic error hypothesis of anxiety  144
Sahib S. Khalsa and Justin S. Feinstein
9 The relevance of interoception for eating behavior
and eating disorders  165
Beate M. Herbert and Olga Pollatos
xvi

xiv Contents

10 Cardiac interoception in neurological conditions and its relevance for

dimensional approaches  187
Adrián Yoris, Adolfo M. García, Paula Celeste Salamone, Lucas Sedeño, Indira
García-​Cordero, and Agustín Ibáñez
11 Interoception, categorization, and symptom perception  212
Omer Van den Bergh, Nadia Zacharioudakis, and Sibylle Petersen
12 Interoceptive appraisal and mental health  227
Norman A. S. Farb and Kyle Logie

Part IV  Toward a philosophy of interoception:

Subjectivity and experience
13 From physiology to experience: Enriching existing conceptions of
“arousal” in affective science  245
Giovanna Colombetti and Neil Harrison
14 Was Descartes right after all? An affective background for bodily
awareness  259
Frédérique de Vignemont
15 Allostasis, interoception, and the free energy principle:
Feeling our way forward  272
Andrew W. Corcoran and Jakob Hohwy
16 Subjectivity as a sentient perspective and the role of interoception  293
Helena De Preester
17 Inside insights: A phenomenology of interoception  307
Drew Leder

Author Index  323

Subject Index  333
 xv

Contributors

Micah Allen Helena De Preester

Institute of Cognitive Neuroscience, School of Arts, University College Ghent,
University College London, United Belgium
Kingdom Department of Philosophy and Moral
Wellcome Centre for Human Sciences, Faculty of Arts and Philosophy,
Neuroimaging, University College Ghent University, Belgium
London, United Kingdom
Frédérique de Vignemont
Qasim Aziz Institut Jean Nicod, ENS-​CNRS-​EHESS,
Barts and the London School of Medicine Department of cognitive studies, École
& Dentistry, Queen Mary University of Normale Supérieure, PSL, Paris, France
London, United Kingdom
Norman A. S. Farb
Mariana Babo-​Rebelo Department of Psychology, University of
Laboratoire de Neurosciences Cognitives, Toronto Mississauga, Mississauga, Canada
Département d’Etudes Cognitives, Ecole Justin S. Feinstein
Normale Supérieure (ENS), PSL Research Laureate Institute for Brain Research,
University, INSERM, Paris, France Tulsa, Oklahoma, United States
Gary G. Berntson Oxley College of Health Sciences,
Department of Psychology, Ohio State University of Tulsa, Tulsa, Oklahoma,
University, United States United States
Giovanna Colombetti Aikaterini Fotopoulou
Department of Sociology, Philosophy Research Department of Clinical,
and Anthropology, University of Exeter, Educational and Health Psychology,
United Kingdom Faculty of Brain Sciences, University
College London, United Kingdom
Andrew W. Corcoran
Cognition & Philosophy Laboratory, Adolfo M. García
Department of Philosophy, Faculty of Laboratory of Experimental Psychology
Arts, Monash University, Melbourne, and Neuroscience (LPEN), Institute of
Australia Cognitive and Translational Neuroscience
(INCYT), INECO Foundation, Favaloro
Hugo D. Critchley
University, Buenos Aires, Argentina
Department of Neuroscience, Brighton
National Scientific and Technical Research
and Sussex Medical School (BSMS),
Council (CONICET), Buenos Aires,
University of Sussex, Brighton, United
Argentina
Kingdom
Faculty of Education, National University
of Cuyo (UNCuyo), Mendoza, Argentina
xvi

xvi Contributors

Indira García-​Cordero Agustín Ibáñez

Laboratory of Experimental Psychology Laboratory of Experimental Psychology
and Neuroscience (LPEN), Institute of and Neuroscience (LPEN), Institute of
Cognitive and Translational Neuroscience Cognitive and Translational Neuroscience
(INCYT), INECO Foundation, Favaloro (INCYT), INECO Foundation, Favaloro
University, Buenos Aires, Argentina University, Buenos Aires, Argentina
National Scientific and Technical Research National Scientific and Technical Research
Council (CONICET), Buenos Aires, Council (CONICET), Buenos Aires,
Argentina Argentina
Universidad Autónoma del Caribe,
Sarah N. Garfinkel
Barranquilla, Colombia
Psychiatry, Department of Neuroscience,
Center for Social and Cognitive
Brighton and Sussex Medical School
Neuroscience (CSCN), School of
(BSMS), University of Sussex, Brighton,
Psychology, Universidad Adolfo Ibañez,
United Kingdom
Santiago, Chile
Sackler Centre for Consciousness Science,
Centre of Excellence in Cognition and its
University of Sussex, Brighton, United
Disorders, Australian Research Council
Kingdom
(ACR), Sydney, Australia
Sussex Partnership NHS Foundation
Trust, United Kingdom Sahib S. Khalsa
Laureate Institute for Brain Research,
Peter J. Gianaros
Tulsa, Oklahoma, United States
Department of Psychology, University of
Oxley College of Health Sciences,
Pittsburgh, United States
University of Tulsa, Tulsa, Oklahoma,
Neil Harrison United States
Department of Neuroscience, Brighton
Drew Leder
& Sussex Medical School, University of
Department of Philosophy, Loyola
Sussex, United Kingdom
University Maryland, Baltimore,
Sussex Partnership NHS Foundation
United States
Trust, Swandean, United Kingdom
Kyle Logie
Beate M. Herbert
Department of Psychology, University of
Clinical Psychology & Psychotherapy,
Toronto Mississauga, Mississauga, Canada
Department of Psychology, Eberhard
Karls University of Tübingen, Germany Karin Meissner
Institute of Medical Psychology, Ludwig-​
Jakob Hohwy
Maximilian University, Munich, Germany
Cognition & Philosophy Laboratory,
Division Integrative Health Promotion,
Department of Philosophy, Faculty of
University of Applied Sciences, Coburg,
Arts, Monash University, Melbourne,
Germany
Australia
Sibylle Petersen
Health Psychology, KU Leuven–​University
of Leuven, Leuven, Belgium
 xvi

Contributors xvii

Olga Pollatos Manos Tsakiris

Clinical & Health Psychology, Institute Lab of Action & Body, Department of
of Psychology and Education, Ulm Psychology, Royal Holloway University of
University, Germany London, United Kingdom
The Warburg Institute, School of
Lisa Quadt
Advanced Study, University of London,
Psychiatry, Department of Neuroscience,
United Kingdom
Brighton and Sussex Medical School
(BSMS), University of Sussex, Brighton, Omer Van den Bergh
United Kingdom Health Psychology, KU Leuven–​University
of Leuven, Leuven, Belgium
James K. Ruffle
Centre for Neuroscience and Trauma, Mariana von Mohr
Blizard Institute, Wingate Institute of Research Department of Clinical,
Neurogastroenterology, Barts and the Educational and Health Psychology,
London School of Medicine & Dentistry, Faculty of Brain Sciences, University
Queen Mary University of London, College London, United Kingdom
London, United Kingdom
Marc Wittmann
Paula Celeste Salamone Institute for Frontier Areas of Psychology
Laboratory of Experimental Psychology and Mental Health, Freiburg, Germany
and Neuroscience (LPEN), Institute of Institute of Medical Psychology, Ludwig-​
Cognitive and Translational Neuroscience Maximilian University, Munich, Germany
(INCYT), INECO Foundation, Favaloro Adrián Yoris
University, Buenos Aires, Argentina
Laboratory of Experimental Psychology
National Scientific and Technical Research and Neuroscience (LPEN), Institute of
Council (CONICET), Buenos Aires,
Cognitive and Translational Neuroscience
Argentina
(INCYT), INECO Foundation, Favaloro
Lucas Sedeño University, Buenos Aires, Argentina
Laboratory of Experimental Psychology National Scientific and Technical Research
and Neuroscience (LPEN), Institute of Council (CONICET), Buenos Aires,
Cognitive and Translational Neuroscience Argentina
(INCYT), INECO Foundation, Favaloro
Nadia Zacharioudakis
University, Buenos Aires, Argentina Health Psychology, KU Leuven–​University
National Scientific and Technical Research of Leuven, Leuven, Belgium
Council (CONICET), Buenos Aires,
Argentina
Catherine Tallon-​Baudry
Laboratoire de Neurosciences Cognitives,
Département d’Etudes Cognitives, Ecole
Normale Supérieure (ENS), PSL Research
University, INSERM, Paris, France
xvi
 1

PART I

Introduction
2
 3

Chapter 1

Interoception and the autonomic

nervous system: Bottom-​up meets
top-​down
Gary G. Berntson, Peter J. Gianaros,
and Manos Tsakiris

1.1 Introduction
Interoception is a multidimensional construct, broadly encompassing the processing of
afferent (sensory) information arising from internal organs, tissues, and cells of the body.
This afference contributes to the regulation of homeostatic reflexes, and, as we will see in
this chapter and throughout this volume, more broadly to the generation and regulation
of cognitive and emotional behaviors.
Interoception can be encompassed by the broader construct of bodily afference.
The latter includes both visceral afference and somatic afference. We use the term vis-
ceral afference to refer to the processing of internal sensory information derived from
interoceptors that are located in the organs and tissues of the main cavities of the body
(i.e. the viscera), as well as from olfactory and gustatory receptors, all being generally
associated with the limbic system and the autonomic nervous system. We use the term
somatic afference to refer to the processing of sensory information (e.g. proprioceptive
input and tactile sensitivity) derived from components of the somatic system (e.g. mus-
cles, joints, skin). This distinction between somatic and visceral afference does not imply a
complete independence. Indeed, in many cases, there is an integration of multiple modes
of bodily or somatosensory information derived, for example, from metabolic changes
in active muscle tissue. Hence, the term somatovisceral afference is more appropri-
ately applied to integrated, multimodal, or otherwise nonspecific internal sensory input
from within the body (e.g. see Yates & Stocker, 1998). In these regards, the construct of
interoception itself is more specifically aligned with that of visceral afference, referring to
the processing of sensory information from interoceptors that are located within the vis-
ceral organs and from interoceptors located elsewhere in the body that provide for local
energy needs. Thus, in contrast to exteroceptors, interoceptors are tuned to sense internal
events (Cameron, 2002).
The so-​called general visceral afferents (GVAs) that relay internal sensory information
from interoceptors are carried by several cranial nerves, the most notable being the vagus
4

4 Interoception and the autonomic nervous system

nerve. These afferents carry information (e.g. pressor receptor activity from blood vessels)
originating from the gut and the viscera more generally (i.e. organs and tissues located
in the thoracic, abdominal, and pelvic cavities, as well as blood vessels and muscles).
By comparison, special visceral afferents (SVAs) convey gustatory senses (i.e. taste) and
olfaction (i.e. smell and pheromonal senses). Although the SVAs detect environmental
stimuli, they do so by virtue of those stimuli impinging on the internal bodily environ-
ment. Hence, they differ from exteroceptors; for example, conveying information related
to touch or audition. Furthermore, the visceral senses have common central projections
to cell groups in the brainstem, including the nucleus tractus solitarius (NTS), midbrain,
and thalamus, that are distinct from those of somatic exteroceptors, and they link an-
atomically and functionally with a distinct set of central neural systems and processes
(Craig, 2014; Saper, 2002). Moreover, they share biochemical markers in common with
GVAs and with autonomic neurons (see Squire et al., 2012). There are other classes of sen-
sory systems, such as proprioceptors, that sense joint position, and vestibuloceptors, that
sense body orientation in gravitational space. These might be considered interoceptors
as they are internal to the body. Yet, they are closely linked with somatic motor systems
anatomically and functionally, and they have biochemical markers more in concert with
somatic motor systems. Hence, they are sometimes considered within the unique class of
proprioceptors, or otherwise just included within the general class of exteroceptors.
What is important to consider is that both exteroceptive and interoceptive informa-
tion can powerfully influence cognitive and emotional processes, and, importantly, vice
versa. Moreover, as will be developed later in this chapter, visceral afferents carrying in-
teroceptive information have a constitutional link with central neural systems underlying
cognitive and emotional processes, and they thus impact these behavioral processes (e.g.
Tsakiris & Critchley, 2016a; see also the entire Theme issue, Tsakiris & Critchley, 2016b).
This is clearly apparent in the positive and negative (e.g. disgust) hedonic effects of tastes
and smells. As detailed in the remaining chapters of this volume, it is thus doubtful that
interoception can be meaningfully parsed or dissociated from cognitive, emotional, and
behavioral processes. Indeed, a recent meta-​analysis of fMRI (functional MRI) studies
revealed considerable overlap in systems co-​activated by interoceptive signaling, emo-
tional regulation and low-​level social cognition, and convergent results were found for the
effects of lesions (Adolfi et al., 2017).

1.2  Historical perspectives

Claude Bernard is generally credited with developing the concept of the “fixity” or rel-
ative stability of the internal fluid matrix (milieu intérieur) as a necessary condition for
what he termed the free and independent life (Bernard, 1878). Walter Cannon, chair
of the Department of Physiology at Harvard around the turn of the twentieth century,
further elaborated on this concept and coined the term homeostasis (Cannon, 1932/​
1939). He believed that the autonomic nervous system (ANS) plays an important part
in maintaining this homeostatic stability. Although Cannon recognized the importance
of visceral afferents in homeostatic reflexes, the predominant view of the ANS was as an
 5

Historical perspectives 5

efferent, lower-​level, reflexive motor system. Cannon further asserted that the ANS is
termed “ ‘autonomic’ because it acts automatically, without direction from the cerebral
cortex” (Cannon, 1939, p.  250). This misconception was perhaps understandable, as it
would be another half century before the existence of direct, monosynaptic projections
from cortical and other telencephalic areas to lower brainstem autonomic source nuclei
and reflex substrates would be documented (e.g. Barbas et al., 2003; Cechetto & Saper,
1990; Dum, Levinthal, & Strick, 2016; Neafsey, 1990).
In retrospect, this misconception is also somewhat surprising in view of the work of
Cannon’s friend, contemporary, and Nobel Laureate, Ivan Pavlov (see Figure 1.1). Here, it
is often underappreciated that much of Pavlov’s work on learning centered on the modifi-
cation of an autonomic and homeostatic reflex that involves the processing of interocep-
tive information in preparation for digestion—​the cephalic vagal reflex, or the cephalic
phase insulin release to a stimulus which had previously been paired with food. Indeed,
this work contributed to a foundation for understanding how interoceptive phenomena
can be powerfully related to learning and other processes instantiated in higher brain
systems that can jointly influence visceral control, including systems within the cere-
bral cortex. This includes experience-​based acquired responses that can both modulate

Figure 1.1  Walter B. Cannon with Ivan Pavlov at the 1929 International Physiological Congress.
Photograph reproduced with the acquiescence of the curator (Harvard Medical Library in the Francis A. Countway
Library of Medicine).
6

6 Interoception and the autonomic nervous system

ongoing visceral processes as well as anticipate and prospectively guide adaptive auto-
nomic, cognitive, and emotional responses (Cameron, 2002; Dworkin, 1993).
Well before the existence of long descending pathways linking brain and viscera had been
established, functional studies revealed autonomic representations in multiple higher-​
level diencephalic and telencephalic areas of the forebrain. The 1949 Nobel Laureate,
Walter Hess, for example, had reported striking autonomic responses elicited by stim-
ulation of telencephalic (e.g. the amygdala and septal area) and diencephalic structures,
including the hypothalamus, which Hess considered the head-​ganglion of the autonomic
nervous system (Hess, 1954; see also Ranson, Kabat, & Magoun, 1935). The Canadian
neurosurgeon, Wilder Penfield, also reported autonomic responses to cortical stimulation
in conscious human patients (Penfield & Jasper, 1954). At this point, we recognize a broad
central autonomic network comprising a number of forebrain areas, including the insular
cortex, cingulate cortex, medial prefrontal cortex, amygdala, and hippocampus, as well as
caudal cell groups in the midbrain periaqueductal gray, pons, cerebellum, and medulla
(Benarroch, 1993; Cechetto & Saper, 1990; Critchley, 2005; Dampney, 2015; Loewy, 1991;
Neafsey, 1990; Saper, 2002; Shoemaker & Goswami, 2015). This network highly overlaps
and interacts with systems implicated in complex cognitive, skeletomotor, and affective
processes (Annoni et al., 2003; Critchley, 2005; Myers, 2017; Saper, 2002).
Another legacy from the Cannon era that did not survive the test of time is the view that
the ANS is primarily an efferent motor system, with minimal visceral afferents necessary
to support homeostatic reflex functions. In his influential book, The Wisdom of the Body,
Cannon stated:  “The nerve fibers of the autonomic nervous system, which are mostly
efferent, pass out of the central nervous system” (Cannon, 1939, p. 252). Similarly, John
Newton Langley, who coined the term autonomic nervous system1 (Langley, 1898) viewed
the ANS largely as a motor system, although he recognized that one might “consider as
afferent autonomic fibers those which give rise to reflexes in autonomic tissues . . . [but
are] . . . incapable of directly giving rise to sensation” (Langley, 1903, p. 2). As early as the
mid-​1930s, however, there were quantitative studies demonstrating that the majority of
fibers in the vagus are sensory (Foley & DuBois, 1937; Heinbecker & O’Leary, 1933). This
is consistent with contemporary estimates that 70–​90% of vagal fibers, about 2–​20% of
fibers in the splanchnic (sympathetic) nerves, and about 2% in spinal nerves are visceral
afferents2 (Berthoud & Neuhuber, 2000; Cameron, 2002; Jänig & Morrison, 1986).
Historically, there were also notable conceptual challenges to the notion that the ANS
is primarily a motor system. In contrast to the view that emotions precede and trigger
bodily reactions, William James (1884) proposed that exciting events induce bodily

“We propose the term ‘autonomic nervous system,’ for the sympathetic system and the allied nervous
1

system of the cranial and sacral nerves, and for the local nervous system of the gut” (p. 270).
Although some of these afferents run with parasympathetic and sympathetic efferents, it is not appro-
2

priate to consider them “parasympathetic” and “sympathetic” afferents (Freire-​Maia & Azevedo,1990;
Jänig & Häbler, 1995). They are general visceral afferents that are not strictly coupled to an autonomic
branch.
 7

A case report 7

changes (including autonomic responses) and that our subsequent feeling of these
changes constitutes the emotion. Shortly thereafter, Carl Lange (1885) independently
proposed a vascular theory of emotion, which held that vasomotor responses are the pri-
mary effects of affectations, and subjective sensations of emotion arise secondarily. Both
of these perspectives converged into a view of visceral afference as fundamental for the
generation and experience of emotion.
This view came under severe assault from two notable figures: Walter Cannon (1927,
1931), often considered the “father of the autonomic nervous system” and Sir Charles
Sherrington (1900)—​a notable turn-​of-​the-​century physiologist and a recipient of the
Nobel Prize in Physiology or Medicine (1932) for “discoveries regarding the functions of
neurons.” At least the strong form of the James–​Lange concept (emotions as the mere per-
ception of visceral feedback) was largely discredited at that time. There were a number
of arguments against the James–​Lange perspective. Cannon, for example, argued that the
viscera have few afferents and are relatively insensate. However, it is now recognized that
visceral afferents in fact outnumber efferents. Additionally, it was argued that (a) visceral
responses are too slow to underlie emotion; (b) similar visceral changes may occur across
different emotions and even non-​emotional states; (c) inducing autonomic responses does
not necessarily invoke emotions; and (d) that autonomic denervations of various types do
not prevent emotional reactions. None of those are particularly telling arguments unless
one wants to assert an identity relationship between visceral afference and emotion. James
(1884), in fact, viewed emotions as being multiply determined, and to include cognitive
contributions. He stipulated in his 1884 article that the only emotions that he will “con-
sider here are those that have a distinct bodily expression” (p. 189)—​the so-​called coarser
emotions. It is well established that there are multi-​level hierarchical and heterarchical
representations in neurobehavioral systems and central autonomic networks (see Berntson,
Cacioppo, & Bosch, 2016; Norman, Berntson, & Cacioppo, 2014), and there are multiple
determinants of affective processes. What will become apparent through the chapters
of this volume is that there are powerful interactions between cognitive and emotional
processes, somatic and autonomic outflows, and interoceptive feedback. Consequently, the
effects of interoceptive feedback would not be expected to be invariant but to show notable
brain-​state and context dependencies (e.g. see Cacioppo, Berntson, & Klein, 1992).

1.3  A case report
Visceral afference can powerfully modulate cognitive and emotional processes, as illus-
trated by the following case report on MM (personal communication). MM is a graphic
artist and videographer who was working on a documentary about a historical kidnapping
and murder. She had extensively researched and documented the story and had located
and interviewed most of the characters involved (except for the perpetrator, who had killed
himself). The story was ready for filming (January 2007), but, alas, filming never happened:
I handed the story back to the producer when it became clear that I could not work the script (kept
repeating the same scenes), as I was able to keep it all in my head for only about 3 pages—​few
8

8 Interoception and the autonomic nervous system

minutes, and no amount of colour coordinated storylines and Post-​It notes were going to save me
when I was not able to make a simple decision on spot (calculating what days we had available for
shooting whom . . . etc.).

What led to this transformation (February 2007)? It was endoscopic thoracic sympathec-
tomy (ETS), the surgical destruction or disabling of the upper spinal sympathetic (auto-
nomic) nerve trunk, for hyperhidrosis (excessive sweating). Thereafter, MM’s life (she was
39 years old at the time) was dramatically changed3:
It is my experience that following this surgery there is a shift in personality and how emotions are
experienced. It is, however, not only emotional blunting but also an impaired impulse control and
disinhibition (as if a grown-​up brain has been replaced by a primitive, and at times manic brain,
that affects higher functioning). I am not sure how to describe it really . . . There is an indifference
and striking lack of fear . . . I witnessed within myself once I got into my car and started driving
around, or in general danger situations any urbanite encounters. My emotions are blunted, and
there is an unsettling deadness and indifference towards my prior life and aspirations, goals. This
indifference and emotional blunting was present as soon as I woke up from the surgery and has
not left me since.
 . . . 
In general, the procedure led to a personality change, in some aspects subtle, in others a profound
shift that I find exceedingly difficult to accept—​a kind of physiological expression of how I was
feeling, zombie-​like.
 . . . 
I was described by one (video) critic as a human seismograph, recording the finest shifts in mood/​
tone . . . (now) I have problems in social settings, where I generally might appear antisocial. I force
myself to ask questions and engage in “banter”, but more often I  forget. I  would say that it has
changed how I relate to people: I do not relate.

Tragically, this outcome was not unique to MM. There is a considerable literature
documenting a range of post-​ sympathectomy complications including cognitive
deficits, altered mood, emotional blunting, fatigue, and neuropathic complications (e.g.
Furlan, Mailis, & Papagapious, 2000; Goldstein, 2012; Mailis & Furlan, 2003). Indeed,
a support group, the Sympathetic Association (FfSo), was formed in Karlstad, Sweden,
by people who found themselves disabled by serious side effects of sympathectomies
(<http://​home.swipnet.se/​sympatiska/​index3.htm>). We will return to the case of MM
in section 1.4.2.

1.4  Central visceral pathways and the visceral cortex

An important integrative site in the forebrain for visceral afference is the insular cortex,
which could be considered a primary visceral cortical site. The insula, in turn, is highly
interconnected with cortical and subcortical areas involved in cognition, emotion, and

There was no pre-​surgical history of psychopathology. The patient elected the procedure to reduce ex-
3

cessive sweating, and to some extent, this was achieved. However, as is common with ETS (Furlan et al.,
2000), she did experience periodic “compensatory” sweating.
 9

Central visceral pathways and the visceral cortex 9

motivation, including the prefrontal cortex, the cingulate cortex, and the amygdala (Allen
et  al., 1991; Augustine, 1996; Nieuwenhuys, 2012; Oppenheimer & Cechetto, 2016).
The insula receives input from all visceral afference and, as will be seen throughout this
volume, contributes to the integration of this afference with neurobehavioral processes
(Tsakiris & Critchley, 2016a), and anomalies in insular function are associated with a
wide range of cognitive, emotional, and behavioral disturbances (Gasquoine, 2014).

1.4.1  Special
visceral afferents: The chemical senses—​Olfaction
and gustation
Olfaction is a special visceral sense closely linked with both positive and negative
hedonics. This is especially true in lower animals where it plays a central role in guiding
behavior. The olfactory system is closely linked with a medial central brain network that
was historically referred to as the rhinencephalon (nose brain). Paul Broca (1878) re-
ferred to the medial central components of the brain as the great limbic lobe (le grand
lobe limbique) because they arch around the central encephalon (“limbique” in French
translates as “hoop” or “curve”). Papez (1937) proposed that limbic areas and associated
structures are an important central network in emotion (often referred to as Papez cir-
cuit). This concept was further developed by Paul MacLean (1954) who coined the term
“limbic system” and viewed this system as an evolutionary heritage (the paleomammalian
brain) that regulates emotion, motivation, and survival-​related behaviors, as well as links
these phenomena with vulnerability to chronic health conditions (e.g. hypertension,
asthma). Olfactory afferents play an important role in emotion, motivation, and survival-​
related behavioral processes. Although the olfactory system more directly projects to a
number of cortical areas, olfactory information is also relayed via the thalamus and other
cortical areas to the insular cortex (which itself is often considered to be a part of the
limbic system). Odors can modulate mood, cognition, and behavior, and many of these
effects appear to be mediated by the insula (for reviews see Miranda, 2012; Saive, Royet,
& Plailly, 2014; Soudry et al., 2011).
The primary gustatory cortex lies in the anterior insula. Gustatory afferents (cranial
nerves VII, IX, and X) terminate in a medullary nucleus, the NTS, and then are relayed via
the midbrain parabrachial nucleus to the ventroposteromedial thalamus, which issues di-
rect projections to the anterior insula (Saper, 2002). A similar functional pattern emerges
in the literature to that of olfaction (Rolls, 2015). There are potent insular contributions to
the processing of taste hedonics and attentional and memorial processes associated with
taste, and insular cortex abnormalities are associated with disturbances in these processes
(Frank, Kullmann, & Veit, 2013; Maffei, Haley, & Fontanini, 2012).

1.4.2  General visceral afferents

As is the case with gustatory afferents, all GVAs in cranial nerves project to the NTS
in the brainstem, to the parabrachial nucleus in the midbrain, and then via the
ventroposteromedial (VPM) nucleus of the thalamus to the insular cortex. In his classic
studies, Wilder Penfield reported that electrical stimulation of the insula induced a
10

10 Interoception and the autonomic nervous system

variety of visceral sensory experiences (Penfield & Jasper, 1954; Penfield & Faulk, 1955).
In addition to cranial nerves, GVAs carrying nociceptive, temperature, and chemosen-
sory information from the body enter the spinal cord via dorsal spinal roots and termi-
nate in the dorsal horn (especially in lamina I). Until around the turn of the twenty-​first
century, the general belief was that small-​diameter nociceptive (i.e. sensory information
about tissue damage)/​temperature afferents were part of the somatosensory system and
were ultimately relayed to the somatosensory cortex in the parietal lobe. Indeed, this view
persists. In their 2016 textbook on neuroscience, Bear, Connors, and Paradiso assert that
the “spinothalamic pathway is the major route by which pain and temperature informa-
tion ascend to the cerebral cortex” (2016, p. 444). In fact, it is now well established that the
small diameter fibers carrying nociceptive, temperature, and chemical senses project from
the VPM not to the somatosensory cortex but to the insula (Craig, 2014; Saper, 2002).
This accounts for the fact that in Wilder Penfield’s studies, patients never reported pain on
stimulation of the somatosensory cortex (Penfield & Jasper, 1954; Penfield & Faulk; 1955).
Moreover, surgeons do not extirpate the somatosensory cortex for pain syndromes. In
contrast, however, Mazzola and colleagues (2012) report induced pain with stimulation
of the insular cortex, and painful “somatosensory” seizures appear to arise not from the
somatosensory cortex but from the opercular-​insular cortex (Montavont et al., 2015).
This general visceral afference, and the top-​down and bottom-​up integration of insular
cortical systems, underlie the cognitive-​emotional processes that reflect the broad inte-
grative contributions of the insula (Tsakiris & Critchley, 2016a). Insula lesions, for ex-
ample, result in diminished emotional arousal to affective pictures, and a reduced ability
to even recognize the affective picture content (Berntson et al., 2011). Although the lit-
erature on insular involvement in emotion and emotional processing is quite consistent
(Uddin, Nomi, & Hébert-​Seropian, 2017), there appears to be some diversity in the effects
of insular lesions.4 In addition to its role in emotion and motivation, the insula appears
to play a pivotal role in the sense of self, agency, and indeed, consciousness (Craig, 2014;
Strigo & Craig, 2016; Tsakiris & Critchley, 2016a, b; see also Chapters 2, 3, and 16 in the
present volume). Thus, insula activation is correlated with the sense of body ownership
and agency (Farrer, Franck, & Georgieff, 2003; Tsakiris et al., 2007). In accord, lesions of
the insula can lead to a disturbed sense of body ownership, including somatoparaphrenia
or the denial of body ownership (Cogliano et al., 2012; Gandola et al., 2012; Karnath &
Baier, 2010; Moro et al., 2016).
These findings and further results addressed in the present volume indicate that visceral
afferent input to the insula appears to be critical in cognitive and emotional processes.

Garcia and colleagues (2016) report minimal cognitive or socio-​emotional deficits in a single case
4

report after extensive vascular lesion damage, including the insular cortex. The authors, however, em-
phasize how unusual this case was as there were also minimal disturbances in sensorimotor and other
functions. The literature on disgust, especially, is quite variable, but Uddin and colleagues (2017) report
consistent socio-​emotional deficits with insular lesions, but they also emphasize the considerable func-
tional heterogeneity in this brain region.
 1

Current applications and implications 11

However, one may see similar deficits with disrupted visceral afference in the absence
of frank insular impairments. Patients with pure autonomic failure, and the associated
blunting of autonomic activity and visceral afference, have been reported to show deficits
in cognitive processing, empathy and emotional reactivity (Chauhan, Mathias, & Critchley,
2008; Critchley, Mathias, & Dolan, 2001; Tsakiris et al., 2006). In this regard, returning to
our case study, although MM did not have a direct insular insult, the surgical sympath-
ectomy would have as a necessary consequence a diminution of both sympathetic and
parasympathetic visceral afference, and this may have contributed to the cognitive and
emotional sequelae she experienced. Critchley and colleagues (2001, p. 207) asserted that
“body state changes, particularly those mediated by the autonomic nervous system, are
crucial to the ongoing emotional experience of emotion,” and Goldstein (2012) reported
that partial cardiac denervation was associated with fatigue, altered mood, blunted emo-
tion, and decreased ability to concentrate. The findings that meditation can increase in-
sular activity, connectivity, gray and white matter volume raise a question as to a potential
therapeutic strategy in visceral denervation syndromes (Gotink et al., 2016; Hernandez
et al., 2016; Laneri et al., 2016).

1.5  Current applications and implications

The central autonomic network (CAN) and neurobehavioral substrates more generally
become re-​represented and elaborated with the evolutionary development of higher-​
level neural systems (Jackson, 1884). This has significant implications for the functional
understanding of the autonomic nervous system, visceral afference, and the multi-​level
representations in the CAN. To elaborate, lower-​level autonomic reflexes, such as those
studied by Cannon, are relatively hardwired and highly regulated by local visceral afference
contributing to the feedback regulation of homeostasis. The baroreceptor reflexes, for
example, are visceral homeostatic control loops that constrain short-​term variations in
blood pressure via rapid autonomic adjustments. The afferent limb of the baroreflexes
encompasses interoceptors positioned in the heart and great arteries. Increases in blood
pressure cause a distortion of their free nerve endings, leading to an increase in afferent
firing and associated afferent input to the NTS. Via relays in the rostral and caudal ven-
trolateral medulla, this NTS afferent input results in a subsequent reflexive reduction
in sympathetic and an increase in parasympathetic control of the heart and cardiovas-
cular system, culminating in an associated compensatory reduction in blood pressure.
The sensitivity of the homeostatic baroreceptor heart rate reflex can be quantified by the
slope of the function relating heart rate to blood pressure. This slope reflects in part the
servocontrol of blood pressure by heart rate decreases with increasing blood pressure
(decreasing cardiac output) and heart rate increases with decreasing blood pressure
(increasing cardiac output). It was Cannon’s student, Philip Bard, however, who noted
that the slope of the baroreflex function could be decreased (flattened) by a typical labora-
tory stressor (reflecting a decrease in sensitivity of the baroreflex). That is to say, stressors
“inhibit” the homeostatic control over blood pressure by the baroreflex, which presum-
ably enables heart rate and blood pressure to rise simultaneously and rapidly to provide
12

12 Interoception and the autonomic nervous system

hemodynamic and metabolic support for adaptive action or stressor coping. Although
stressor-​evoked effects on the baroreflex have been widely seen across species, such effects
raise a number of basic questions insofar as they reflect “anti-​homeostatic” actions that
appear to be implemented by higher levels of the CAN. That is, while homeostasis is an
important contribution of the ANS, ANS effects are not always homeostatic across behav-
ioral states (see Berntson, Cacioppo, & Bosch, 2016). Indeed, it may be maladaptive to
maintain a “fixity” of the internal milieu by lower reflex substrates of the CAN. In this way
and in the face of adaptive challenges, higher levels of the CAN may modulate or “reset”
lower substrates for reflex control to implement contextually appropriate, anticipatory, or
otherwise “adaptive” increases in both blood pressure and heart rate (Dampney, 2017).
Sterling and Eyer (1988) and later Schulkin (2003) introduced the concept of allostasis
to reflect the fact that homeostasis is not necessarily static, but can assume different regu-
latory levels (setpoints), to adapt to survival challenges. An example of this is fever, which
unlike the “anti-​homeostatic” baroreflex effects of stress, represents a true adoption of
a higher regulatory setpoint, which is monitored and actively defended both physio-
logically and behaviorally. Bruce McEwen (2012) subsequently introduced the concept
of allostatic load to reflect the fact that while short-​term allostatic adjustments may be
adaptive, sustained, long-​term allostatic adjustments may have cumulative and delete-
rious health consequences. However, deviations from homeostasis may not always en-
tail simply an altered setpoint level. Berntson, Cacioppo, and Bosch (2016) advanced the
concept of heterodynamic regulation in which higher level CAN and neurobehavioral
substrates, integrating somatovisceral afference, cognitive and emotional processing, can
dynamically regulate autonomic outflows and therefore somatovisceral afference in a flex-
ible, dynamic fashion to achieve more optimal adaptive outcomes that are appropriate to
given behavioral contexts.
These considerations suggest that classical concepts of the autonomic nervous system,
which focus on autonomic reflexes and homeostasis, are inadequate for a full under-
standing of the contributions of the ANS and its afference to neurobehavioral processes.
An illustration of the latter point is evident in contemporary perspectives on the role
of interoceptive processes in physiological stress reactivity and recovery. More precisely,
physiological stress reactivity and recovery have received widespread and long-​standing
attention because of their presumptive relationships with aspects of physical and mental
health across the lifespan (Cohen , Gianaros, & Manuck, 2016). For example, people with
phenotypes to exhibit exaggerated and prolonged rises in heart rate and blood pressure
that are mediated by the autonomic nervous system are at elevated risk for hypertension,
stroke, myocardial infarction, and early death (Ginty, Kraynak, Fisher,et al., 2017).
The central substrates for such patterns of stressor-​evoked physiological reactivity and
recovery have been studied for over a century in human and non-​human animal studies.
Notwithstanding, a historically neglected dimension of these substrates is how they are
influenced by stressor-​evoked interoceptive information encoded in peripheral physi-
ology (Gianaros & Wager, 2015). To elaborate, stressors are thought to engage higher
neural substrates of the CAN, including the insula and anterior cingulate cortex (ACC),
 13

Current applications and implications 13

which may appraise stressors and in turn issue descending visceral motor commands
for rapid autonomic adjustments to cardiovascular physiology. These centrally deter-
mined and stressor-​evoked adjustments may entail simultaneous rises in blood pressure
and heart rate, with accompanying modifications to the baroreflex, to provide metabolic
support for behavioral action and stressor coping (Gianaros et  al., 2012). This central
linkage of behavior with metabolically supportive changes in cardiovascular physiology is
exemplified in the cardiac-​somatic coupling hypothesis of Obrist (1981) and the “central
command hypothesis” within the field of exercise physiology (Fisher et al., 2015). More
recent perspectives on this linkage now emphasize that central substrates for periph-
eral stress reactivity most likely issue visceral motor commands in a predictive fashion,
providing metabolic support for behavior that is anticipated in the future (Ginty et al.,
2017; Gianaros & Jennings, 2018). Moreover, these substrates may also predict patterns
of expected visceral (interoceptive) feedback in a way that serves to calibrate periph-
eral physiology with behavior and the metabolic demands of a given context (Barrett &
Simmons, 2015).
Understood in this way, “mismatches” between actual and predicted metabolic demands
can be viewed as visceral prediction errors. These errors may manifest in the magnitude
or patterning of stressor-​evoked changes in peripheral physiology. For example, a rise in
blood pressure in excess of 40 mmHg in preparing for a public speech can be seen as a
visceral prediction error—​insofar as it is a hemodynamic change that is disproportionate
to the actual metabolic needs of the context. Likewise, sustained or prolonged changes in
cardiovascular physiology that far outlast the ending of a given stressor can be viewed as
metabolically disproportionate or otherwise contextually unnecessary and inappropriate.
Visceral prediction errors of these types may be quantified by integrating laboratory stress
reactivity testing with methods of exercise physiology, wherein changes in cardiovascular
physiology that are in excess of oxygen consumption and metabolic requirements of a
context can be computed (see Ginty et al., 2017; Gianaros & Jennings, 2018).
The presumptive bases for visceral prediction errors may partly involve the resetting
or modulation of homeostatic functions by substrates of the central autonomic network,
allowing for context-​dependent changes in visceral control via the autonomic nervous
system. Visceral prediction errors underlying observable patterns of stress physiology
may also involve insensitivity to interoceptive and visceral feedback as well (Ginty et al.,
2017; Gianaros & Jennings, 2018; see also Chapter 17 in the present volume). For example,
as a result of such insensitivity, the visceral feedback provided by baroreceptors about
“exaggerated” and stressor-​evoked rises in blood pressure may not serve to minimize fu-
ture visceral prediction errors that manifest as exaggerated stress reactivity. Moreover, in-
teroceptive information from stressor-​evoked changes in cardiovascular physiology (e.g.
relayed by the baroreceptors) is capable of powerfully shaping the appraisal of threatening
and painful information (Garfinkel & Critchley, 2016; see also Chapter 7 in this volume).
Put simply, stressor-​evoked changes in peripheral physiology do not happen in a vacuum,
having “bottom-​up” effects on higher neural substrates. Open questions in this domain
extend to other parameters of physiology beyond the cardiovascular system that change
14

14 Interoception and the autonomic nervous system

with stress (e.g. immune and neuroendocrine functions), as well as how visceral feed-
back from multiple physiological parameters are integrated by higher neural substrates to
shape behavioral states as they unfold across contexts.

1.6  From interoception to interoceptive awareness

The multiple reciprocal links between interoception and psychological function, as
highlighted in the preceding sections, do not necessarily imply a conscious awareness or
intervention. However, several research strands across psychological sciences and cog-
nitive neuroscience have recently focused on our ability to become aware of interocep-
tive states and the importance that such states of interoceptive awareness have for the
awareness of the self and of others. Earlier psychological research has shown how higher
levels of interoceptive accuracy that is typically quantified in behavioral tasks that re-
quire participants to pay attention to interoceptive states such as heartbeats (Schandry,
1981), respiration (Daubenmier et al., 2013), or feelings of fullness and gastric sensitivity
(Herbert et al., 2012) influence emotional processing. For example, higher levels of in-
teroceptive accuracy are associated with more intense emotional experiences and better
emotion regulation (see Critchley & Harrison, 2013)
Capitalizing on such findings, more recent studies have expanded their focus to ask
questions about the role that interoceptive awareness may play for body representations
(for reviews see Craig, 2009; Tsakiris, 2010). Historically, the perception of one’s own
body from the outside (e.g. self-​recognition) and the perception of the body from within
(e.g. of signals coming from the visceral organs) have largely been studied independently.
For example, the question of how the brain produces the experience of body ownership
has focused mainly on multisensory integration. In the Rubber Hand Illusion (RHI), one
of the most influential experimental models of embodiment, watching a rubber hand
being stroked synchronously with one’s own unseen hand causes the rubber hand to be
experienced as part of one’s body (Botvinick & Cohen, 1998; Tsakiris, 2010). These results
speak in favor of an exteroceptive model of the self within which self-​awareness is highly
malleable, subject to the perception of the body from the outside. However, exteroceptive
input represents only one set of channels of information available for body awareness. We
are also interoceptively aware of our body.
To address how interoceptive signals are integrated with exteroceptive signals to create
an integrated sense of the bodily self, Tsakiris, Tajadura-​Jiménez, and Costantini (2011)
measured and quantified Interoceptive Accuracy (IAcc) with the heartbeat-​counting task
and compared this with the change in body ownership caused by multisensory stimu-
lation, using the RHI as a paradigmatic case of the exteroceptive self. Participants with
lower IAcc experienced a stronger illusory sense of body ownership, suggesting that in
the absence of accurate interoceptive representations one’s model of self is predominantly
exteroceptive. While others had shown how a change in the body ownership during
RHI affects homeostatic regulation (Moseley et al., 2008), we now had evidence showing
that both the experience of body ownership, and subsequent changes in homeostatic
 15

Future perspectives 15

regulation, depend partly on levels of IAcc. Consistent with these behavioral findings,
neuroimaging and neuropsychological observations on the critical role of the insular
cortex for body awareness support the view that the ways in which we perceive our body
from the inside interact with our perception of the body from the outside. Right ante-
rior insula activity correlates with performance in interoceptive accuracy tasks (Critchley
et al., 2004). A rare single-​case study shows that heartbeat awareness decreased after in-
sular resection (Ronchi et al., 2015), and Couto and colleagues (2015) report impaired
interoceptive awareness with insular cortical or white matter lesions. Right mid-​posterior
insula activity correlates with the body ownership experienced during the Rubber Hand
Illusion, a paradigm that uses exteroceptive input (e.g. vision and touch) to study, in a
controlled way, the bodily self (Tsakiris et al., 2007). This same area seems to be the critical
lesion site for somatoparaphrenia—​a striking loss of body ownership (Karnath & Baier,
2010). These findings suggest that the interoceptive and the exteroceptive representations
of the body are integrated from the posterior to anterior subregions across the insular
cortex (Farb, Segal, & Anderson, 2013; Simmons et al., 2013). Moreover, this integration
appears to underpin the experience of my body as mine—​an experience that is the hall-
mark of the bodily self (Gallagher, 2000).
Such approaches paved the way for a large number of psychophysiological and neu-
roimaging studies (Aspell et al., 2013; Blefari et al., 2017; Crucianelli et al., 2017; Park
et  al., 2017; Ronchi et  al., 2017; Schauder et  al., 2015; Sel, Azevedo, & Tsakiris, 2016;
Shah, Catmur, & Bird, 2017; Suzuki et al., 2013; Tajadura-​Jiménez & Tsakiris, 2014) that
corroborate the basic hypothesis about the crucial psychological role that interoception
plays for self-​awareness and awareness of other people (see also Chapter 2 in the present
volume). Across different domains, from emotion processing (Dunn et al., 2010; Pollatos
et al. 2007) to body image (see Badoud & Tsakiris, 2017 for a review), and social cogni-
tion (Shah et al. 2017), the representation that one has of her internal body seems to be
crucial for the representation of the “material me”, as Sherrington would put it, in relation
to others (Fotopoulou & Tsakiris, 2017).

1.7  Future perspectives

The complexity of the central autonomic network, its vast interconnectivity with other
brain systems, and the broad functional impact of visceral afference all pose challenges to
progress in the interoceptive canon. Although there may certainly be a place for studies
of single interoceptive dimensions or measures, a full understanding is most likely to re-
quire the consideration of multiple parameters and physiological patterns, from a multi-​
system, multi-​level interacting (top-​down and bottom-​up) perspective. Thus, important
directions for the future include a focus on multiple interoceptive dimensions, interdis-
ciplinary perspectives (cognitive/​behavioral, neural, physiological, etc.) and interactions
among interoceptive processes and a broader range of affective-​cognitive processes. An
illustration of this point comes from the literature on autonomic specificity of emotions
and the role of visceral afference.
16

16 Interoception and the autonomic nervous system

Consider Cannon’s arguments against the James–​Lange view, that: (a) visceral responses
are too slow to underlie emotion; (b) similar visceral changes may occur across different
emotions and even non-​emotional states; (c)  inducing autonomic responses does not
necessarily invoke emotions; and (d) that autonomic denervations of various types do not
prevent emotional reactions. There clearly is not a simple isomorphism between auto-
nomic responses and emotions, but that really should not be expected; there are multiple
determinants of emotion and multiple levels of organization in affective substrates and
the central autonomic network. Emotions can be triggered cognitively, as well as from
external stimuli and context. These may well have different patterns of activation within
central networks. The arguments that visceral responses are too slow and that autonomic
denervations do not eliminate emotions would only be arguments against the view that
visceral afference is the only determinant of emotion. We now know this not to be the case.
Moreover, visceral afference is also likely to be “fast” enough to exert influences along the
lines of what James and Lange envisioned. Neurally mediated baroreflex influences on
heart rate and blood pressure control, for example, happen within milliseconds. These
influences extend to the processing of and reactivity to affective and nociceptive stimuli
on a heartbeat-​to-​heartbeat basis (Garfinkel and Critchley, 2016).
Canon’s two remaining arguments, that autonomic arousal may not lead to emotional
states and that similar autonomic responses may be associated with different emotions,
are also not compelling. As noted earlier, the effects of visceral afference are likely to
be varied and brain-​state dependent (e.g. see also Chapter  10 in the present volume).
A given pattern of visceral afference in one context (be that environmental, psychological,
or neurophysiological) may well have quite distinct effects. How we perceive a stimulus
or context is very much determined by expectations, goals, and attentional focus, in the
same way that classic visual illusions (e.g. the old woman/​young woman illusion; Boring,
1930) depend on attentional focus, expectations, priming, or other variables.
A given pattern of visceral afference, for example, may be functionally perceived, and
have different outcomes depending on the context or neurobiological or neurobehavioral
(brain) state. This has been termed the somatovisceral model of emotion (SAME)
wherein the same pattern of visceral afference may be associated with different emotions
(see Cacioppo et al., 1992; Norman et al., 2014). As with research on interoception, there
is consensus that research on interoceptive awareness must develop a wider and more
grounded measurement model, a richer theoretical framework that is at the same time
biologically plausible and psychologically meaningful, and a fuller characterization of
the links between different interoceptive systems and across interoceptive and extero-
ceptive systems. This contextual and brain-​state dependency needs to be considered in
interpreting and understanding visceral afference and our awareness of it, across different
fields, from basic research to clinical applications and from theoretical and computa-
tional perspectives. The chapters included in this volume intend to address at least some
of these aims.
 17

References 17

References
Adolfi, F., Couto, B., Richter, F., Decety, J., Lopez, J., Sigman, M., et al. (2017). Convergence of
interoception, emotion, and social cognition: A twofold fMRI meta-​analysis and lesion approach.
Cortex, 88, 124–​42. doi:10.1016/​j.cortex.2016.12.019.
Allen, G. V., Saper, C. B., Hurley, K. M., and Cechetto, D. F. (1991). Organization of visceral and limbic
connections in the insular cortex of the rat. Journal of Comparative Neurology, 311,  1–​16.
Annoni, J. M., Ptak, R., Caldara-​Schnetzer, A. S., Khateb, A., and Pollermann B. Z. (2003).
Decoupling of autonomic and cognitive emotional reactions after cerebellar stroke. Annals of
Neurology, 53,  654–​8.
Aspell, J. E., Heydrich, L., Marillier, G., Lavanchy, T., Herbelin, B., and Blanke, O. (2013). Turning
body and self inside out: Visualized heartbeats alter bodily self-​consciousness and tactile perception.
Psychological Science, 24, 2445–​53. <https://​doi.org/​10.1177/​0956797613498395>
Augustine, J. R. (1996). Circuitry and functional aspects of the insular lobe in primates including
humans. Brain Research Reviews, 22, 229–​44.
Badoud, D. and Tsakiris, M. (2017). From the body’s viscera to the body’s image: Is there a link between
interoception and body image concerns? Neuroscience & Biobehavioral Reviews, 77, 237–​46.
doi:10.1016/​j.neubiorev.2017.03.017.
Barbas, H., Saha, S., Rempel-​Clower, N., and Ghashghaei, T. (2003). Serial pathways from primate
prefrontal cortex to autonomic areas may influence emotional expression. BMC Neuroscience, 4, 25.
Barrett, L. F. and Simmons, W.K. (2015). Interoceptive predictions in the brain. Nature Reviews
Neuroscience, 16, 419–​29.
Bear, M.F., Connors, B.W., and Paradiso, M. A. (2016). Neuroscience: Exploring the Brain. Philadelphia,
PA: Wolters Kluwer.
Benarroch, E. E. (1993). The central autonomic network: Functional organization, dysfunction, and
perspective. Mayo Clinic Proceedings, 68(10), 988–​1001.
Bernard, C. (1878). Les Phénomènes de la Vie. Paris: Ballière.
Berntson, G. G., Bechara, A., Damasio, H., Tranel, D., Norman, G.J., and Cacioppo, J.T. (2011). The
insula and evaluative processes. Psychological Science, 22,  80–​6.
Berntson, G. G., Cacioppo, J. T., and Bosch, J. A. (2016). From homeostasis to allodynamic regulation.
In J. T. Cacioppo, L. G. Tassinary, and G. G. Berntson (eds). Handbook of Psychophysiology, 4th edn.
Cambridge: Cambridge University Press, pp. 401–​26.
Berthoud, H. R. and Neuhuber, W. L. (2000). Functional and chemical anatomy of the afferent vagal
system. Autonomic Neuroscience, 85,  1–​17.
Blefari, M. L., Martuzzi, R., Salomon, R., Bello-​Ruiz, J., Herbelin, B., Serino, A., et al. (2017). Bilateral
Rolandic operculum processing underlying heartbeat awareness reflects changes in bodily self-​
consciousness. European Journal of Neuroscience, 45(10), 1300–​12. <https://​doi.org/​10.1111/​
ejn.13567>
Boring, E. G. (1930). A new ambiguous figure. The American Journal of Psychology, 42, 444.
Botvinick, M. and Cohen, J. (1998). Rubber hands “feel” touch that eyes see. Nature, 391, 756.
doi:10.1038/​35784.
Boucher, O., Rouleau, I., Lassonde, M., Lepore, F., Bouthillier, A., and Nguyen, D. K. (2015). Social
information processing following resection of the insular cortex. Neuropsychologia, 71, 1–​10.
doi:10.1016/​j.neuropsychologia.2015.03.008.
Broca, P. (1878). Anatomie comparée des circonvolutions cérébrales: Le grand lobe limbique et la
scissure limbique dans la série des mammifères. Revue d’Anthropologie, 1, 385–​498.
18

18 Interoception and the autonomic nervous system

Cacioppo, J. T., Berntson, G. G., and Klein, D.J. (1992). What is an emotion? The role of somatovisceral
afference, with special emphasis on somatovisceral “illusions”. Review of Personality and Social
Psychology, 14,  63–​98.
Cameron, O. G. (2002). Visceral Sensory Neuroscience: Interoception. Oxford: Oxford University Press.
Cannon, W. B. (1927). The James–​Lange theory of emotions: A critical examination and an alternative
theory. American Journal of Psychology, 39, 106–​24.
Cannon, W. B. (1931). Again, the James–​Lange and the thalamic theory of emotion. Psychological
Review, 38, 281–​95.
Cannon, W. B. (1932/​1939). The Wisdom of the Body. London: Kegan Paul, Trench, Trubner & Co.
Cechetto, D. F. and Saper, C. B. (1990). Role of the cerebral cortex in autonomic function. In: A. D.
Loewy and K. M. Spyer (eds), Central Regulation of Autonomic Functions. New York, NY: Oxford
University Press, pp. 208–​33.
Chauhan, B., Mathias, C. J., and Critchley, H. D. (2008). Autonomic contributions to
empathy: Evidence from patients with primary autonomic failure. Autonomic Neuroscience, 140,
96–​100.
Cogliano, R., Crisci, C., Conson, M., Grossi, D., and Trojano, L. (2012). Chronic
somatoparaphrenia: A follow-​up study on two clinical cases. Cortex, 48,  758–​67
Cohen, S., Gianaros, P.J., and Manuck, S.B. (2016). A stage model of stress and disease. Perspectives on
Psychological Science, 11, 456–​63.
Couto, B., Adolfi, F., Sedeño, L., Salles, A., Canales-​Johnson, A., Alvarez-​Abut, et al. (2015).
Disentangling interoception: Insights from focal strokes affecting the perception of external and
internal milieus. Frontiers in Psychology, 6, 503. doi:10.3389/​fpsyg.2015.00503.
Craig, A. D. (2009). How do you feel-​now? The anterior insula and human awareness. Nature Reviews
Neuroscience, 10,  59–​70
Craig, A. D. (2014). How Do You Feel? An Interoceptive Moment with Your Neurobiological Self.
Princeton, NJ: Princeton University Press.
Critchley, H. D., Mathias, C. J., and Dolan R. J. (2001). Neuroanatomical basis for first-​and second-​
order representations of bodily states. Nature Neuroscience, 4, 207–​12.
Critchley, H. D. (2005). Neural mechanisms of autonomic, affective, and cognitive integration. Journal
of Comparative Neurology, 493, 154–​66.
Critchley, H. D., Wiens, S., Rotshtein, P., Ohman, A., Dolan, R. J., Öhman, A., et al. (2004). Neural
systems supporting interoceptive awareness. Nature Neuroscience, 7(2), 189–​95. <https://​doi.org/​
10.1038/​nn1176>
Critchley, H. D. and Harrison, N. A. (2013). Visceral influences on brain and behavior. Neuron, 77,
624–​38. <https://​doi.org/​10.1016/​j.neuron.2013.02.008>
Crucianelli, L., Krahé, C., Jenkinson, P. M., and Fotopoulou, A. (2017). Interoceptive ingredients of
body ownership: Affective touch and cardiac awareness in the rubber hand illusion. Cortex. <https://​
doi.org/​10.1016/​j.cortex.2017.04.018>
Daubenmier, J., Sze, J., Kerr, C. E., Kemeny, M. E., and Mehling, W. (2013). Follow your
breath: Respiratory interoceptive accuracy in experienced meditators. Psychophysiology, 50(8), 777–​
89. <https://​doi.org/​10.1111/​psyp.12057>
Dampney, R. A. (2015). Central mechanisms regulating coordinated cardiovascular and respiratory
function during stress and arousal. American Journal of Physiology. Regulatory, Integrative and
Comparative Physiology, 309, R429–​43.
Dampney, R. A. (2017). Resetting of the baroreflex control of sympathetic vasomotor activity
during natural behaviors: Description and conceptual model of central mechanisms. Frontiers in
Neuroscience, 11, 461. <https://​doi.org/​10.3389/​fnins.2017.00461>
 19

References 19

Dum, R. P., Levinthal, D. J., and Strick, P. L. (2016). Motor, cognitive, and affective areas of the cerebral
cortex influence the adrenal medulla. Proceedings of the National Academy of Sciences, 113(35),
9922–​7.
Dunn, B. D., Galton, H. C., Morgan, R., Evans, D., Oliver, C., Meyer, M., et al. (2010). Listening
to your heart. How interoception shapes emotion experience and intuitive decision making.
Psychological Science, 21(12), 1835–​44. <https://​doi.org/​10.1177/​0956797610389191>
Dworkin, B. R. (1993). Learning and Physiological Regulation. Chicago, IL: University of Chicago Press.
Farb, N. A. S., Segal, Z. V., and Anderson, A. K. (2013). Attentional modulation of primary interoceptive
and exteroceptive cortices. Cerebral Cortex, 23(1), 114–​26. <https://​doi.org/​10.1093/​cercor/​bhr385>
Farrer, C., Franck, N., Georgieff, N., Frith, C. D., Decety, J., and Jeannerod M. (2003). Modulating the
experience of agency: A positron emission tomography study. Neuroimage, 18, 324–​33.
Fisher, J. P., Young, C. N., and Fadel, P. J. (2015). Autonomic adjustments to exercise in humans.
Comprehensive Physiology, 5, 475–​512.
Foley, J. O. and DuBois, F. S. (1937). Quantitative studies of the vagus nerve in the cat. I. The ratio of
sensory to motor fibers. Journal of Comparative Neurology, 67,  49–​64.
Fotopoulou, A. and Tsakiris, M. (2017). Mentalizing homeostasis: The social origins of interoceptive
inference. Neuropsychoanalysis, 19,  3–​28
Frank, S., Kullmann, S., and Veit, R. (2013). Food related processes in the insular cortex. Frontiers in
Human Neuroscience, 7, 499.
Freire-​Maia, L. and Azevedo, A. D. (1990). The autonomic nervous system is not a purely efferent
system. Medical Hypotheses, 32,  91–​9.
Furlan, A. D., Mailis, A., and Papagapiou, M. (2000). Are we paying a high price for surgical
sympathectomy? A systematic literature review of late complications. Journal of Pain, 1, 245–​57.
Gandola, M., Invernizzi, P., Sedda, A., Ferrè, E.R., Sterzi, R., Sberna, M., et al. (2012). An anatomical
account of somatoparaphrenia. Cortex, 48, 1165–​78.
Gallagher, S. (2000). Philosophical conceptions of the self: Implications for cognitive science. Trends in
Cognitive Sciences, 4,  14–​21.
García, A. M., Sedeño, l., Murcia, E. H., Couto, B., and Ibáñez, A. (2016). A lesion-​proof brain?
Multidimensional sensorimotor, cognitive, and socio-​affective preservation despite extensive
damage in a stroke patient. Frontiers in Aging Neuroscience, 8, 335.
Garfinkel, S. N. and Critchley, H. D. (2016). Threat and the body: How the heart supports fear
processing. Trends in Cognitive Science, 20,  34–​46.
Gasquoine, P. G. (2014). Contributions of the insula to cognition and emotion. Neuropsychology Review,
24,  77–​87.
Gianaros, P. J., Onyewuenyi, I. C., Sheu, L. K., Christie, I. C., and Critchley, H. D. (2012). Brain
systems for baroreflex suppression during stress in humans. Human Brain Mapping, 33, 1700–​16.
Gianaros, P. J. and Wager, T. D. (2015). Brain–​body pathways linking psychological stress and physical
health. Current Directions in Psychological Science, 24, 313–​21.
Gianaros, P. J. and Jennings, J. R. (2018). Host in the machine: A neurobiological perspective on
psychological stress and cardiovascular disease. American Psychologist. doi:10.1037/amp0000232.
Ginty, A. T., Kraynak T. E., Fisher J. P., and Gianaros P. G. (2017). Cardiovascular and autonomic
reactivity to psychological stress: Neurophysiological substrates and links to cardiovascular disease.
Autonomic Neuroscience. doi:10.1016/​j.autneu.2017.03.003.
Goldstein, D. S. (2012). Neurocardiology. Cardiovascular Therapeutics, 30, e89–​e106.
Gotink, R. A., Meijboom, R., Vernooij, M. W., Smits, M., and Hunink, M. G. (2016). 8-​week
mindfulness based stress reduction induces brain changes similar to traditional long-​term
meditation practice—​A systematic review. Brain and Cognition, 108,  32–​41.
20

20 Interoception and the autonomic nervous system

Heims, H. C., Critchley, H. D., Martin, N. H., Jäger, H. R., Mathias, C. J., and Cipolotti, L. (2006).
Cognitive functioning in orthostatic hypotension due to pure autonomic failure. Clinical Autonomic
Research, 16, 113–​20.
Heinbecker, P. and O’Leary, J. (1933). The mammalian vagus nerve—​afunctional and histological study.
American Journal of Physiology, 106, 623–​46.
Herbert, B. M., Muth, E. R., Pollatos, O., and Herbert, C. (2012). Interoception across modalities: On
the relationship between cardiac awareness and the sensitivity for gastric functions. PLoS One, 7(5),
e36646.
Hernández, S. E., Suero, J., Barros, A., González-​Mora, J. L., and Rubia, K. (2016). Increased grey
matter associated with long-​term sahaja yoga meditation: A voxel-​based morphometry study. PLoS
One, 11(3).
Hess, W. R. (1954). Diencephalon: Autonomic and Extrapyramidal Functions. New York, NY: Grune &
Stratton.
Jackson, J. H. (1884). On the evolution and dissolution of the nervous system. Croonian Lectures 3, 4
and 5 to the Royal Society of London. Lancet, 1, 555–​739.
James, W. (1984). What is emotion. Mind, 9, 188–​205.
Jänig, W. and Morrison, J. F. B. (1986). Functional properties of spinal visceral afferents supplying
abdominal and pelvic organs with special emphasis on visceral nociception. In: F. Cervero and J.
F. B. Morrison (eds). Visceral sensation. Progress in Brain Research, 67, 87–​114.
Jänig W. and Häbler, H.-​J. (1995). Visceral-​autonomic integration. In: G. F. Gebhart (ed.). Visceral pain.
Progress in Pain Research and Management, Vol. 4. Seattle, WA: IASP Press, pp. 311–​48.
Karnath, H.-​O. and Baier, B. (2010). Right insula for our sense of limb ownership and self-​
awareness of actions. Brain Structure and Function, 214(5–​6), 411–​17. <https://​doi.org/​10.1007/​
s00429-​010-​0250-​4>
Lange, C. G. (1885). Om sindsbevaegelser: et psycho-​fysiologisk studie. Copenhagen: Jacob Lunds.
Laneri, D., Schuster, V., Dietsche, B., Jansen, A., Ott, U., and Sommer, J. (2016). Effects of long-​term
mindfulness meditation on brain’s white matter microstructure and its aging. Frontiers in Aging
Neuroscience, 7, 254.
Langley, J. N. (1898). On the union of cranial autonomic (visceral) fibers with the nerve cells of the
superior cervical ganglion. The Journal of Physiology, 23, 240–​70.
Langley, J. N. (1903). The autonomic nervous system. Brain, 26,  1–​26.
Loewy, A. D. (1991). Forebrain nuclei involved in autonomic control. Progress in Brain Research, 87,
253–​68.
MacLean, P. D. (1954). Psychosomatic Medicine. Alphen aan den Rijn: Wolters Kluwer Health, Inc.
Maffei, A., Haley, M., and Fontanini, A. (2012). Neural processing of gustatory information in insular
circuits. Current Opinion in Neurobiology, 22, 709–​16.
Maffei, A. M., Haley, M., Fontanini, A., and Miranda, M. I. (2012). Taste and odor recognition
memory: The emotional flavor of life. Review of Neuroscience, 23, 481–​99.
Mailis, A. and Furlan, A. (2003). Sympathectomy for neuropathic pain. The Cochrane Database of
Systematic Reviews, 2, CD002918.
Mazzola, L., Isnard, J., Peyron, R., and Mauguière, F. (2012). Stimulation of the human cortex and the
experience of pain: Wilder Penfield’s observations revisited. Brain, 135, 631–​40.
McEwen, B. S. (2012). Brain on stress: How the social environment gets under the skin. Proceedings of
the National Academy of Sciences of the USA, 16, 109 Suppl 2, 17180–​5.
Montavont, A., Mauguière, F., Mazzola, L., Garcia-​Larrea, L., Catenoix, H., Ryvlin, P., et al. (2015). On
the origin of painful somatosensory seizures. Neurology, 84, 594–​601.
 21

References 21

Moro, V., Pernigo, S., Tsakiris, M., Avesani, R., Edelstyn, N. M., Jenkinson, P.M., et al. (2016).
Motor versus body awareness: Voxel-​based lesion analysis in anosognosia for hemiplegia and
somatoparaphrenia following right hemisphere stroke. Cortex, 83,  62–​77.
Moseley, G. L., Olthof, N., Venema, A., Don, S., Wijers, M., Gallace, A., et al. (2008). Psychologically
induced cooling of a specific body part caused by the illusory ownership of an artificial counterpart.
Proceedings of the National Academy of Sciences of the USA, 105(35), 13169–​73. <https://​doi.org/​
10.1073/​pnas.0803768105>
Myers, B. (2017). Corticolimbic regulation of cardiovascular responses to stress. Physiology & Behavior,
172,  49–​59.
Neafsey, E. J. (1990). Prefrontal cortical control of the autonomic nervous system: Anatomical and
physiological observations. Progress in Brain Research, 85, 147–​65.
Nieuwenhuys, R. (2012). The insular cortex: A review. Progress in Brain Research, 195, 123–​63.
Norman, G. J., Berntson, G. G., and Cacioppo, J. T. (2014). Emotion, somatovisceral afference and
autonomic regulation. Emotion Review, 6, 113–​23.
Obrist, P. A. (1981). Cardiovascular Psychophysiology: A Perspective. New York, NY: Plenum Press.
Oppenheimer, S. M. and Cechetto, D. F. (2016). Insular cortex and the regulation of cardiac function.
Comprehensive Physiology, 6, 1081–​133.
Papez, J. (1937). A proposed mechanism of emotion. Reprinted in Journal of Neuropsychiatry and
Clinical Neuroscience (1995) 7, 103–​12.
Park, H.-​D., Bernasconi, F., Salomon, R., Tallon-​Baudry, C., Spinelli, L., Seeck, M., et al. (2017).
Neural sources and underlying mechanisms of neural responses to heartbeats, and their role in
bodily self-​consciousness: An intracranial EEG Study. Cerebral Cortex, 116, 1–​14. <https://​doi.org/​
10.1093/​cercor/​bhx136>
Penfield, W. and Jasper, H. (1954). Epilepsy and the Functional Anatomy of the Human Brain. New York,
NY: Little, Brown & Co.
Penfield, W., and Faulk, M. E. (1955). The insula; further observations on its function. Brain, 78,
445–​70.
Pollatos, O., Gramann, K., and Schandry, R. (2007). Neural systems connecting interoceptive
awareness and feelings. Human Brain Mapping, 28(1), 9–​18. <https://​doi.org/​10.1002/​hbm.20258>
Ranson, S. W., Kabat, H., and Magoun, H. W. (1935). Autonomic responses to electrical stimulation of
hypothalamus, preoptic region and septum. Archives of Neurology & Psychiatry, 33, 467–​77.
Rolls, E. T. (2015). Taste, olfactory, and food reward value processing in the brain. Progress in
Neurobiology, 127–​8,  64–​90.
Ronchi, R., Bello-​Ruiz, J., Lukowska, M., Herbelin, B., Cabrilo, I., Schaller, K., et al. (2015).
Right insular damage decreases heartbeat awareness and alters cardio-​visual effects on bodily
self-​consciousness. Neuropsychologia, 70, 11–​20. <https://​doi.org/​10.1016/​j.neuropsycholo
gia.2015.02.010>
Ronchi, R., Bernasconi, F., Pfeiffer, C., Bello-​Ruiz, J., Kaliuzhna, M., and Blanke, O. (2017).
Interoceptive signals impact visual processing: Cardiac modulation of visual body perception.
NeuroImage, 158, 176–​85. <https://​doi.org/​10.1016/​j.neuroimage.2017.06.064>
Saive, A-​L, Royet, J-​P, and Plailly, J. (2014). A review on the neural bases of episodic odor
memory: From laboratory-​based to autobiographical approaches. Frontiers in Behavioral
Neuroscience, 8, 240.
Saper, C. B. (2002). The central autonomic nervous system: Conscious visceral perception and
autonomic pattern generation. Annual Review of Neuroscience, 25, 433–​69.
Schandry, R. (1981). Heart beat perception and emotional experience. Psychophysiology, 18(4),  483–​8.
2

22 Interoception and the autonomic nervous system

Schauder, K. B., Mash, L. E., Bryant, L. K., and Cascio, C. J. (2015). Interoceptive ability and body
awareness in autism spectrum disorder. Journal of Experimental Child Psychology, 131, 193–​200.
<https://​doi.org/​10.1016/​j.jecp.2014.11.002>
Schulkin, J. (2003). Rethinking Homeostasis: Allostatic Regulation in Physiology and Pathophysiology.
Cambridge, MA: MIT Press.
Sel, A., Azevedo, R. T., and Tsakiris, M. (2016). Heartfelt self: Cardio-​visual integration affects self-​face
recognition and interoceptive cortical processing. Cerebral Cortex. <https://​doi.org/​10.1093/​cercor/​
bhw296>
Shah, P., Catmur, C., and Bird, G. (2017). From heart to mind: Linking interoception, emotion, and
theory of mind. Cortex. April 28. Pii, S0010-​9452(17)30060-​6. doi:10.1016/​j.cortex.2017.02.010.
Sherrington, C. S. (1900). Experiments on the value of vascular and visceral factors for the genesis of
emotion. Proceedings of the Royal Society (London) B, 66, 390–​403.
Shoemaker, J. K. and Goswami, R. (2015). Forebrain neurocircuitry associated with human reflex
cardiovascular control. Frontiers in Physiology, 6, 240. doi:10.3389/fphys.2015.00240. <https://www.
frontiersin.org/articles/10.3389/fphys.2015.00240/full>
Simmons, W. K., Avery, J. A., Barcalow, J. C., Bodurka, J., Drevets, W. C., and Bellgowan, P. (2013).
Keeping the body in mind: Insula functional organization and functional connectivity integrate
interoceptive, exteroceptive, and emotional awareness. Human Brain Mapping, 34, 2944–​58.
<https://​doi.org/​10.1002/​hbm.22113>
Soudry, Y., Lemogne, C., Malinvaud, D., Consoli, S.-​M., and Bonfils, P. (2011). Olfactory system and
emotion: Common substrates. European Annals of Otorhinolaryngology, Head and Neck Diseases,
128(1),  18–​23.
Squire, L., Berg, D., Bloom, F. E., du Lac Anirvan, S., Nicholas, G., and Spitzer, C. (2012).
Fundamental Neuroscience. Oxford: Elsevier.
Sterling, P. and Eyer, J. (1988). Allostasis: A new paradigm to explain arousal pathology. In: S. Fisher
and J. Reason (eds), Handbook of Life Stress, Cognition and Health. New York, NY: John Wiley &
Sons, pp. 629–​49.
Strigo, I. A. and Craig, A. D. (2016). Interoception, homeostatic emotions and sympathovagal balance.
Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 371, 1708.
Suzuki, K., Garfinkel, S. N., Critchley, H. D., and Seth, A. K. (2013). Multisensory integration across
exteroceptive and interoceptive domains modulates self-​experience in the rubber-​hand illusion.
Neuropsychologia, 51, 2909–​17. <https://​doi.org/​10.1016/​j.neuropsychologia.2013.08.014>
Tajadura-​Jiménez, A. and Tsakiris, M. (2014). Balancing the “inner” and the “outer” self: Interoceptive
sensitivity modulates self-​other boundaries. Journal of Experimental Psychology: General, 143(2).
<https://​doi.org/​10.1037/​a0033171>
Tsakiris, M., Hesse, M. D., Boy, C., Haggard, P., and Fink, G. R. (2007). Neural signatures of body
ownership: A sensory network for bodily self-​consciousness. Cerebral Cortex, 17, 2235–​44.
Tsakiris, M. (2010). My body in the brain: A neurocognitive model of body-​ownership.
Neuropsychologia, 48(3), 703–​12. <https://​doi.org/​10.1016/​j.neuropsychologia.2009.09.034>
Tsakiris, M., Tajadura-​Jiménez, A., and Costantini, M. (2011). Just a heartbeat away from one’s
body: Interoceptive sensitivity predicts malleability of body-​representations. Proceedings. Biological
Sciences/​The Royal Society, 278, 2470–​6. <https://​doi.org/​10.1098/​rspb.2010.2547>
Tsakiris, M. and Critchley, H. (2016a). Interoception beyond homeostasis: Affect, cognition and
mental health. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences,
371,  1–​6.
Tsakiris, M. and Critchley, H. (eds) (2016b). Theme issue: “Interoception beyond homeostasis: Affect,
cognition and mental health”. Philosophical Transactions of the Royal Society of London. Series B, 371,
issue 1708.
 23

References 23

Uddin, L. Q., Nomi, J. S., Hébert-​Seropian, B., Ghaziri, J., and Boucher O. (2017). Structure and
function of the human insula. Journal of Clinical Neurophysiology, 34, 300–​6. doi:10.1097/​
WNP.0000000000000377.
Yates, B. J. and Stocker, S. D. (1998). Integration of somatic and visceral inputs by the
brainstem: Functional considerations. Experimental Brain Research, 119, 269–​75.
24
 25

PART II

Mentalizing interoception
Advances and challenges
26
 27

Chapter 2

The body as first prior: Interoceptive

predictive processing and the primacy
of self-​models
Micah Allen and Manos Tsakiris

2.1 Introduction
Behind every human experience lies a tapestry of embodied sensations; the beating of
the heart, air filling the lungs, or the skin of the chest pulling taught with each breath
are all instances of such sensations. Fascinatingly, although the sensations of the living,
moving body are ever-​available, they can also go easily unnoticed; in reading the pre-
vious sentence, you probably became more aware of them. Our heartbeat is usually silent
unless we have exerted physical effort; in which case, its thrumming beat comes to dom-
inate, however briefly, our self-​awareness. In this sense, one can characterize our embod-
iment as a kind of ever-​present, yet just behind-​the-​scenes kind of experience. What can
cognitive neuroscience reveal about this special relationship between the body and our
conscious mind?
Phenomenologists typically describe the transparency of the body as a preconditioning
factor for awareness of the self, world, and others, emphasizing that all experience takes
place from within this embodied subjective frame (Gallagher & Zahavi, 2013; Heidegger,
1996; Merleau-​Ponty, 1996). Until recently however, the foundational role of embodied
sensations in shaping our perceptual, cognitive, and socio-​affective activity has been
largely neglected by cognitive and computational neuroscience. This neglect may be in
part due to a paucity of mechanistic theories equipped to generate testable hypotheses
concerning the neural integration of visceral and exteroceptive bodily sensations. Now,
resurgent interest in Bayes-​inspired neuroscience is driving an explosion of empirical
data and theoretical models which places the visceral milieu, its homeostatic functioning,
and our interoceptive awareness thereof back on center stage. In this chapter, we provide
an overview of the latest developments in this area, considering the developing notions
of bodily precision and computational “self-​models.” At the core of our approach is the
hypothesis that homeostatic priors are “first priors” in the sense that, given the impetus
for the organism to maintain homeostasis, they are afforded a privileged status within
the cortical hierarchy. Having explained their privileged status for ensuring the stability
of organism, we then turn to the psychological importance that interoceptive priors and
28

28 The body as first prior

precision have for self-​awareness, in particular for our understanding of domain-​specific

(i.e. body-​awareness) as well as more global self-​models (i.e. metacognition).

2.2  Predictive processing and the hierarchical

optimization of precision
New developments in computational and cognitive neuroscience drive an increased em-
phasis on top-​down, prediction-​based theories of mind and brain function. These col-
lectively describe the brain as a “prediction engine” which embodies or implements a
statistically generative model describing the hidden causes of sensory inputs (Clark, 2013;
Friston, 2010; Hohwy, 2013). Although so-​called predictive processing (PP) approaches
have a long history in neuroscience, beginning with early models of motor control (for
review, see Miall & Wolpert, 1996), this renewed interest is driven in part by their recent
generalization to domains such as social (Friston & Frith, 2015; Gallagher & Allen, 2016;
Koster-​Hale & Saxe, 2013), affective (Barrett & Simmons, 2015; Joffily & Coricelli, 2013;
Seth, 2013), and embodied cognition (Ainley et al., 2016; Allen & Friston, 2016; Apps &
Tsakiris, 2014). In particular, this new wave emphasizes the role of the visceral body in
maintaining homeostasis as a fundamental aspect of the predictive mind.
The essential elements of the PP framework have been elaborated in numerous arti-
cles (Clark, 2013; Friston, 2005, 2010; Hohwy, 2014; see also Chapter  15 in the present
volume); here we rehearse the basic elements to inform a discussion of their applica-
tion to interoception and embodiment. The most general, formally articulated theory of
predictive processing is found in the Free Energy Principle (FEP) as developed by Karl
Friston (Friston, 2009, 2010, 2013). The FEP dictates that an organism can only survive
by minimizing the entropy of internal states. That is, an entity which is subject to un-
bounded entropy—​such as a snowflake—​will not survive for very long. In contrast to the
snowflake, which can only increase in entropy over time, a living organism like you and
me, or an amoeba, can take self-​sustaining (or autopoietic) actions which, for a time, resist
entropy. This entails that an organism is fundamentally comprised of a Markov Blanket,
constituted by an interlocking causal web of self-​sustaining actions, new sensations, and
an internal model interlinking the two. The Markov Blanket thus demarcates internal from
external states and enables an organism to resist entropy both by acting on the world so as
to render it closer to one’s predictions (active inference) or through updating one’s internal
states (perceptual inference). In this way, the FEP furnishes a basic normative principle by
which to guide our understanding of the embodied mind; biological organisms persist by
virtue of their ability to maintain a stable autopoietic self, and to predict the impact of fu-
ture actions on that homeostatic imperative. However, understanding how this normative
principle plays out in specific neurobiological process theories requires specifying a partic-
ular architecture; for example, at what level of the cortical hierarchy do signals conveying
successful homeostasis exert their influence on cognition, and in what manner, precisely?
Here it is useful to appeal to the varieties of Bayesian Predictive Coding (BPC) found in
the predictive processing literature. As a general process theory, BPC-​based approaches
 29

Predictive processing and the hierarchical optimization of precision 29

describe the brain as a hierarchical network implementing a distributed Bayesian infer-

ence scheme. Within this network, “top-​down” signals encode a prior probability dis-
tribution (prediction); incoming sensory data are understood as a statistical likelihood,
and the precision-​weighted posterior difference between these two inputs—​the predic-
tion error—​is passed upward to the next level of the hierarchy.1 Neurobiologically, this
scheme is thought to be accomplished by a canonical microcircuit motif, in which deep
(infra-​granular) pyramidal cells receive prediction errors from lower levels and superfi-
cial (agranular) cells encode predictions. This motif is assumed to be replicated across
the brain2 such that the prediction error originating from the deep layer of, for example,
primary visual cortex (V1) will be integrated with a prediction originating from the su-
perficial layer of secondary visual cortex (V2), and so on. As higher-​order circuits must in-
tegrate increasingly multimodal and abstract inputs (e.g. multisensory parietal areas), the
predictions and prediction errors at each level become subsequently more removed from
the fast temporal dynamics of sensory stimulation, and more intrinsically self-​related.3
Thus, each canonical microcircuit integrates top-​down expectations and bottom-​up
sensory prediction errors according to a cascading application of approximate Bayesian
inference. This entails that top-​down predictions and bottom-​up prediction errors are
neurobiologically encoded in such a way as to reflect their precision (i.e. confidence or
inverse variance), with the interaction between the two governed by whichever input is
more precise (Feldman & Friston, 2010; Kanai et al., 2015; Moran et al., 2013). As an in-
tuitive example of how precision-​weighting works, consider a scientist, Daisy, who sets
out to test the hypothesis that the average height of Dutch people is 182 cm (6 ft). If Daisy
is unsure about her prediction—​perhaps previous studies have had divergent results, or
there is a lack of prior data—​then she will assign a lower prior precision.4 Conversely, if
Daisy visits Holland and only manages to sample very few individuals, or perhaps uses
an unreliable measurement device, then the precision of the sensory data (or likelihood)
will be very low. Bayes’ law captures the intuitive notion then that one should update one’s
theory or belief according to whichever aspect of inference—​prior knowledge or empir-
ical observations—​is more reliable.
The hierarchical optimization of precision naturally motivates an account of both
endogenous and exogenous attentional spotlight effects (Fardo et  al., 2017; Feldman
& Friston, 2010). For example, the classical Posner cueing task has been modeled as
inducing an expectation of enhanced expected precision on the right versus left visual
hemifield (Feldman & Friston, 2010). Here, descending neuromodulation encoding

1
The distinction of likelihood vs prediction error is here relative to the hierarchical level one describes;
the prediction error at level 1 becomes the inferential likelihood at level 2, and so on (see Figure 2.1).
2
Or plausibly, within a particular hierarchy if one does not subscribe to radically predictive processing
(i.e. the notion that all systems everywhere in the brain are predictive).
3
In the sense of describing increasingly intrinsic (as opposed to extrinsic) processes: level B describes
Level A, level C describes levels A and B, and so on.
4
Statistically, the prior probability distribution will have a wider variance or standard deviation.
30

30 The body as first prior

expected precision enhances the precision of sensory prediction errors on the expected
field, causing them to dominate in perception, while simultaneously reducing the preci-
sion of stimuli presented to the unattended visual field, weakening their ability to attract
attention. Similarly, exogenous attentional effects, or “salience,” are also well-​captured by
the model, as bottom-​up changes in stimulus intensity will drive stronger post-​synaptic
depolarization, increasing bottom-​up gain to result in highly precise prediction errors
which can override top-​down predictions. Thus, according to most BPC-​related schemes,
both expected (top-​down) and sensory precision play a crucial role in dictating the pri-
ority of information processing in the brain.
These computational tools give us a fresh window into understanding both visceral
perception itself and the integration of visceral signals within a more general self-​model.
For example, respiratory or cardiac control, awareness, and disruptions thereof can be re-
lated to various levels of a predictive coding hierarchy in visceromotor brain areas (Faull,
Hayen, & Pattinson, 2017; Stephan et al., 2016). Visceral awareness in general has been
proposed to depend upon the precision (or confidence, more on this later) of the inter-
oceptive hierarchy (Ainley et al., 2016). As any hierarchical network will necessarily en-
code more abstract, supramodal representations at higher levels, others have emphasized
a more general role of interoceptive predictions in emotion, self-​awareness, and motor
control (Seth, 2013; Seth, Suzuki, & Critchley, 2012). The homeostatic axioms of the FEP
have further led to increased emphasis on the global priority of visceral predictive codes
in conditioning cortical representations and self-​awareness (Allen & Friston, 2016; Barrett
& Simmons, 2015). To improve our understanding of these views, we now introduce the
concept of “embodied precision” (see also Ainley et al., 2016; Fotopoulou, 2013) and the
related notion of homeostasis in predictive processing.

2.3  Embodied precision and homeostatic “first priors”

The FEP sets homeostatic and allostatic function as the defining core of organismic life
(Friston, 2013). This emphasis on an imperative to maintain internal states in the face of
uncertainty renders FEP-​based process theories embodied in the first order. For the free-​
energy minimizing agent, “value” or “optimality” should be understood as an imperative
to maintain homeostasis (Friston et  al., 2015). This is accomplished not by subserving
some external economic principle but instead through modelling the self, the world, and
their interaction in such a way as to maximize the chance of survival. In this sense, it
might be said that the autopoetic principle at the basis of FEP acts as a kind of “first prior”;
Bayesian inference is always bootstrapped from within the subjective needs of the agent
and its embodied econiche within the world (Bruineberg, Kiverstein, & Rietveld, 2016;
Bruineberg & Rietveld, 2014).
In terms of a more specific process theory, there are multiple ways in which visceral,
tactile, and proprioceptive bodily signals might be afforded some privileged status within
the cortical hierarchy. The FEP dictates that organisms are themselves “models of the
world” (in the cybernetic sense); those models are then passed on through phylogenetic
evolution as genetic “priors” which constrain every aspect of biological function from
 31

Embodied precision and homeostatic “first priors” 31

the DNA-​driven development of synaptic weights and cortical networks to the very mor-
phology of the agent itself. In this sense, the priors one is born with5 are also imbued
with an inherited prior precision.6 Interoceptive signaling, perception, and control are
perhaps the phylogenetically oldest of all sensory motor systems; all vertebrates possess
the basic homeostatic biomechanics of pumping blood, respiratory oxygenation, and so
on. Further, in terms of importance for organism survival, the stability and reliability of
our visceral function is vital. Quite literally, if our internal organs become unreliable, the
immediate result is death. It makes sense then that evolutionarily speaking, the afferents
of the interoceptive hierarchy are likely to be afforded the highest expected precision of
any sensorimotor channel.
There are several rather profound implications of this observation. Recall that the opti-
mization of precision across the hierarchy can be unpacked as a mechanism for both top-​
down, endogenous attention and bottom-​up saliency (Feldman & Friston, 2010). Here,
salience is nothing more than the relative precision of incoming prediction errors. If the
visceral body is afforded, a priori, the highest expected precision then interoceptive pre-
diction errors should dominate the perceptual hierarchy (see Figure 2.1). This means that
perceptual and value-​based salience are, for the homeostatic organism, always relative to
the impact some stimulus or decision has on the organisms’ visceral prediction.
Note that this account goes much further than previous theories such as the somatic
marker hypothesis (Damasio, 1999). The somatic marker hypothesis states in essence
that subjective value is determined by adding or weighting some bodily representation
to a value computation. Instead, according to our account, salience is literally defined
by whatever has the most (or least) impact on visceral and autonomic homeostasis.
Visceral sensations here are the dominant basis to which perceptual-​and value-​based
computations are added; through active inference, cognition is enslaved to embodiment,
rather than the other way around. In the perceptual domain, the implication is that un-
expected deviations in heart rate, gut response, or other systems may literally change
the way we perceive the world and our metacognitive uncertainty about such percepts.
Indeed, it has recently been shown that unexpected disgust signals can reverse the impact
of sensory precision on the exteroceptive perception of uncertainty itself (Allen et  al.,
2016), demonstrating the priority of visceral surprise as dictated by homeostatic prior
precision.
The presumed a priori hyper-​precision of visceral channels may explain a variety of
classical cognitive phenomenon, beginning with substantive work done by mid-​century
Soviet psychologists who, inspired by the pioneering work of Pavlov and Sokolov, exhaus-
tively investigated “interoceptive conditioning” and its relationship to the orienting reflex
(Razran, 1961; Sokolov, 1963). In an impressive body of work combining various means of
stimulating cardiac, respiratory, and enteric nervous systems with classical conditioning

5
For example, consider the tendency to perceive certain conjunctions of circles as “faces” with an inher-
ently social meaning, or to inhabit certain environments.
6
Also known as “hyper-​priors,”—​priors dictating an expected level of sensory precision.
32

32 The body as first prior

Key
(a) Precision-Weighted Inference
Tactile
Prior
Likelihood

Visual
Likelihood
Likelihood

Posterior
Visceral Input
(Prediction Error)

(b) Accumulated Visceral Precision (c) Expected Visceral Precision

(sensory-weighted) (prior-weighted)

Figure 2.1  Visceral influences on precision-​weighted inference. (a) Under the principles of

Bayesian Inference, prediction errors are weighted by their statistical confidence (i.e. inverse
variance or precision). Both “top-​down” predictions and “bottom-​up” prediction errors can be
more or less precise. In this example, a precise prior belief (in red, relative to likelihood or sensory
precision, in blue) has shifted a perceptual prediction error (dotted black line) towards the
prediction. Note that the precision of the prediction error itself is the product of both prior and
sensory precision; some models posit that an additional system must read out the precision of
prediction errors to guide subjective sensitivity and/​or confidence. (b) This example is expanded
to consider how visceral information may enter a low-​level multisensory “self-​model” combining
prior beliefs (red) with visual (green) and tactile (blue) sensory inputs. In this example, visceral
signals are inherently more precise than other signals (see text), causing sensory signals to
overwhelm prior expectations. (c) Here, highly precise visceral signals strengthen the influence
of prior beliefs rather than feed-​forward prediction errors. Note that in both examples, the
hypothetical hyper-​precision of visceral signals causes the prediction error to be highly stable,
regardless whether prior or likelihood are weighted.

paradigms and biophysiological recording, these scientists sought to understand how vis-
ceral processes contribute to both implicit and explicit cognitive processes. This approach
enabled them to study how exteroceptive unconditioned stimuli could be transferred to
conditioned interoceptive responses, and vice versa. This revealed not only that subjects
could quickly learn to associate internal sensations with exteroceptive phenomena but
also that interoceptive learning itself was much faster to acquire and harder to extinguish
than purely external learning.
This is exactly what one would expect if the brain affords higher a priori precision to
internal changes than to external stimuli. This enhanced precision would lead to predic-
tion errors on the interoceptive channel dominating how we learn about various cues,
particularly when those highly precise prediction errors are integrated with other sen-
sory channels. In a sense, this can be recast as an account of how interoceptive signals
themselves provide a source of reliability in our self-​perception and conscious awareness.
 3

Self-models and interoceptive predictions 33

Interoceptive signals rise and fall according to a highly predictable pattern of slowly
oscillating circadian and hormonal biorhythms. This low volatility, high precision stream
of prediction errors may therefore provide a natural kind of “pacemaker” by which the
brain can bind together a more coherent, potentially conscious stream of experience from
the more noisy, unpredictable exteroceptive senses (see also Chapter  3 in the present
volume).
In the brain, this mechanism of enhanced interoceptive precision neatly explains
both phenomena within the laboratory, such as the interoceptive conditioning findings
outlined earlier, but also more relatable real-​world experiences. Nearly everyone has had
the unfortunate experience of eating some food which has gone off, and then suffered
through a day of food poisoning. Typically, this will result in a strong, lasting preference
never to eat that food again—​a form of highly precise learning which is often quite gen­
eral as well. For example, if a bad bit of venison burger makes you sick, you are likely
never to look at any burger (even a veggie one) in quite the same way again.
From the brain’s point of view, this sudden surge in highly precise interoceptive pre-
diction errors represents a potentially fatal event and affords for an equally precise alter-
ation of internal models. To overwhelm the precise interoceptive prediction errors, the
brain must essentially engage in “one-​shot learning,” sending an equally precise cascade
of predictions to alter the global neural networks governing appetite, aversion, and eating
behavior. Through interoceptive active inference, the brain thus ensures that the next time
you see a burger, or anything even closely related to it, you immediately feel some revul-
sion or perhaps even “see” the meat as green.
Interoceptive signals thus play a very important role in regulating the cortical hier-
archy and how we learn from aversive, potentially life-​threatening events. However, not
all is bad within the domain of the viscera. Indeed, highly pleasurable sensations can be
obtained from the interoceptive encoding of social touch and sexual intercourse. It is
interesting to note here the parallel between how the precision of visceral pain or dis-
tress may subserve the evolutionary priority to stay alive, and a potentially equal a priori
precision for affective or sexual sensations may subserve the principle of doing whatever
makes one more likely to reproduce, and therefore ensure the survival of one’s “species
model” (see Chapter 6 in the present volume). Also, of course, beyond aversive or affective
interoceptive prediction errors associated with specific experiences, interoceptive signals,
as first priors, may provide, as we explain in the following sections, the necessary basis
for the experience of “the same body always there” as James aptly put it, the experience,
in other words, of body ownership which seems to be central feature of the bodily self.

2.4  Self-​models and interoceptive predictions

An important emerging extension of the predictive coding literature is its application to
the notion of a predictive self-​model (Allen & Friston, 2016; Apps & Tsakiris, 2014) that
can explain fundamental aspects of selfhood such as body ownership and agency. At least
two predominant self-​model process theories can be motivated from EPP approaches;
a more domain-​constrained model which specifies how exteroceptive and interoceptive
34

34 The body as first prior

sensations are bound together to identify a sensation as originating from “me” (i.e.
having to do with body ownership and agency), and a more global or emergent concept
emphasizing the role of the hierarchical brain in iteratively predicting its own neural ac-
tivity. We discuss these in turn.

2.4.1  The multisensory self and body ownership

One of the first attempts towards a predictive coding self-​model focused strictly on the
domain of body awareness, namely the sense of body ownership (Apps & Tsakiris, 2014).
This model is posited entirely within the multisensory integration of bodily and extero-
ceptive sensations and thus describes a lower-​order self-​model situated between the in-
tegrative and sensori-​motor layers. The Rubber Hand Illusion (RHI), one of the most
influential and well-​studied experimental paradigms of body ownership of embodiment,
illustrates how predictions and predictions errors are used to compute the probability that
a body part belongs to me. In the RHI, watching a rubber hand being stroked synchro-
nously with one’s own unseen hand causes the rubber hand to be experienced as part of
one’s body (Tsakiris, 2010 for a review). The illusion and its striking effects (for review see
Tsakiris, 2010) illustrate how multisensory stimulation can lead to changes in how one’s
body is processed, leading to an update of what is experienced as “my body.”
How can a predictive coding account of the self explain changes in body ownership
that are driven by multisensory stimulation? During the induction of the illusion there
is considerable bottom-​up sensory surprise evoked in one’s sensory system. The somat-
osensory experience of touch on one’s hand that is temporally congruent with the vision
of touch on the rubber hand is surprising, as prior to stimulation participants cannot
see the touch on their own hand and would not predict that touch on the rubber hand
would evoke a sensation of touch. Instead they would predict that the vision of touch and
felt touch should come from the same single body part. This surprise will be explained
away by top-​down effects from multisensory areas where the relative precision weighting
of vision over touch may explain away the prediction error, updating the prediction of
what and where my “real” hand now is. In turn, perceptual learning processes will update
representations of one’s body, such that the probabilistic representation of one’s body is
different after synchronous multisensory stimulation. Therefore, one’s body, at least as
perceived exteroceptively, seems to be processed in a probabilistic manner as “the most
likely to be me” (Apps & Tsakiris, 2014). Such probabilistic representations are created
through the integration of top-​down predictions about the body with bottom-​up pre-
diction errors from unimodal sensory systems that are explained away by priors. These
results speak in favor of an exteroceptive model of the self, within which body ownership,
a key aspect of self-​awareness, is highly malleable, subject to the perception of the body
from the outside. However, exteroceptive input represents only one set of channels of
information available for self-​awareness. We are also interoceptively aware of our body
(Craig, 2002).
What is then the role of the interoceptive and bodily hierarchies in such predic-
tive self-​models? As one’s body is not simply perceived exteroceptively but is also felt
 35

Self-models and interoceptive predictions 35

interoceptively, a unifying model of the self must account for the integration of these
modalities. For example, during the RHI, the exteroceptive evidence suggests that what
I am looking at (i.e. the rubber hand) is my hand. However, if this is my hand, then there
are interoceptive prediction errors between how my true hand feels (i.e. interoceptive
prediction) and the fact that I cannot feel the rubber hand interoceptively. Therefore, ex-
teroceptive and interoceptive streams must be integrated to explain away these prediction
errors before the body can be represented as “self.”
Beyond the domain of predictive processing, numerous theorists have emphasized a
foundational role of visceral-​sensory integration in providing a basis for the “embodied
subjective frame” in the notion of interoceptive first priors (see section 2.3). In an
attempt to address the apparent dichotomy that existed in the field between a purely
exteroceptively driven model of the self (Apps & Tsakiris, 2014; Tsakiris, 2010) and an
interoceptively focused model (Seth, 2013), an FEP-​inspired model of the bodily self is
needed in which exteroceptive (tactile and visual) prediction errors are bound together
with interoceptive prediction errors. Indeed, the renewed focus on the potential role of
interoception for self-​awareness, beyond emotional processing, resulted in new lines of
research that attempted to study in parallel the integration of exteroceptive with intero-
ceptive predictions errors, thereby adding the important interoceptive element of how
“my body feels” to exteroceptive evidence (Aspell et al., 2013; Blefari, Martuzzi, Salomon
et al., 2017; Filippetti & Tsakiris, 2017; Ronchi et al., 2015; Sel et al., 2017; Suzuki et al.,
2013; Tsakiris, Jimenez, & Costantini, 2011).
Cardiac interoceptive signals and their cortical signatures have been related to bodily
self-​awareness in several recent studies. Inspired by classic bodily illusions, such as the
RHI (Botvinick & Cohen, 1998), the Full-​Body Illusion (Lenggenhager et al., 2007) and
the Enfacement Illusion (Tsakiris, 2008) that rely on multisensory (e.g. visuo-​tactile) stim-
ulation to induce transient yet striking changes in body ownership, body identification,
and self-​face recognition, several studies have focused on the potential impact of cardiac
interoceptive signals for these dimensions of self-​awareness. Early studies focused on cor-
relational designs to show that levels of explicit interoceptive accuracy were negatively
correlated with the strength of the illusory experience of alterations in self-​awareness
such that individuals with lower interoceptive accuracy tend to experience a stronger
RHI (Schauder et al., 2015; Tsakiris et al., 2011). These studies highlighted the dynamic
balance between exteroceptive and interoceptive influences for self-​awareness. The obser-
vation that participants with lower interoceptive accuracy experienced a stronger illusory
sense of body ownership in the Rubber Hand Illusion suggested that in the absence of
accurate interoceptive representations one’s model of self is predominantly exterocep-
tive. Importantly, this finding suggested an antagonism between interoceptive and ex-
teroceptive cues in bodily self-​awareness. More recent studies have used cardiac signals
as inducers of such changes in body ownership. In particular, the substitution of purely
exteroceptive visuo-​tactile stimulation by cardio-​visual stimulation (i.e. a combination of
interoceptive and exteroceptive signals) allowed researchers to probe a more active and
potentially causal role of interoceptive signals for body-​awareness. For example, Suzuki,
36

36 The body as first prior

Garfinkel, Critchley, and colleagues (2013) showed that looking at a virtual hand that
pulsates in synchrony with one’s heartbeat can lead to the same changes in body owner-
ship as the ones reported in the classic version of the Rubber Hand Illusion. Such findings
were soon extended to the Full Body Illusion (Aspell et al., 2013) and the Enfacement
Illusion (Sel, Azevedo, & Tsakiris, 2016). The effect of the synchronicity between cardiac
and visual events that induce the illusory changes in self-​awareness was also reflected
in the attenuated amplitude of the neural response to heartbeats (i.e. Heartbeat Evoked
Potentials; HEP), an effect that also correlated with explicit interoceptive accuracy. More
excitingly, Park and colleagues confirmed the insula as the neural source of HEPs and
demonstrated HEP modulations in the insula during self-​identification using intracranial
electroencephalography (EEG) (Park Bernasconi et al., 2017).
Taken together, these findings suggest that if exteroceptive influences highlight the
malleability of body awareness, interoceptive signals seem to serve the stability of body
awareness in response to exteroceptive stimulation, reflecting a psychological conse-
quence of the biologically necessary function of homeostasis. Therefore, this attempt to
unify the two sides of embodiment concerns equally the physiological and psycholog-
ical basis of selfhood. At the physiological level, it is interoceptive autonomic signaling
that ensures the stability (homeostasis) of the organism. At the psychological level, as
visceral information can be in some sense described as inherently self-​related (unlike ex-
teroceptive sensations, which are externally caused), both predictive and non-​predictive
accounts suggest that the feed-​forward integration of visceral with exteroceptive infor-
mation provides some inherent selfhood and affect to conscious experiences. In other
words, the high expected precision of visceral signals make them ideal for providing a
continuous estimate of “self-​stability.”

2.4.2  The metacognitive self-​model and visceral precision

The notion of a more global or emergent concept emphasizing the role of the hierar-
chical brain in iteratively predicting its own neural activity emerges naturally from con-
sidering the predictive brain as a whole, rather than as a single hierarchy or collection
of hierarchies (auditory, visceral, etc.). Although cortical hierarchies are often depicted
as a “pyramid,” with one executive homunculus sitting atop relatively few inputs, recent
advances in human tractography, functional connectivity, and monkey tracing studies
challenge the plausibility of such a view. Instead, increasing evidence suggests that the
global topology of brain connectivity is best described as a kind of centrifugal hierarchy
(Figure 2.2).
From this global or bird’s-​eye view of the predictive brain, it is clear that a variety of
models can be spun to explain how visceral sensations are propagated throughout the
hierarchy. For example, in the model architecture depicted earlier, visceral sensations
are treated as equal with other sensorimotor hierarchies. In this case “selfhood” emerges
solely from the metacognitive nature of the deep hierarchy; that is, it is the global ex-
pectation of precision that embeds an inherently self-​related notion in information pro-
cessing. To put this another way, consider that the global self-​model is essentially the
 37

Self-models and interoceptive predictions 37

The Global Predictive Hierarchy

Motor
Expected-Precision
‘shadow hierarchy’
Olfaction Self-Model
Audition Multisensory

Unisensory

Gustation

Vision

Somatosensation

Proprioception Canonical Microcircuit

Prediction
Interoception

Prediction Error

Figure 2.2  The Dynamic Self-​Model. By depicting the brain as a “centrifugal” hierarchy, the
self-​model is revealed. At the “periphery” of the brain we find high-​speed, sensory-​motor
hierarchies with a relatively high modularity and functional specialization. At the next level,
multisensory cortices combine prediction errors from divergent sensory-​viscero-​motor hierarchies
with descending “self-​predictions” originating in the centermost layer. At the “center” of the
hierarchy sits the “rich club,” a network of hubs integrating multi-​sensory prediction errors with
global “self-​predictions” oscillating at narrative timescales. The metastable interaction of these
layers is further controlled by a neuromodulatory “shadow hierarchy” which estimates and
optimizes expected precision from the most global (self-​related) predictions down to the most
local level of cortical processing.

brain’s prediction of its own reliability. In this sense, global “experience predictions” (i.e.
that I am this kind of agent, with these kinds of beliefs, who will experience some stimulus
in a particular way) are inherently self-​related and constrain lower-​level brain dynamics.
However, there are numerous reasons to think that visceral sensations may enjoy some
privileged status within the global hierarchy. Both neuroanatomically and functionally
speaking, primary viscero-​sensory and viscero-​motor control centers such as the insular
and cingulate cortices sit firmly within the “middle” integration layer and exhibit greater
multisensory sensitivity than other “primary” sensory cortices. Further, these areas are
38

38 The body as first prior

rich in neuromodulatory neurons thought to encode expected precision, and visceral

sensations themselves may have an a priori hyper-​precision. Thus, rather than being a
mere “arm” of the predictive hierarchy, the visceral brain may in fact form the backbone
of the “shadow hierarchy” responsible for mediating perceptual and affective salience via
precision optimization. In line with this hypothesis, recent studies have highlighted the
far-​reaching effects of the visceral brain on a global self-​model by considering the role of
visceral precision on metacognition.
The metacognitive ability to monitor our own experiences, actions, and beliefs and
to form an accurate feeling of confidence is a crucial part of cognition in general and
decision-​making in particular. If our subjective beliefs become very divorced from reality,
then poor decisions are sure to follow. Indeed, if perception itself is a kind of “Bayesian
hallucination,” then overly precise expectations may be doubly so; I may perceive friendly
faces as hostile, or systematically undervalue my own competence, if my predictions are
too precise. In general, there is an obvious overlap between the statistical notion of preci-
sion, and the metacognitive notion of confidence, but how do the two relate? Are we more
confident when our brain is very precise? At what hierarchical level (i.e. self, integrative,
or sensorimotor) does precision matter most for our subjective self-​awareness? Most im-
portantly for the current issue, how do visceral signals come into this picture?
Until now, these questions have been pursued primarily from the perspective of signal
detection theory (SDT). Here, metacognition and confidence are presumed to arise
from the feed-​forward monitoring of decision-​relevant information only. That is to say,
according to the SDT model, all that really matters for subjective confidence is the (ex-
teroceptive) evidence upon which my decision itself is based. Yet this view of the brain
as a feed-​forward device is clearly in contradiction to the predictive processing views
explained in sections 2.2 and 2.3, where both feedback and multisensory integration are
pervasive features of self-​experience. Why then should our metacognition be so rationally
secluded from the homeostatic visceral function of the body?
In contrast to this view, we argue that metacognition arises generally from the inter-
action of the global self-​model, with more task-​or performance-​specific modules lower
in the hierarchy. In this sense, global expectations of self-​precision are a kind of met-
acognitive model of how accurate I expect to be on an upcoming trial, given both my
visceral and exteroceptive precision. Put another way: if I expect to be very fatigued on
an upcoming test, then I will also expect to be less confident on that material even if it is
relatively easy. The metacognitive self-​model thus emerges from the deepest levels of the
cortical hierarchy, and enables the brain to both estimate and control its own self-​efficacy
in the face of an ever-​changing environment. In this sense, descending expected precision
signals are not only a crucial aspect of selfhood but also of our ability to monitor our per-
formance and change our behavior when it is no longer adaptive.
In support of this view, it has recently been demonstrated that subjective exterocep-
tive confidence depends both on sensory precision, and on visceral surprise itself. To
demonstrate this, Allen and colleagues utilized a novel task in which participants viewed
moving dots on a screen and judged whether their average motion was to the left or right,
 39

An after-thought: The ontogenetic origins of interoceptive precisions 39

and also rated their confidence (Allen et  al., 2016). Unbeknownst to the participants,
on every trial an unconscious neutral or disgust-​related cue was presented using careful
masking techniques. By carefully controlling the difficulty of the dot judgment, this study
demonstrated that a sudden increase in interoceptive prediction error can reverse the
impact of sensory precision, resulting in higher confidence for noisy trials and lower
confidence for precise trials. In a complementary study, Hauser and colleagues (2017a)
followed up on these results by administering the beta-​blocking drug propranolol, which
inhibits the activity of noradrenaline both in the central and autonomic nervous system,
crucial for interoceptive arousal. Interestingly, this study found blocking noradrenaline
actually improved the correspondence of perceptual accuracy and confidence, suggesting
that blocking signals from the body renders metacognition “less embodied” and more
concerned solely with exteroceptive evidence. In this sense, altering bodily signals is likely
to influence a variety of biases in both health and disordered decision-​making. This may
have profound implications for how we understand and treat psychiatric disorders such
as obsessive–​compulsive disorder (OCD), anxiety, and psychosis, where aberrant visceral
precision may result in overly precise (or imprecise) metacognitive self-​model (Bliksted
et al., 2017)), locking the patient into an unrealistically uncertain world.
Ultimately, the emerging understanding of a metacognitive self-​model integrating in-
teroceptive and exteroceptive sensations is still in its infancy. It remains to be seen, for ex-
ample, the extent to which visceral signals influence perceptual or value-​based decisions
in a variety of different domains. Nevertheless, these sorts of findings are exactly what one
would expect if visceral sensations are afforded higher a priori precision, and they suggest
that in many domains our explicit, conscious self-​awareness may be biased in subtle ways
by the ongoing activity and predictability of the visceral body.

2.5  An after-​thought: The ontogenetic origins

of interoceptive precisions
In section 2.4, we described some key properties of domain-​specific and more global
self-​models. In the future, it will be crucial to use a model-​based approach in order to
map how the varieties of self models (e.g. global, multisensory, metacognitive) interact
with one another and to pit competing hypotheses about how visceral signals contribute
to these models (e.g. by enhancing expected precision or prediction error) against one
another. This will not only greatly accelerate the computational neuroscience of embod-
iment but also provide valuable tools to ask important questions, for instance, about the
ontogenetic development of self-​models the lifespan, as well as about the role of vis-
ceral signals for social awareness—​the awareness of other people in relation to the self.
It is precisely these ontogenetic and more social aspects of interoception that remain
poorly understood. A large body of developmental research on the development of self-​
awareness (Lewis, 2001; Rochat, 2003) documents the importance of multisensory inte-
gration processes that will enable infants to match their mirror reflection with their own
body and eventually to master delayed self-​recognition, indicative of a more diachronic
40

40 The body as first prior

self-​representation by the third year. In addition, an intersubjective approach suggests

that infants learn self/​other differentiation through social interactions of mutual attention
(Reddy, 2003)  where they detect the caregiver’s contingent response to their emotion
expression (see also the social-​biofeedback model; Gergely & Watson, 1999). More re-
cently, intersubjective approaches on the development of self-​awareness have been ex-
tended to interoception (Fotopoulou & Tsakiris, 2017), whereby the development of the
visceral and emotional circuit depends on a caregiver–​infant relationship (Fox et  al.,
2007; Rinaman, 2009), often conceptualized as one of homeostatic regulation (Rinaman,
2009). Human infants are born immature, and they do not have the ability to perform by
themselves the actions needed for addressing their internal sensations and needs. From
eating and drinking to thermoregulation and even sleep, their bodily regulation depends
largely on caregivers. Furthermore, the first months post-​partum are characterized by rel-
ative instability of key cardiovascular variables (e.g. Heart Rate Variability (HRV), Vagal
Tone) that become moderately stable by the end of the first year (Fox et al., 2007), due to
physiological maturation (Fracasso et al., 1994). Importantly, their levels depend on care-
giving (McLaughlin et  al., 2015)  such as parent–​infant contingency during interaction
(Feldman, 2007; Feldman et  al., 2011), and are predictive of self-​regulation abilities in
three-​year-​old children (Fracasso et al., 1994).
From the point of view of interoceptive awareness, progress in this area has been
hindered by the lack of appropriate experimental methods that would allow to measure
interoceptive sensitivity in early infancy. Recently, Maister, Tang, and Tsakiris have
tackled this issue by developing the iBEAT task, a non-​verbal behavioral task that meas-
ures infants’ sensitivity to their own heartbeat. During the task, five-​month-​old infants
were shown an animated character that either moved in synchrony with their own heart-
beat or out of synchrony with their heartbeat. Using this sequential looking time task,
it was observed that infants spent longer more time looking at the character that was
moving out of synchrony than the one moving in synchrony, suggesting that even at
this early age infants are, at least implicitly, sensitive to their own interoceptive signals.
Moreover, the infants’ performance in the iBEAT task revealed interesting individual
differences. Infants who displayed greater interoceptive sensitivity in the iBEAT task
also had larger HEP amplitudes in a subsequent task when infants viewed short video
clips of facial expressions, a finding that conforms to HEP measures in adults (Pollatos &
Schandry, 2004). However, the question of what processes can account for such individual
differences remains unanswered.
Methodological advances such as the development of the iBEAT task will enable the
field to study the ontogenetic origins and longitudinal development of interoception as the
infant enters an intersubjective world of carer–​infant embodied and affective interactions.
These interactions are thought to be critical for providing the means by which the infant
will learn to relate specific homeostatic needs (e.g. pain) and their behavioral expres-
sion (e.g. crying) to contingent allostatic responses from the carer (e.g. soothing rather
than feeding; see Fotopoulou & Tsakiris, 2017). If these coupled intersubjective embodied
iterations turn out to be critical for developing a sense of one’s self inside her body, as well
 41

Conclusion 41

as interoceptive inferences with higher precision and better metacognition, we will then
be in a position to test empirically the causal and crucial role that interoception plays
for development and its importance for mentalizing one’s own bodily states and those
of others (Fotopoulou & Tsakiris, 2017; Shah, Catmur, & Bird, 2017). In particular, this
approach could be integrated with new methods for studying metacognitive processes in
infants (Goupil &  Kouider, 2016), revealing how changes in visceral sensitivity contribute
to the emergence of accurate self-​monitoring and advance our understanding of develop-
mental disorders such as autism (Brewer et al., 2015; Quattrocki & Friston, 2014), eating
(Badoud & Tsakiris, 2017; see also the Chapter  9 in the present volume) and affective
disorders (see the Chapter 8 in the present volume).

2.6 Conclusion
One can distinguish among different kinds of “self models” within the predictive mind.
The exteroceptive and interoceptive models, in a more feed-​forward sense, integrate the
action of a visceral hierarchy with ascending exteroceptive signals to provide the basic
embodied context for the self. The second, by representing confidence itself, enforces
systematic interactions between the different exteroceptive and interoceptive hierarchies
(see also Garfinkel et al., 2015). Collectively, both are likely to contribute uniquely to our
embodied self-​experience. Considering the a priori high precision of interoception, in
favour of which we argued in this chapter, the feed-​forward multisensory integration of
prediction errors may be a crucial mechanism for binding together conscious contents
in an embodied way. The high priority of visceral sensations may provide a naturally
reliable “anchor” on which to lodge a more permanent feeling of bodily self. However,
these processes are themselves likely to be cognitive impenetrable, in the sense that they
occur below the metacognitive self-​inferences required for conscious access. Thus, the
top-​down representation of both visceral and exteroceptive confidence (or precision) is a
likely precursor for explicit learning in these domains, and may be a critical target for on-
togenetic and cultural development of social norms, expected behaviors, and ultimately
our empathic understanding of other minds as embodied agents.
The hypothesis that interoception plays an important role in cognitive and emotional
processes is well established in the fields of experimental psychology, cognitive neuro-
science, and clinical psychology and psychiatry. The proposal we put forward about the
psychological role of interoception, namely that interoception serves the stability of self-​
awareness, because of the primacy of a priori high precision of visceral channels, is in
line with FEP-​inspired models of self-​awareness. This role is informed by the homeo-
static principles that keep the organism within certain ranges of physiological function
in adaptive response to external changes. At the same time, one should be aware of the
long-​lasting dichotomies in our disciplines that have fractionalized the self. FEP-​inspired
models of the bodily self can, in principle, go beyond such dichotomies to integrate the
influential traditions that until now have been kept separate:  the awareness of the self
from the outside and from within. Rather than focusing on the apparent differences of
these sources of information for the self, our field should consider their antagonism from
42

42 The body as first prior

a predictive coding perspective that can explain their integration, and how their relation-
ship is reflected in the balance between stability and adaptation. In reference to classic
embodied cognition approaches, which in certain ways depict a surface body, it was about
time, as the growing body of evidence reviewed in this chapter shows, to ground the self
to a non-​hollow body with a dimension of visceral depth. This step may pave the way for
new investigations that can fulfil a long-​standing aim of psychology to link the physiolog-
ical to the psychological self.

Acknowledgments
MA would like to thank Francesca Fardo and Matthew Apps for productive discussions
on this work. MA was supported by a Wellcome grant 100227 (MA). The Wellcome Trust
Centre for Neuroimaging is supported by core funding from Wellcome (091593/​Z/​10/​Z).
MT is supported by the European Research Council (ERC-​2016-​CoG-​724537) under the
FP7 and the NOMIS Foundation Distinguished Scientist Award.

References
Ainley, V., Apps, M. A. J., Fotopoulou, A., and Tsakiris, M. (2016). “Bodily precision”: A predictive
coding account of individual differences in interoceptive accuracy. Philosophical Transactions of the
Royal Society B, 371, 20160003.
Allen, M., Frank, D., Schwarzkopf, D. S., Fardo, F., Winston, J. S., Hauser, T. U., et al. (2016).
Unexpected arousal modulates the influence of sensory noise on confidence. eLife, 5, e18103.
Allen, M. and Friston, K. J. (2016). From cognitivism to autopoiesis: Towards a computational
framework for the embodied mind. Synthese,  1–​24.
Apps, M. A. J. and Tsakiris, M. (2014). The free-​energy self: A predictive coding account of self-​
recognition. Neuroscience and Biobehavioral Reviews, 41,  85–​97.
Aspell, J. E., Heydrich, L., Marillier, G., Lavanchy, T., Herbelin, B., and Blanke, O. (2013). Turning
body and self inside out: Visualized heartbeats alter bodily self-​consciousness and tactile perception.
Psychological Science, 24, 2445–​53.
Badoud, D. and Tsakiris, M. (2017). From the body’s viscera to the body’s image: Is there a link between
interoception and body image concerns? Neuroscience and Biobehavioral Reviews, 77, 237–​46.
<https://​doi.org/​10.1016/​j.neubiorev.2017.03.017>
Barrett, L. F. and Simmons, W. K. (2015). Interoceptive predictions in the brain. Nature Reviews
Neuroscience, 16, 419–​29.
Blefari, M. L., Martuzzi, R., Salomon, R., Bello-​Ruiz, J., Herbelin, B., Serino, A., et al. (2017). Bilateral
Rolandic operculum processing underlying heartbeat awareness reflects changes in bodily self-​
consciousness. European Journal of Neuroscience, 45(10), 1300–​12. <https://​doi.org/​10.1111/​
ejn.13567>
Bliksted, V., Samuelsen, E., Sandberg, K., Bibby, B. M., and Overgaard, M. S. (2017). Discriminating
between first-​and second-​order cognition in first-​episode paranoid schizophrenia. Cognitive
Neuropsychiatry, 22, 95–​107.
Botvinick, M. and Cohen, J. (1998). Rubber hand feels touch that eyes see. Nature, 391(February), 756.
<https://​doi.org/​10.1038/​35784>
Bruineberg, J., Kiverstein, J., and Rietveld, E. (2016). The anticipating brain is not a scientist: The free-​
energy principle from an ecological-​enactive perspective. Synthese,  1–​28.
 43

References 43

Brewer, R., Happé, F., Cook, R., and Bird, G. (2015). Commentary on “Autism, oxytocin and
interoception”: Alexithymia, not autism spectrum disorders, is the consequence of interoceptive
failure. Neuroscience and Biobehavioral Reviews, 56, 348–​53. <https://​doi.org/​10.1016/​
j.neubiorev.2015.07.006>
Bruineberg, J. and Rietveld, E. (2014). Self-​organization, free energy minimization, and optimal grip on
a field of affordances. Frontiers in Human Neuroscience, 8, 599.
Bruineberg, J., Kiverstein, J., and Rietveld, E. (2016). The anticipating brain is not a scientist: The free-​
energy principle from an ecological-​enactive perspective. Synthese, 1–​28. <https://​doi.org/​10.1007/​
s11229-​016-​1239-​1>
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science.
Behavioral and Brain Sciences, 36, 181–​204.
Damasio, A. R. (1999). The Feeling of What Happens: Body and Emotion in the Making of Consciousness.
New York, NY: Harcourt Brace.
Fardo, F., Auksztulewicz, R., Allen, M., Dietz, M. J., Roepstorff, A., and Friston, K. J. (2017).
Expectation violation and attention to pain jointly modulate neural gain in somatosensory cortex.
NeuroImage, 153, 109–​21.
Faull, O., Hayen, A., and Pattinson, K. (2017). Breathlessness and the body: Neuroimaging
evidence for the inferential leap. bioRxiv:117408. <https://www.biorxiv.org/content/biorxiv/
early/2017/03/24/117408.full.pdf>
Feldman, H. and Friston, K. J. (2010). Attention, uncertainty, and free-​energy. Frontiers in Human
Neuroscience, 4, 215
Feldman, R. (2007). Infant biological foundations synchrony and developmental outcomes. Current
Directions in Psychological Science, 16(6),  340–​5.
Feldman, R., Magori-​Cohen, R., Galili, G., Singer, M., and Louzoun, Y. (2011). Mother and infant
coordinate heart rhythms through episodes of interaction synchrony. Infant Behavior and
Development, 34(4),  569–​77
Filippetti, M. L. and Tsakiris, M. (2017). Heartfelt embodiment: Changes in body-​ownership and self-​
identification produce distinct changes in interoceptive accuracy. Cognition, 159,  1–​10.
Fotopoulou, A. (2013). Beyond the reward principle: Consciousness as precision seeking.
Neuropsychoanalysis, 15,  33–​8.
Fotopoulou, A. and Tsakiris, M. (2017). Mentalizing homeostasis: The social origins of interoceptive
inference. Neuropsychoanalysis, 19(1),  3–​28.
Fox, N. A., Schmidt, L. A., Henderson, H. A., and Marshall, P. J. (2007). Developmental
psychophysiology: Conceptual and methodological issues. Handbook of Psychophysiology, (1995),
453–​81.
Fracasso, M. P., Porges, S. W., Lamb, M. E., and Rosenberg, A. A. (1994). Cardiac activity in infancy:
Reliability and stability of individual differences. Infant Behavior & Development, 17(3), 277–​84.
Friston, K. (2005). A theory of cortical responses. Philosophical Transactions of the Royal Society B, 360,
815–​36.
Friston, K. (2009). The free-​energy principle: A rough guide to the brain? Trends in Cognitive Sciences,
13, 293–​301.
Friston, K. (2010). The free-​energy principle: A unified brain theory? Nature Reviews Neuroscience, 11
127–​38.
Friston, K. (2013). Life as we know it. Journal of the Royal Society Interface, 10, 20130475.
Friston, K. and Frith, C. (2015). A duet for one. Consciousness and Cognition, 36, 390–​405
Friston, K., Rigoli, F., Ognibene, D., Mathys, C., Fitzgerald, T., and Pezzulo, G. (2015). Active
inference and epistemic value. Cognitive Neuroscience, 6, 187–​214.
4

44 The body as first prior

Gallagher, S. and Zahavi, D. (2013). The Phenomenological Mind. London: Routledge.

Gallagher, S. and Allen, M. (2016). Active inference, enactivism and the hermeneutics of social
cognition. Synthese,  1–​22.
Garfinkel, S. N., Seth, A. K., Barrett, A. B., Suzuki, K., and Critchley, H. D. (2015). Knowing your own
heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology,
104,  65–​74.
Gergely, G. & Watson, J. S. (1999). Early social-emotional development: Contingency perception and
the social biofeedback model. In: P. Rochat (ed.), Early Social Cognition. Hillsdale, NJ: Erlbaum,
pp. 101–36.
Goupil, L. & Kouider, S. (2016). Behavioral and neural indices of metacognitive sensitivity in preverbal
infants. Current biology, 26(22), 3038–45.
Hauser, T. U., Fiore, V. G., Moutoussis, M., and Dolan, R.J. (2016). Computational psychiatry of
ADHD: Neural gain impairments across marrian levels of analysis. Trends in Neuroscience,
39,  63–​73.
Hauser, T. U., Allen, M., Purg, N., Moutoussis, M., Rees, G., and Dolan, R.J. (2017a). Noradrenaline
blockade specifically enhances metacognitive performance. eLife, 6, e24901.
Hauser, T. U., Allen, M., Rees G, and Dolan, R. J. (2017b). Metacognitive impairments extend
perceptual decision making weaknesses in compulsivity. Scientific Reports, 7, 6614.
Heidegger, M. (1996). Being and Time: A Translation of Sein und Zeit. New York, NY: SUNY Press.
Hohwy, J. (2013). The Predictive Mind. Oxford: Oxford University Press.
Hohwy, J. (2014). The neural organ explains the mind. In: T. Metzinger and J. M. Windt
(eds), Open MIND. Frankfurt am Main: MIND Group. <https://open-mind.net/papers/
the-neural-organ-explains-the-mind>
Joffily, M. and Coricelli, G. (2013). Emotional valence and the free-​energy principle. PLoS
Computational Biology, 9, e1003094.
Kanai, R., Komura, Y., Shipp, S., and Friston, K. (2015). Cerebral hierarchies: Predictive processing,
precision and the pulvinar. Philosophical Transactions of the Royal Society B, 370, 20140169.
Koster-​Hale, J. and Saxe, R. (2013). Theory of mind: A neural prediction problem. Neuron, 79, 836–​48.
Lenggenhager, B., Tadi, T., Metzinger, T., and Blanke, O. (2007). Video ergo sum: Manipulating bodily
self-​consciousness. Science, 317(5841), 1096–​9. <https://​doi.org/​10.1126/​science.1143439>
Lewis, M. (2011). The origins and uses of self-​awareness or the mental representation of me.
Consciousness and Cognition, 20(1),  120–​9.
McLaughlin, K. A., Sheridan, M. A., Tibu, F., Fox, N. A., Zeanah, C. H., and Nelson, C. A. (2015).
Causal effects of the early caregiving environment on development of stress response systems in
children. Proceedings of the National Academy of Sciences of the USA, 112(18), 5637–​42. <https://​doi.
org/​10.1073/​pnas.1423363112>
Meessen, J., Mainz, V., Gauggel, S., Volz-​Sidiropoulou, E., Sütterlin, S., and Forkmann, T. (2016). The
relationship between interoception and metacognition: A pilot study. Journal of Psychophysiology,
30(2),  76–​86.
Merleau-​Ponty, M. (1996). Phenomenology of Perception. New Delhi: Motilal Banarsidass Publisher.
Miall, R. C. and Wolpert, D. M. (1996). Forward models for physiological motor control. Neural
Networks, 9, 1265–​79.
Moran, R. J., Campo, P., Symmonds, M., Stephan, K. E., Dolan, R. J., and Friston, K. J. (2013). Free
energy, precision and learning: The role of cholinergic neuromodulation. Journal of Neuroscience,
33, 8227–​36.
 45

References 45

Park, H.-​D., Bernasconi, F., Salomon, R., Tallon-​Baudry, C., Spinelli, L., Seeck, M., et al. (2017).
Neural sources and underlying mechanisms of neural responses to heartbeats, and their role in
bodily self-​consciousness: An intracranial EEG study. Cerebral Cortex, 116,  1–​14.
Pollatos, O. and Schandry, R. (2004). Accuracy of heartbeat perception is reflected in the amplitude of
the heartbeat-​evoked brain potential. Psychophysiology, 41(3), 476–​82.
Quattrocki, E. and Friston, K. (2014). Autism, oxytocin and interoception. Neuroscience and
Biobehavioral Reviews, 47, 410–​30.
Razran, G. (1961). The observable and the inferable conscious in current Soviet psychophysiology:
Interoceptive conditioning, semantic conditioning, and the orienting reflex. Psychological Review,
68, 81–​147.
Reddy, V. (2003). On being the object of attention: Implications for self–​other consciousness. Trends in
Cognitive Sciences, 7(9), 397–​402.
Rinaman, L. and Koehnle, T. J. K. (2009). The development of central visceral circuits. In: M. Blumberg,
J. Freeman, and S. Robinson (eds), Oxford Handbook of Developmental Behavioral Neuroscience.
Oxford: Oxford University Press, pp. 298–​322.
Rochat, P. (2003). Five levels of self-​awareness as they unfold early in life. Consciousness and Cognition,
12(4), 717–​31.
Ronchi, R., Bello-​Ruiz, J., Lukowska, M., Herbelin, B., Cabrilo, I., Schaller, K., et al. (2015). Right
insular damage decreases heartbeat awareness and alters cardio-​visual effects on bodily self-​
consciousness. Neuropsychologia, 70,  11–​20.
Schauder, K. B., Mash, L. E., Bryant, L. K., and Cascio, C. J. (2015). Interoceptive ability and body
awareness in autism spectrum disorder. Journal of Experimental Child Psychology, 131, 193–​200.
Sel, A., Azevedo, R. T., and Tsakiris, M. (2017). Heartfelt self: Cardio-​visual integration affects self-​face
recognition and interoceptive cortical processing. Cerebral Cortex, 27(11), 5144–​55.
Seth, A. K., Suzuki, K., and Critchley, H. D. (2012). An interoceptive predictive coding model of
conscious presence. Frontiers in Psychology, 2, 395.
Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive
Sciences, 17, 565–​73.
Shah, P., Catmur, C., and Bird, G. (2017). From heart to mind: Linking interoception, emotion, and
theory of mind. Cortex, 93,  220–​3.
Sokolov, E. N. (1963). Higher nervous functions: The orienting reflex. Annual Review of Physiology, 25,
545–​80.
Stephan, K. E., Manjaly, Z. M., Mathys, C. D., Weber, L.A.E., Paliwal, S., Gard, T., et al. (2016).
Allostatic self-​efficacy: A metacognitive theory of dyshomeostasis-​induced fatigue and depression.
Frontiers in Human Neuroscience, 10, 550.
Suzuki, K., Garfinkel, S. N., Critchley, H. D., and Seth, A. K. (2013). Multisensory integration across
exteroceptive and interoceptive domains modulates self-​experience in the rubber-​hand illusion.
Neuropsychologia, 51, 2909–​17.
Tsakiris, M. (2008). Looking for myself: Current multisensory input alters self-​face recognition. PLoS
One, 3(12), e4040.
Tsakiris, M. (2010). My body in the brain: A neurocognitive model of body-​ownership.
Neuropsychologia, 48(3), 703–​12.
Tsakiris, M., Jimenez, A. T.-​, and Costantini, M. (2011). Just a heartbeat away from one’s body:
Interoceptive sensitivity predicts malleability of body-​representations. Proceedings of the Royal
Society B, 278(1717), 2470–​6.
46

Chapter 3

Interoceptive signals, brain dynamics,

and subjectivity
Mariana Babo-​Rebelo and Catherine Tallon-​Baudry

3.1  The embodied self: Visceral inputs as

self-​specifying signals

3.1.1  The self is rooted in the body

The question of the bodily basis of the self has long been debated by philosophers. Descartes
(1641/​1989) had a dualistic point of view, where the self is purely mental and thus de-
tached from the body. However, his approach fails to explain the interaction between the
mental and the physical domains. For instance, remembering a vivid, scary event from
our past increases our heart rate. To solve this issue, Spinoza (1677/​2005) proposed that
mind and body form one unique entity so that any mental event is accompanied by bodily
changes. Therefore, the self, as the subject of mental activity, has a bodily basis. William
James (1890/​1931) further elaborated the concept of self. He defined a hierarchy of the
self where the bodily self is at the bottom and underlies more cognitive and higher-​level
forms of self, such as the social self or the spiritual self. For him, “our entire feeling of spir-
itual activity . . . is really a feeling of bodily activities” (Vol. 1, pp. 301–​2, 1931). Every epi-
sode of our lives is accompanied by particular bodily feelings that remain associated with
the memories we retain from those episodes. These bodily feelings give the memories the
“warmth” necessary for memories to be felt as our own and a sense of continuity of the
self along time as being one and the same.
More recently, the idea of the bodily basis of the self has been translated into biological
terms, notably by Antonio Damasio (1994, 1999). Damasio defines the first level of self,
the proto-​self, as representing the unconscious monitoring of the moment by moment
state of the body. This proto-​self is non-​conscious, and corresponds to a mechanism
shared by all living organisms. When an object (a face, a melody, etc.) interacts with the
organism, it modifies the organism’s state, thereby modifying the proto-​self. These new
maps representing the interaction between the organism and the object can become con-
scious and generate the core-​self. Because a multiplicity of objects is constantly interacting
with the organism, the core-​self is constantly generated and continuous in time. The au-
tobiographical self is then composed of the collection of experiences of the core-​self. The
autobiographical self is extended in time and it places the subject at a given point in their
 47

The embodied self 47

personal history, with a past that constitutes their identity, and a perspective of the future.
In this hierarchical model of the self, each level depends on the lower level, and therefore
all depend on bodily mechanisms. For Damasio, the body ensures the stability of the
self over time; that is, this feeling that we stay the same person throughout our lifetime.
Because the range of internal bodily states compatible with life is actually limited, bodily
representations are stable and so is the self.
While those proposals have been influential, they have also been met with criticism.
In particular, on both James and Damasio’s proposals, selfhood is rooted in changes in
bodily states. For James, the “feeling of bodily activities” constitutes the core definition of
the self, while for Damasio the proto-​self is generated by the brain response to changes in
the organism internal state. This is reminiscent of the James–​Lange theory of emotions,
where physiological changes generate the feeling of an emotion. This theory has been
criticized on several grounds, but the most relevant point is that altering bodily states
does not necessarily induce an emotion. Similarly, altering bodily states does not alter the
self. For instance, the self is not deeply altered when stimulating the viscera by eating. The
proposal we outline aims at overcoming this criticism. Briefly, we propose that organs
endowed with pacemaker properties, such as the heart or the stomach, function as con-
tinuous sources of signals, sending the message to the central nervous system that a body
is there—​whatever the bodily state is. This information would then be used at the central
level to generate an egocentric reference frame, from which first-​person perspective can
arise (Park & Tallon-​Baudry, 2014; Tallon-​Baudry et al., 2017). First-​person perspective
can be viewed as a basic building block necessary to all aspects of selfhood. First-​person
perspective is intrinsic to the simplest, pre-​reflective aspect of the self that defines the
subject who acts, perceives, or feels. In more complex aspects of the self, such as the au-
tobiographical or social self, the self includes also a reflective component where the self is
the object of introspection (Christoff et al., 2011; Gallagher, 2012; Legrand & Ruby, 2009;
Zahavi, 2005).
In the first part of the chapter, we develop the idea that the neural monitoring of visceral
signals could constitute the ground for selfhood. Then, we describe cardiac and gastric
pacemaker properties as well as known anatomical pathways from the viscera to the cen-
tral nervous system. In the third part of the chapter, we review the recent experimental
evidence directly relating neural responses to heartbeats to different facets of the self—​the
bodily self, the self who thinks, and the self who perceives, and finally we discuss the hy-
pothesis in light of these results.

3.1.2  Theneural monitoring of visceral signals as the ground

for the self
3.1.2.1  The importance of visceral electrical signals from the heart and
gastrointestinal tract
In the search for a bodily basis of selfhood, emphasis has mostly been placed on the role
of sensory and motor signals (from the skin, limbs, joints, vestibular system) in agency
(Frith, Blakemore, & Wolpert, 2000) or bodily self-​consciousness (Blanke & Metzinger,
48

48 Interoceptive signals, brain dynamics, and subjectivity

2009) studies, leading to the proposal that the bodily self results from multi-​sensory in-
tegration (Blanke, Slater, & Serino, 2015). While sensory and motor signals certainly play
a role, they are mostly emitted in response to a particular stimulus or as a feedback from a
motor action, whereas first-​person perspective should be continuously defined. The case
of locked-​in patients, who are fully paralyzed but nevertheless conscious, also shows that
the self can exist despite massive disruption of proprioceptive signals (Tononi & Koch,
2008). Conversely, asomatognosic patients cannot sense their somatosensory body but
are aware of their visceral functions, such as breathing, the heart beating, or digestion.
Such a patient described the strangeness of her bodily self but her first-​person perspective
remained intact; she was able to say “I” (Damasio, 1994). From this clinical case, Damasio
hypothesized that “some body representations may be of greater value than others to
ground the mind, namely, those that pertain to the organism’s interior, specifically to the
viscera and internal milieu” (2003, p. 193).
The heart and the gastrointestinal tract have a very distinctive property: they contain
pacemaker cells that continuously and autonomously generate an electrical signal. This
is well known for the heart, and results in cardiac contraction, at a rate of about one
beat per second. The gastrointestinal tract is lined with a specific pacemaker cell type,
the interstitial cells of Cajal (Sanders, Koh, & Ward, 2006), that intrinsically generate a
slow electrical rhythm at rates varying between 3 cycles per minute in the stomach to 12
cycles per minute in the duodenum. This slow electrical rhythm controls the frequency
of contraction of the gastrointestinal tract smooth muscle, but is generated at all times,
even when fasting (Bozler, 1945). These signals, continuously emitted by the heart and
gastrointestinal tract, are relayed up to the central nervous system, as will be detailed in
section 3.2.1. The brain is constantly monitoring visceral signals which most of the time
are processed implicitly and remain unnoticed. Visceral signals can sometimes become
conscious to indicate a large deviation from homeostasis in order to trigger a protec-
tive behavior. However, such events remain quite rare, whereas the self is continuously
present. It thus seems more likely that the self is rooted in the automatic, unconscious
monitoring of visceral inputs, rather than in sparse, transient, conscious events.

3.1.2.2  Proposal for the biological implementation of the self

Our proposal is that the integration of the various visceral signals in the brain generates a
body-​centered reference frame (Park & Tallon-​Baudry, 2014; Tallon-​Baudry et al., 2017).
Ascending visceral signals would be in this sense self-​specifying because they define the
self at the biological level. This reference frame would then be used by the brain to tag
mental processes as being subjective or self-​related.
One could argue that there is no need for a mechanism tagging my mental processes
as being mine, since they happen in my brain and not someone else’s. However, not all
mental processes have “mineness.” The vast literature on unconscious processing shows
numerous examples where mental processes remain unconscious, without first-​person
perspective. Thus, not all neural processes are by default subjective or self-​related, and a
tagging mechanism might be useful to define “mineness” (Tallon-​Baudry et al., 2017). In
 49

Visceral signals 49

this view, first-​person perspective would not be defined at the neural level by activity in
specific brain regions but would be based on the neural monitoring of visceral afferents,
which might take place in a quite extended network as detailed in the following.

3.2  Visceral signals: Pathways and mechanisms

3.2.1  Pathways from the viscera to the brain
Which neuronal pathways carry visceral sensory information? What is the nature of
visceral information and how is it represented? Surprisingly, we know much less about
visceroception than about other senses such as vision or audition. One reason is probably
that it has attracted less attention than classical senses but it also seems that the organiza-
tion of visceral pathways is more complex, with more parallel pathways and more cortical
targets, than for exteroceptive senses. However, the existence of multiple pathways also
speaks about the importance of visceral representations in the brain.

3.2.1.1  From the periphery to the brainstem

Information about heart contraction is transduced by baroreceptors, neurons sensing
changes in pressure following blood ejection. This system is mostly dedicated to the
regulation of arterial blood pressure. Baroreceptors are located in the heart walls, the
aortic arch and the carotid bodies and, depending on their location, send ascending
projections either through the vagus nerve or glossopharyngeal nerve and discharge at
different moments of the cardiac cycle (Klabunde, 2012). Both vagal and glossopharyn-
geal pathways converge in the Nucleus Tractus Solitarius (NTS), where they also con-
verge with spinal afferents (Nosaka et al., 1995). Neurons responding to the stimulation
of baroreceptor afferents have been observed in the rat NTS (Nosaka et al., 1995). It is
worth underlining that the baroreceptor route is not the only pathway that can be em-
ployed. Somatosensory receptors are likely to convey heartbeat-​related information as
well, although detailed mechanisms have not yet been elucidated. Indeed, the chest is
innervated by somatosensory afferents that might respond to the impact of the heart on
the chest, at each heartbeat (Khalsa et al., 2009). Finally, it has recently been shown that,
at least in vitro, vascular events can alter directly firing rate (Kim et al., 2016), suggesting
the possibility of a direct route from vessels to neurons that would bypass the peripheral
mechanotransduction of cardiac contraction.
Regarding the stomach, the interstitial cells of Cajal, which generate the slow gastric
basal rhythm, connect not only to smooth muscles but also to sensory neurons (Powley
& Phillips, 2011). Sensory information from the gastrointestinal tract is, as cardiac infor-
mation, relayed up to the NTS (Furness et al., 2013).

3.2.1.2  Central targets of visceral inputs

Visceral inputs reach a number of cortical targets. The insular cortex receives vagal inputs
in a number of mammalian species (Saper, 2002)  and appears viscerotopically organ-
ized, with cardiopulmonary inputs targeting the posterior (granular) insula (Cechetto
& Saper, 1987). While the insula is sometimes presented as the primary visceral cortex,
50

50 Interoceptive signals, brain dynamics, and subjectivity

visceral information is also mapped in a number of other regions, that should not be
neglected: the ventral anterior cingulate cortex (Vogt, Pandya, & Rosene, 1987) and so-
matosensory cortices SI and SII (Kern et al., 2013; Korn, Wendt, & Albe-​Fessard, 1966;
Newman, 1962). More recently, Dum, Levinthal, and Strick (Dum et al., 2009) showed
that motor cingulate regions receive inputs from the spinothalamic tract.
While anatomical tracing or stimulation studies show the existence of multiple cor-
tical targets of visceral inputs, much emphasis has been placed on the role of the insula,
following the proposal by Craig (Craig, 2002, 2009). In his view, interoceptive informa-
tion would be represented in the posterior insula, and then integrated with other sen-
sory and cognitive processes in a gradient progressing toward the anterior insula, thereby
generating self-​awareness. Following this influential theory, a number of brain imaging
studies used a region-​of-​interest approach targeting the insula, but did not test other can-
didate regions. There is no doubt that the right anterior insula is important to explic-
itly report interoceptive feelings, as elegantly demonstrated by (Critchley et  al., 2004).
However, as will be detailed in section 3.3.4, the role of the insula appears more limited
in the absence of heartbeat-​related tasks (Babo-​Rebelo, Richter, & Tallon-​Baudry, 2016a;
Babo-​Rebelo et al., 2016b; Park et al., 2016; Park et al., 2014).

3.2.1.3  An impact of visceral inputs on brain dynamics?

Visceral information can reach a number of different cortical targets but does it have an
impact on brain dynamics? Evidence accumulates to suggest a strong cross-​talk between
brain dynamics and physiological signals, such as skin conductance levels (Fan et al., 2012;
James, Henderson, & Macefield, 2013; Nagai et al., 2004), respiration fluctuations (Yuan
et al., 2013), or heart rate (de Munck et al., 2008). Classically, physiological measures have
been considered as reflecting brain outputs. The most well-​known case is fluctuations in
pupil diameter, which are thought to be directly reflecting noradrenergic release by the
locus coeruleus (Aston-​Jones & Cohen, 2005; Joshi et al., 2016), or electrodermal activity,
reflecting sympathetic output (Beissner et al., 2013). More generally, there is a strong bias
to think that physiological measures reflect mostly top-​down autonomic control, from
the brain to the peripheral organs. In human brain imaging, physiological signals are
mostly considered as a source of noise that needs to be regressed out of the data (Birn,
2012; Glover, Li, & Ress, 2000; Shmueli et al., 2007).
The importance of a bottom-​up influence of visceral inputs on brain dynamics should
nonetheless be considered. Indeed, up to 80% of the fibers of the vagus nerve are as-
cending (Agostoni et  al., 1957), suggesting that the bottom-​up transfer of information
is prioritized relative to top-​down transfer of information. Besides, while there is no
doubt that pupil diameter is under strong central control, other organs generate their
own activity. As mentioned earlier, the stomach intrinsically generates a slow (~0.05 Hz)
electrical wave, known as the gastric basal rhythm (Sanders et al., 2006). A recent study
investigated a potential link between the gastric basal rhythm with resting-​state brain
dynamics (Richter et  al., 2017). There it was shown that the phase of the gastric basal
rhythm constrains the amplitude of alpha oscillations in the right anterior insula and
 51

Neural responses to heartbeats encode the self 51

occipito-​parietal regions, with the gastric slow rhythm accounting for 8% of the vari-
ance of alpha rhythm amplitude fluctuations. Importantly, the coupling was mostly due
to an ascending influence from the stomach on the brain. So-​called intrinsic brain activity
might thus be—​at least partly—​constrained by visceral pacemakers.

3.2.2  Neural responses to heartbeats

To demonstrate a link between the self and ascending visceral information, one needs
a measure of visceral processing. We will thus focus here on the heart for which such a
measure is available: the heartbeat-​evoked response.
Neural responses to heartbeats are computed by averaging brain activity, recorded with
electroencephalography (EEG) or magnetoencephalography (MEG), locked to heartbeats
(Figure 3.1). Heartbeat-​evoked responses (HERs) were firstly explored in the 1980s by the
group of Schandry (Schandry, Sparrer, & Weitkunat, 1986).
HERs appear to be generated in an impressive range of brain areas. Initial source lo-
calization (Pollatos, Kirsch, & Schandry, 2005) found dipole locations compatible with
the anterior cingulate cortex, the medial frontal gyrus, the right insula, and the left
somatosensory cortex. A more recent EEG study found sources in posterior cingulate
cortex/​supplementary motor area (Park et  al., 2016). MEG recordings, which have a
better spatial resolution than EEG, have shown the involvement of the right inferior
parietal lobule, ventral posterior cingulate cortex, and ventromedial prefrontal cortex
in the generation of HERs (Babo-​Rebelo et al., 2016a; Park et al., 2014). HERs were also
found in the insula, using a region of interest analysis of high-​density EEG (Couto et al.,
2015) or MEG data (Babo-​Rebelo et al., 2016a), as well as in intracranial EEG (Babo-​
Rebelo et al., 2016b; Park et al., 2017). Intracranial electro-​corticographic (EcoG) grid
recordings also revealed prominent HERs in the primary somatosensory cortex (Kern
et  al., 2013). Many studies reported different effect latencies, ranging from 200–​650
ms after the R-​peak (Kern et al. 2013, for a review), potentially reflecting the different
latencies at which baroreceptors can fire in a cardiac cycle, or the existence of different
cortical relays.

3.3  Neural responses to heartbeats encode the self

Neural responses to heartbeats are a measure of the neural monitoring of cardiac activity.
We now review recent evidence showing a link between these responses and three facets
of the self: bodily awareness, the self in spontaneous thoughts, and the self as the subject
of visual perception.

3.3.1  Neural responses to heartbeats and bodily awareness

Direct evidence for a link between HERs and bodily awareness comes from a study on full-​
body illusions (Park et al., 2016), induced by stroking the participant’s back (Lenggenhager
et al., 2007) in synchrony with the strokes applied to a virtual body displayed in front
of the participant. If the stroking is asynchronous, participants do not experience the
52

52 Interoceptive signals, brain dynamics, and subjectivity

Figure 3.1  What is a Heartbeat-​Evoked Response (HER)? Example from real data. Example

of raw data in a single participant, with an electrocardiogram (ECG, top row), recorded here
between the right clavicle and left abdomen, magnetoencephalographic (MEG) raw data (middle
raw), and after correction of the cardiac field artefact (bottom row), from the MEG sensor
highlighted in white on the topographical maps. The ECG shows major deflections or R-​waves
(corresponding to ventricular depolarization and the beginning of myocardial contraction),
followed by T-​waves (corresponding to ventricular repolarization). The interval between R peaks
is quite variable in healthy participants. The R and T waves are also detected on MEG raw data,
a phenomenon called the cardiac-​field artefact (light-​grey shaded areas). One thus has to be
cautious when searching for brain responses to heartbeats: MEG (or EEG) data includes cardiac
signals. The cardiac field artefact typically affects sensors located in temporal regions, as shown
on the topographic map. We combine two solutions to avoid including the cardiac field artefact
in our analysis. First, we analyze only the portion of MEG data corresponding to an electrical
“silence” of the heart, when the heart relaxes and refills, after the end of the T wave (dark-​grey
shaded area). We also correct the MEG data from the cardiac-​field artefact (bottom row), using
independent component analysis (ICA). The heartbeat-​evoked response (HER) is then computed
as the average of the data segments in the dark-​grey shaded area.

illusion. HER amplitude was found to differ depending on whether the stroking was syn-
chronous or asynchronous and correlated with the strength of the experienced illusion
(Park et al., 2016). This effect was associated with the posterior cingulate cortex/​supple-
mentary motor area (PCC-​SMA) (Figure 3.2a). Similar findings were obtained with a
modified version of the enfacement illusion (Sel, Azevedo, & Tsakiris, 2016). Participants
watched a morphed face which decreased in luminance synchronously or asynchronously
with their heartbeats. This manipulation induced a bias in a subsequent self-​recognition
task where the other person’s face was included to a greater extent in the representation of
 53

(a) Baseline Stroking Pause Questionnaire Full-body illusion

Questionnaire

PCC-SMA

(b) I’ll go shopping I’m thirsty

He’s coming
tomorrow “I”
vPrec-PCC

“I” scale
– + – + – +
“Me” scale
External Me External Me External Me vmPFC

Skip Skip Skip

“Me”

Visual perception

(c)
vACC/vmPFC vACC/vmPFC

Fixation Warning Stimulus Delay Response:

Seen or Unseen?

HERs in
Hits vs Misses rIPL

Figure 3.2 Summary of paradigms and results involving Heartbeat-​Evoked Responses (HERs).

(a)—​HERs covary with the bodily self (Park et al., 2016). Participants were stroked on the back,
while visualizing their own back being stroked in front of them, in a virtual reality setting. When
the stroking was synchronous with the visual input, participants may experience the illusion of
being located closer to the virtual body. The amplitude of HERs recorded during the stroking
phase varied with illusion strength. This effect was located in the posterior cingulate cortex-​
supplementary motor area (PCC-​SMA). (b)—​HERs encode the self-​relatedness of spontaneous
thoughts (Babo-​Rebelo et al., 2016a; Babo-​Rebelo et al., 2016b). Participants had to mind-​
wander while fixating. They were interrupted at random intervals and had to evaluate the self-​
relatedness of the interrupted thought on the “I” scale (how much were they involved as the
subject or agent in the thought) and on the “Me” scale (was the thought referring to themselves,
the “Me,” or to something external). Participants could skip the answer if they did not know
how to rate the thought. HERs in the ventral precuneus-​posterior cingulate cortex (vPrec-​PCC)
encoded the involvement of the “I” in the thought, while HERs in the ventromedial prefrontal
cortex (vmPFC) independently encoded the “Me” dimension. (c)—​HERs predict conscious visual
perception (Park et al., 2014). Participants fixated a central point and were presented with a faint
visual stimulus (annulus grating at threshold of visibility). They had to report if they had seen the
stimulus or not. The amplitude of HERs before stimulus onset predicted whether the stimulus was
going to be subsequently seen or not. This effect was located in the ventral anterior cingulate
cortex/​ventromedial prefrontal cortex (vACC/​vmPFC) and in the right inferior parietal lobule (rIPL).
Note that differential neural responses to heartbeats were observed in the default network during
fixation (spontaneous thoughts, panel b; perception at threshold before stimulus onset, panel c)
while in the virtual body experiment (panel a), differential responses to heartbeats are maximal in
premotor regions.
54

54 Interoceptive signals, brain dynamics, and subjectivity

one’s own face after synchronous stimulation. HER amplitude differed between synchro-
nous and asynchronous conditions, in centro-​parietal electrodes (Sel et al., 2016).
Changes in bodily awareness are thus associated with changes in HER amplitude. Those
results speak in favor of the hypothesis that the neural monitoring of visceral signals is
linked to self-​awareness. Still, it remains to be determined whether neural responses to
heartbeats are related to other facets of the self, beyond bodily awareness.

3.3.2  Neural
responses to heartbeats and the self
in spontaneous thoughts
The most basic level of self is the self as the subject who acts, feels, and perceives from a
first-​person perspective. Actions, feelings, and perceptions can then be reflected upon.
This corresponds to a second level of self, where the self is the object of introspection.
This distinction between the self as the subject and the self as the object is reminiscent
of phenomenological theories distinguishing a pre-​reflective from a reflective form of
self (Christoff et al., 2011; Gallagher, 2012; Legrand and Ruby, 2009; Zahavi, 2005). This
idea can be operationalized in spontaneous thoughts, by distinguishing the “I” from the
“Me.” The “I” is always present, but can be more or less engaged. The “I” is engaged when
one is the agent, the first-​person subject of the thought, for example in a thought like “I
will go to the supermarket,” as opposed to “It’s raining” or “He has to go to the super-
market.” In contrast, the “Me” corresponds to a reflective dimension of thoughts, when
one is thinking about oneself, about one’s feelings or bodily state for instance, as in “I am
tired” (Babo-​Rebelo et al., 2016a; Babo-​Rebelo et al., 2016b). Could HERs correspond to
the biological implementation of the “I”, of the “Me”, or both?
This question was assessed in a thought-​sampling paradigm, where participants
were asked to mind-​wander while fixating on a screen (Babo-​Rebelo et  al., 2016a).
They were interrupted at random intervals and had to report whether they were en-
gaged as the “I” and as the “Me” during the interrupted thought. HER amplitude was
shown to correlate with the self-​relatedness of thoughts, but in different regions for
the “I” and the “Me.” The amplitude of HERs in the ventral precuneus and posterior
cingulate regions (vPrc-​PCC) co-​varied with the engagement of the “I” in the ongoing
thought (Figure 3.2b). The “Me” dimension was associated with HERs in the ventro-
medial prefrontal cortex (vmPFC) (Figure 3.2b). Although the “I” and the “Me” were
often combined in a given thought, we could demonstrate that the association between
the “I” and HERs in vPrc-​PCC was orthogonal to the association between the “Me”
and HERs in vmPFC.
These MEG results were then replicated with intracranial recordings, by showing a trial-​
by-​trial parametrical modulation of HERs along with the level of involvement of each
self-​dimension in each sampled thought (Babo-​Rebelo et  al., 2016b). The main effects
were found in midline regions of the default-​network, but a region of interest analysis
additionally revealed that HERs in the right anterior insula were modulated by the degree
of engagement of the “I” in thoughts. HERs thus encode the self-​relatedness of thoughts,
along both the “I” and the “Me” self-​dimensions, but in distinct regions.
 5

Neural responses to heartbeats encode the self 55

3.3.3  Neuralresponses to heartbeats and the self as the

subject of visual perception
The existence of a first-​person subject underlying any conscious perception has been
overlooked in cognitive neuroscience (Tallon-​Baudry et al., 2017). Indeed, the hallmark
of conscious vision is that the participant reports his/​her perception: for conscious per-
ception to exist, there must be a subject of experience (Park and Tallon-​Baudry, 2014;
Tallon-​Baudry et al., 2017). In the experiment by Park and colleagues (Park et al., 2014),
participants had to report the presence or absence of near-​threshold gratings. HER ampli-
tude in the ventral anterior cingulate/​ventromedial prefrontal cortex (vACC/​vmPFC) and
the right posterior inferior parietal lobule (rIPL), before stimulus presentation, predicted
whether the visual stimulus was going to be consciously perceived or not (Figure 3.2c)
(Park et al., 2014). This effect corresponded to changes in visual sensitivity (d’) and not
to changes in the criterion of response, suggesting that information carried by neural
responses to heartbeats is used as sensory evidence in the seen/​unseen decision.
Differential HERs predicting conscious vision can here be interpreted as corresponding
to fluctuations in the engagement of the self as the experiencing subject, the one who
wonders whether or not he/​she has seen the stimulus. Note that only weak, subthreshold
differences in HERs could be detected in the right insula.

3.3.4  Reconsidering the role of the insula

In the three experiments mentioned above (Babo-​Rebelo et al., 2016a; Babo-​Rebelo et al.,
2016b; Park et al., 2016, 2014) that directly relate interoceptive processing with the “I,” the
“Me,” or the bodily self, the insula did not appear to be the key player. This raises questions
about the “hypothesis that the anterior insula engenders human awareness” (Craig, 2009).
There is no doubt that the insula is involved in cardiac processing, and indeed neural
responses to heartbeats were observed in the posterior (Park et al., 2017) and anterior
(Babo-​Rebelo et al., 2016b) insula using iEEG. fMRI studies further suggest that (a) the
insula is involved in cardiac-​visual integration (Blefari et al., 2017; Ronchi et al., 2015;
Salomon et al., 2016), (b) the insula is activated when explicitly paying attention to intero-
ceptive signals (Critchley et al., 2004; Farb, Segal, & Anderson, 2013), and (c) the anterior
insula is involved in self-​related bodily processing (Blanke et al., 2014; Brass & Haggard,
2010; Heydrich & Blanke, 2013; Karnath & Baier, 2010). When combined, these three sets
of results could suggest a link between cardiac processing and the interoceptive/​bodily
self in the insula.
However, there are several reasons why the role of the insula might have been
overestimated. First, it is not clear whether there is a true anatomical overlap between
interoception and self-​related processes. This is all the more important that the insula
is an extended structure, with numerous subdivisions (Deen, Pitskel, & Pelphrey, 2011).
Second, even if there were an overlap between self-​related activations and interoception,
it does not necessarily follow that there is a true interaction between interoception
and the self. For instance, it could be that paying attention to interoceptive signals and
paying attention to some aspects of the self both require saliency estimation, a pro-
cess implemented in the insula for a variety of tasks, independently from the self or
56

56 Interoceptive signals, brain dynamics, and subjectivity

interoceptive processing (Menon & Uddin, 2010). Last, the reverse inference suggesting
that insular activity is a signature of interoceptive processing might be flawed, as any re-
verse inference (Poldrack, 2011). The insula is activated by a wealth of different processes,
not necessarily related to the self nor to interoception (Chang et  al., 2013; Duncan &
Owen, 2000).

3.3.5  Changesin neural responses to heartbeats in the absence

of physiological changes
Theories rooting the self in the body have focused on the role of bodily changes (see section
3.1.1). Importantly, none of the results reported earlier were associated with differences in
cardiorespiratory parameters (heart rate, heart rate variability, peripheral blood pressure,
respiratory patterns), suggesting that cardio-​respiratory state did not change. This is in
line with the hypothesis that bodily information exists and impacts neural processing
even in the absence of bodily changes. In addition, several arousal measures, such as the
pupil diameter, interbeat interval, heart rate variability, electrodermal activity, or alpha
power did not differ between conditions, ruling out the possibility that HER differences
reflected distinct global arousal states. Altogether, this suggests that the amplitude changes
of HERs encoding the self are related to neural, rather than bodily, fluctuations, as fur-
ther discussed in section 3.4. A heartbeat could be considered as a physically constant
stimulus generating neural responses whose amplitude varies from one stimulus to the
next, a phenomenon which is well known for exteroception. For instance, the repeated
presentation of a physically identical visual stimulus can elicit brain responses of different
amplitudes, depending on neural and cognitive parameters.
The fact that neural responses to heartbeats can vary independently from changes in
cardiac parameters seems incompatible with interoceptive prediction error accounts of
selfhood (Seth, 2013; Seth & Friston, 2016). In this framework, cardiac inputs should vary
and deviate from the corresponding homeostatic prediction, thereby generating predic-
tion error signals at the origin of a sense of self.

3.4  Towards a biological implementation of the self?

We have presented evidence that neural responses to heartbeats encode changes in bodily
awareness, encode self-​relatedness in spontaneous thoughts, and predict the seen or un-
seen fate of a visual stimulus. Such results are in line with the hypothesis that neural
responses to heartbeats constitute a marker of the first-​person perspective inherent to
self-​consciousness, bodily consciousness, and perceptual consciousness. However, the
results show a correlation between HER amplitude and the self, while the hypothesis that
neural responses to heartbeats play a causal role in the generation of first-​person perspec-
tive remains to be directly tested. It is thus worth examining the different interpretations
of the correlations between HER amplitude and the self.
A first interpretation of the changes in amplitude of HERs is that HERs are a non-​
specific marker of enhanced brain activity. Any region being particularly active at a given
 57

Conclusion 57

moment would then show changes in HER amplitude. In this view, if two conditions
induce different levels of activity in a given region, responses to heartbeats in this region
differ as well. The fact that differential HERs are found in regions which are expected to
be differently activated, such as default-​network regions during the resting state or passive
fixation (Babo-​Rebelo et al., 2016a; Babo-​Rebelo et al., 2016b; Park et al., 2014), or bodily/​
motor regions during full-​body illusions (Park et al., 2016), could support this interpre-
tation. However, in the aforementioned three experiments, HER differences could not be
trivially explained by different levels of neural activity (e.g. alpha power, slow fluctuations)
differing between conditions. Still, not all features of brain activity (e.g. high-​frequency
bands) were investigated. Ruling out definitively the hypothesis that neural responses
to heartbeats are a non-​specific marker of cortical reactivity will prove difficult, since it
implies measuring all aspects of cortical reactivity and demonstrating the absence of a
difference, which is a notoriously difficult task.
A second hypothesis is that ascending visceral signals themselves differ between
conditions, leading in turn to differential neural responses to heartbeats. In this hypoth­
esis, information about the self can be found directly in cardiac data. For instance, car-
diac afferent signals would determine whether or not a stimulus is likely to be seen, or
whether a spontaneous thought is self-​related. However, there were no differences in the
measured cardiac parameters (electrocardiogram, heart rate, heart rate variability, and
blood pressure), but more subtle differences may have gone unnoticed. Other parameters,
such as volume of blood ejected at each heartbeat, were not measured and might vary.
The third hypothesis is that self-​related information is present in brain responses to
ascending visceral signals, not in the visceral signals themselves. Cardiac or gastric as-
cending signals would function as constant sources of signal, indicating that a body is
there but not necessarily conveying more specific information. Note that other organs or
systems might also contribute, but the pacemaker properties of the heart and the gastro-
intestinal tract make them particularly relevant. These ascending signals could interact
with ongoing brain activity, anchoring some aspects of brain activity in a body-​centered
reference frame and hence implementing first-​person perspective. The mechanisms un-
derlying this interaction remain to be investigated. Cardiac signals would contribute to a
body-​centered reference frame, which would be used by the brain to anchor thoughts or
percepts to the self (Park & Tallon-​Baudry, 2014). The amplitude of these neural responses
would then be a marker of the self-​relatedness of neural activity in this region. Note that
given that visceral signals target a large number of cortical areas, this mechanism could
take place in many different brain regions, wherever a self versus non-​self distinction is
relevant for the task at play.

3.5 Conclusion
Building on a long tradition relating the self to bodily signals, we have proposed to revise
this hypothesis and to focus on visceral inputs. Such signals display interesting charac-
teristics, notably the fact that they are continuously emitted, in the absence of bodily
58

58 Interoceptive signals, brain dynamics, and subjectivity

movement but also in the absence of bodily state changes. Visceral inputs would indicate
to the brain the mere presence of a body, and could thus act as continuous self-​specifying
signals. Neural responses to heartbeats, which can be measured non-​invasively in humans,
offer a valuable tool to experimentally probe the link between the neural monitoring of
visceral inputs and the self. Recent results show correlations between neural responses to
heartbeats and the self, not only, as initially hypothesized, when the self is implicit (the
self as the agent, the self experiencing a visual input) but also in situations where the self is
explicit or reflective (bodily awareness, thinking about oneself). These results are compat-
ible with our proposal that the integration of visceral signals generates a subject-​centered
reference frame underlying the different facets of the self. However, the results so far are
only correlational, and moving to causation will prove an important, but difficult, step. In
addition, some questions remain open, for instance how neural responses to heartbeats
interact with stimulus or thought content.

References
Agostoni, E., Chinnock, J. E., Daly, M. D. B., and Murray, J. G. (1957). Functional and histological
studies of the vagus nerve and its branches to the heart, lungs and abdominal viscera in the cat.
Journal of Physiology, 135(1), 182–​205. <https://​doi.org/​10.1113/​jphysiol.1957.sp005703>
Aston-​Jones, G. and Cohen, J. D. (2005). An integrative theory of locus coeruleus-​norepinephrine
function: Adaptive gain and optimal performance. Annual Review of Neuroscience, 28(1), 403–​50.
<https://​doi.org/​10.1146/​annurev.neuro.28.061604.135709>
Babo-​Rebelo, M., Richter, C., and Tallon-​Baudry, C. (2016a). Neural responses to heartbeats in the
default network encode the self in spontaneous thoughts. Journal of Neuroscience, 36(30), 7829–​40.
<https://​doi.org/​https://​doi.org/​10.1523/​JNEUROSCI.0262-​16.2016>
Babo-​Rebelo, M., Wolpert, N., Adam, C., Hasboun, D., and Tallon-​Baudry, C. (2016b). Is the cardiac
monitoring function related to the self in both the default network and right anterior insula?
Philosophical Transactions of the Royal Society B, 371(1708), 1–​13. <https://​doi.org/​10.1098/​
rstb.2016.0004>
Beissner, F., Meissner, K., Bär, K.-​J., and Napadow, V. (2013). The autonomic brain: An activation
likelihood estimation meta-​analysis for central processing of autonomic function. Journal of
Neuroscience, 33(25), 10503–​11. <https://​doi.org/​10.1523/​JNEUROSCI.1103-​13.2013>
Birn, R. M. (2012). The role of physiological noise in resting-​state functional connectivity. NeuroImage,
62(2), 864–​70. <https://​doi.org/​10.1016/​j.neuroimage.2012.01.016>
Blanke, O. and Metzinger, T. (2009). Full-​body illusions and minimal phenomenal selfhood. Trends in
Cognitive Sciences, 13(1), 7–​13. <https://​doi.org/​10.1016/​j.tics.2008.10.003>
Blanke, O., Pozeg, P., Hara, M., Heydrich, L., Serino, A., Yamamoto, A., et al. (2014). Neurological and
robot-​controlled induction of an apparition. Current Biology, 24, 2681–​6. <https://​doi.org/​10.1016/​
j.cub.2014.09.049>
Blanke, O., Slater, M., and Serino, A. (2015). Behavioral, neural, and computational principles of bodily
self-​consciousness. Neuron, 88(1), 145–​66. <https://​doi.org/​10.1016/​j.neuron.2015.09.029>
Blefari, M. L., Martuzzi, R., Salomon, R., Bello-​Ruiz, J., Herbelin, B., Serino, A., et al. (2017). Bilateral
rolandic operculum processing underlying heartbeat awareness reflects changes in bodily self-​
consciousness. European Journal of Neuroscience, 45, 1300–​12. <https://​doi.org/​10.1111/​ijlh.12426>
Bozler, E. (1945). The action potentials of the stomach. American Journal of Physiology, 144, 693–​700.
<http://​ajplegacy.physiology.org/​content/​144/​5/​693.article-​info>
 59

References 59

Brass, M. and Haggard, P. (2010). The hidden side of intentional action: The role of the anterior insular
cortex. Brain Structure and Function, 214, 603–​10. <https://​doi.org/​10.1007/​s00429-​010-​0269-​6>
Cechetto, D. F. and Saper, C. B. (1987). Evidence for a viscerotopic sensory representation in the cortex
and thalamus in the rat. Journal of Comparative Neurology, 262(1), 27–​45. <https://​doi.org/​10.1002/​
cne.902620104>
Chang, L. J., Yarkoni, T., Khaw, M. W., and Sanfey, A. G. (2013). Decoding the role of the insula in
human cognition: Functional parcellation and large-​scale reverse inference. Cerebral Cortex, 23(3),
739–​49. <https://​doi.org/​10.1093/​cercor/​bhs065>
Christoff, K., Cosmelli, D., Legrand, D., and Thompson, E. (2011). Specifying the self for
cognitive neuroscience. Trends in Cognitive Sciences, 15(3), 104–​12. <https://​doi.org/​10.1016/​
j.tics.2011.01.001>
Couto, B., Adolfi, F., Velasquez, M., Mesow, M., Feinstein, J., Canales-​Johnson, A., et al. (2015). Heart
evoked potential triggers brain responses to natural affective scenes: A preliminary study. Autonomic
Neuroscience: Basic and Clinical, 193, 132–​7. <https://​doi.org/​10.1016/​j.autneu.2015.06.006>
Craig, A. D. (2002). How do you feel? Interoception: The sense of the physiological condition of the
body. Nature Reviews Neuroscience, 3(8), 655–​66. <https://​doi.org/​10.1038/​nrn894>
Craig, A. D. (2009). How do you feel—​now? The anterior insula and human awareness. Nature Reviews
Neuroscience, 10(1), 59–​70. <https://​doi.org/​10.1038/​nrn2555>
Critchley, H. D., Wiens, S., Rotshtein, P., Ohman, A., and Dolan, R. J. (2004). Neural systems supporting
interoceptive awareness. Nature Neuroscience, 7(2), 189–​95. <https://​doi.org/​10.1038/​nn1176>
Damasio, A. R. (1994). Descartes’ Error. Emotion, Reason, and the Human Brain. New York,
NY: Avon Books.
Damasio, A. R. (1999). The Feeling of what Happens: Body, Emotion and the Making of Consciousness.
New York, NY: Harcourt.
Damasio, A. R. (2003). Looking for Spinoza. Joy, Sorrow, and the Feeling Brain. New York, NY: Harcourt.
de Munck, J. C., Gonçalves, S. I., Faes, T. J. C., Kuijer, J. P. A., Pouwels, P. J. W., Heethaar, R. M.,
et al. (2008). A study of the brain’s resting state based on alpha band power, heart rate and fMRI.
NeuroImage, 42(1), 112–​21. <https://​doi.org/​10.1016/​j.neuroimage.2008.04.244>
Deen, B., Pitskel, N. B., and Pelphrey, K. a. (2011). Three systems of insular functional connectivity
identified with cluster analysis. Cerebral Cortex, 21(7), 1498–​506. <https://​doi.org/​10.1093/​cercor/​
bhq186>
Descartes, R. (1989, orig. pub. 1641). Discourse on Method and Meditations. (trans. J. Veitch). New York,
NY: Amherst.
Dum, R. P., Levinthal, D. J., and Strick, P. L. (2009). The spinothalamic system targets motor and
sensory areas in the cerebral cortex of monkeys. Journal of Neuroscience, 29(45), 14223–​35. <https://​
doi.org/​10.1523/​JNEUROSCI.3398-​09.2009>
Duncan, J. and Owen, A. M. (2000). Common regions of the human frontal lobe recruited by
diverse cognitive demands. Trends in Neurosciences, 23(10), 475–​83. <https://​doi.org/​10.1016/​
S0166-​2236(00)01633-​7>
Fan, J., Xu, P., Van Dam, N. T., Eilam-​Stock, T., Gu, X., Luo, Y., et al. (2012). Spontaneous brain activity
relates to autonomic arousal. Journal of Neuroscience, 32(33), 11176–​86. <https://​doi.org/​10.1523/​
JNEUROSCI.1172-​12.2012>
Farb, N. a S., Segal, Z. V, and Anderson, A. K. (2013). Mindfulness meditation training alters cortical
representations of interoceptive attention. Social Cognitive and Affective Neuroscience, 8(1), 15–​26.
<https://​doi.org/​10.1093/​scan/​nss066>
Frith, C. D., Blakemore, S. J., and Wolpert, D. M. (2000). Abnormalities in the awareness and control of
action. Philosophical Transactions of the Royal Society of London B, 355(1404), 1771–​88. <https://​doi.
org/​10.1098/​rstb.2000.0734>
60

60 Interoceptive signals, brain dynamics, and subjectivity

Furness, J. B., Rivera, L. R., Cho, H.-​J., Bravo, D. M., and Callaghan, B. (2013). The gut as a sensory
organ. Nature Reviews Gastroenterology & Hepatology, 10, 729–​40. <https://​doi.org/​10.1038/​
nrgastro.2013.180>
Gallagher, S. (2012). Phenomenology. London: Palgrave Macmillan.
Glover, G. H., Li, T. Q., and Ress, D. (2000). Image-​based method for retrospective correction of
physiological motion effects in fMRI: RETROICOR. Magnetic Resonance in Medicine, 44(1), 162–​7.
<https://​doi.org/​10.1002/​1522-​2594(200007)44:1<162::AID-​MRM23>3.0.CO;2-​E>
Heydrich, L. and Blanke, O. (2013). Distinct illusory own-​body perceptions caused by damage to
posterior insula and extrastriate cortex. Brain, 136, 790–​803. <https://​doi.org/​10.1093/​brain/​
aws364>
James, C., Henderson, L., and Macefield, V. G. (2013). Real-​time imaging of brain areas involved in the
generation of spontaneous skin sympathetic nerve activity at rest. NeuroImage, 74, 188–​94. <https://​
doi.org/​10.1016/​j.neuroimage.2013.02.030>
James, W. (1931, orig. pub. 1890). The Principles of Psychology. New York, NY: Henry Holt and
Company. <https://​doi.org/​10.1037/​11059-​000
Joshi, S., Li, Y., Kalwani, R. M., and Gold, J. I. (2016). Relationships between pupil diameter and
neuronal activity in the locus coeruleus, colliculi, and cingulate cortex. Neuron, 89(1), 221–​34.
<https://​doi.org/​10.1016/​j.neuron.2015.11.028>
Karnath, H. O. and Baier, B. (2010). Right insula for our sense of limb ownership and self-​awareness of
actions. Brain Structure and Function, 214, 411–​17. <https://​doi.org/​10.1007/​s00429-​010-​0250-​4>
Kern, M., Aertsen, A., Schulze-​Bonhage, A., and Ball, T. (2013). Heart cycle-​related effects on
event-​related potentials, spectral power changes, and connectivity patterns in the human ECoG.
NeuroImage, 81, 178–​90. <https://​doi.org/​10.1016/​j.neuroimage.2013.05.042>
Khalsa, S. S., Rudrauf, D., Feinstein, J. S., and Tranel, D. (2009). The pathways of interoceptive
awareness. Nature Neuroscience, 12(12), 1494–​6. <https://​doi.org/​10.1038/​nn.2411>
Kim, K. J., Diaz, J. R., Iddings, J. A., and Filosa, J. A. (2016). Vasculo-​neuronal coupling: Retrograde
vascular communication to brain neurons. Journal of Neuroscience, 36(50), 12624–​39. <https://​doi.
org/​10.1523/​JNEUROSCI.1300-​16.2016>
Klabunde, R. E. (2012). Cardiovascular Physiology Concepts. Philadelphia, PA: Lippincott, Williams &
Wilkins.
Korn, H., Wendt, R., and Albe-​Fessard, D. (1966). Somatic projection to the orbital cortex of the cat.
Electroencephalography and Clinical Neurophysiology, 21(3), 209–​26. <https://​doi.org/​10.1016/​
0013-​4694(66)90071-​X>
Legrand, D. and Ruby, P. (2009). What is self-​specific? Theoretical investigation and critical review of
neuroimaging results. Psychological Review, 116(1), 252–​82. <https://​doi.org/​10.1037/​a0014172>
Lenggenhager, B., Tadi, T., Metzinger, T., and Blanke, O. (2007). Video ergo sum: Manipulating bodily
self-​consciousness. Science, 317(5841), 1096–​9. <https://​doi.org/​10.1126/​science.1143439>
Menon, V. and Uddin, L. Q. (2010). Saliency, switching, attention and control: A network model
of insula function. Brain Structure and Function, 214(5–​6), 655–​67. <https://​doi.org/​10.1007/​
s00429-​010-​0262-​0>
Nagai, Y., Critchley, H. D., Featherstone, E., Trimble, M. R., and Dolan, R. J. (2004). Activity in
ventromedial prefrontal cortex covaries with sympathetic skin conductance level: A physiological
account of a “default mode” of brain function. NeuroImage, 22(1), 243–​51. <https://​doi.org/​10.1016/​
j.neuroimage.2004.01.019>
Newman, P. P. (1962). Single unit activity in the viscero-​sensory areas of the cerebral cortex. Journal of
Physiology, 160(2), 284–​97. <https://​doi.org/​10.1113/​jphysiol.1962.sp006846>
Nosaka, S., Murase, S., Murata, K., and Inui, K. (1995). “Aortic baroreceptor” neurons in the nucleus
tractus solitarius in rats: Convergence of cardiovascular inputs as revealed by heartbeat-​locked
 61

References 61

activity. Journal of the Autonomic Nervous System, 55(1–​2), 69–​80. <http://​www.ncbi.nlm.nih.gov/​

entrez/​query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_​uids=8690854>
Park, H.-​D. and Tallon-​Baudry, C. (2014). The neural subjective frame: From bodily signals to
perceptual consciousness. Philosophical Transactions of the Royal Society of London B, 369(1641),
20130208. <https://​doi.org/​10.1098/​rstb.2013.0208>
Park, H.-​D., Correia, S., Ducorps, A., and Tallon-​Baudry, C. (2014). Spontaneous fluctuations in
neural responses to heartbeats predict visual detection. Nature Neuroscience, 17(4), 612–​18. <https://​
doi.org/​10.1038/​nn.3671>
Park, H.-​D., Bernasconi, F., Bello-​Ruiz, J., Pfeiffer, C., Salomon, R., and Blanke, O. (2016). Transient
modulations of neural responses to heartbeats covary with bodily self-​consciousness. Journal of
Neuroscience, 36(32), 8453–​60. <https://​doi.org/​10.1523/​JNEUROSCI.0311-​16.2016>
Park, H.-​D., Bernasconi, F., Salomon, R., Tallon-​Baudry, C., Spinelli, L., Seeck, M., et al. (2017).
Neural sources and underlying mechanisms of neural responses to heartbeats, and their role in
bodily self-​consciousness: An intracranial EEG study. Cerebral Cortex, 1–​14. <https://​doi.org/​
10.1093/​cercor/​bhx136>
Poldrack, R. A. (2011). Inferring mental states from neuroimaging data: From reverse inference to
large-​scale decoding. Neuron, 72(5), 692–​7. <https://​doi.org/​10.1016/​j.neuron.2011.11.001>
Pollatos, O., Kirsch, W., and Schandry, R. (2005). Brain structures involved in interoceptive awareness
and cardioafferent signal processing: A dipole source localization study. Human Brain Mapping,
26(1), 54–​64. <https://​doi.org/​10.1002/​hbm.20121>
Powley, T. L. and Phillips, R. J. (2011). Vagal intramuscular array afferents form complexes with
interstitial cells of Cajal in gastrointestinal smooth muscle: Analogues of muscle spindle organs?
Neuroscience, 186, 188–​200. <https://​doi.org/​10.1016/​j.neuroscience.2011.04.036>
Richter, C. G., Babo-​Rebelo, M., Schwartz, D., and Tallon-​Baudry, C. (2017). Phase-​amplitude
coupling at the organism level: The amplitude of spontaneous alpha rhythm fluctuations varies with
the phase of the infra-​slow gastric basal rhythm. NeuroImage, 146, 951–​8. <https://​doi.org/​10.1016/​
j.neuroimage.2016.08.043>
Ronchi, R., Bello-​Ruiz, J., Lukowska, M., Herbelin, B., Cabrilo, I., Schaller, K., et al. (2015).
Right insular damage decreases heartbeat awareness and alters cardio-​visual effects on bodily
self-​consciousness. Neuropsychologia, 70, 11–​20. <https://​doi.org/​10.1016/​j.neuropsycholo
gia.2015.02.010>
Salomon, R., Ronchi, R., Donz, J., Bello-​Ruiz, J., Herbelin, B., Martet, R., et al. (2016). The insula
mediates access to awareness of visual stimuli presented synchronously to the heartbeat. Journal of
Neuroscience, 36(18), 5115–​27. <https://​doi.org/​10.1523/​JNEUROSCI.4262-​15.2016>
Sanders, K. M., Koh, S. D., and Ward, S. M. (2006). Interstitial cells of Cajal as pacemakers in the
gastrointestinal tract. Annual Review of Physiology, 68(1), 307–​43. <https://​doi.org/​10.1146/​annurev.
physiol.68.040504.094718>
Saper, C. B. (2002). The central autonomic nervous system: Conscious visceral perception and
autonomic pattern generation. Annual Review of Neuroscience, 25, 433–​69. <https://​doi.org/​10.1146/​
annurev.neuro.25.032502.111311>
Schandry, R., Sparrer, B., and Weitkunat, R. (1986). From the heart to the brain: A study of heartbeat
contingent scalp potentials. International Journal of Neuroscience, 30(4), 261–​75. <https://​doi.org/​
10.3109/​00207458608985677>
Sel, A., Azevedo, R. T., and Tsakiris, M. (2016). Heartfelt self: Cardio-​visual integration affects self-​face
recognition and interoceptive cortical processing. Cerebral Cortex, 1–​12. <https://​doi.org/​10.1093/​
cercor/​bhw296>
Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive
Sciences, 17(11), 565–​73. <https://​doi.org/​10.1016/​j.tics.2013.09.007>
62

62 Interoceptive signals, brain dynamics, and subjectivity

Seth, A. K. and Friston, K. J. (2016). Active interoceptive inference and the emotional brain.
Philosophical Transactions of the Royal Society B, 371(20160007). <https://​doi.org/​10.1098/​
rstb.2016.0007>
Shmueli, K., van Gelderen, P., de Zwart, J. A., Horovitz, S. G., Fukunaga, M., Jansma, J. M., et al.
(2007). Low-​frequency fluctuations in the cardiac rate as a source of variance in the resting-​
state fMRI BOLD signal. NeuroImage, 38(2), 306–​20. <https://​doi.org/​https://​doi.org/​10.1016/​
j.neuroimage.2007.07.037>
Spinoza, B. de. (2005, orig. pub. 1677). Ethics (trans. E. Curley). London: Penguin Classics.
Tallon-​Baudry, C., Campana, F., Park, H.-​D., and Babo-​Rebelo, M. (2017). The neural monitoring
of visceral inputs, rather than attention, accounts for first-​person perspective in conscious vision.
Cortex, 1–​11. <https://​doi.org/​10.1016/​j.cortex.2017.05.019>
Tononi, G. and Koch, C. (2008). The neural correlates of consciousness: An update. Annals of the
New York Academy of Sciences, 1124, 239–​61. <https://​doi.org/​10.1196/​annals.1440.004>
Vogt, B. A., Pandya, D. N., and Rosene, D. L. (1987). Cingulate cortex of the rhesus monkey: I.
Cytoarchitecture and thalamic afferents. Journal of Comparative Neurology, 262(2), 256–​70. <https://​
doi.org/​10.1002/​cne.902620207>
Yuan, H., Zotev, V., Phillips, R., and Bodurka, J. (2013). Correlated slow fluctuations in respiration,
EEG, and BOLD fMRI. NeuroImage, 79, 81–​93. <https://​doi.org/​10.1016/​j.neuroimage.2013.04.068>
Zahavi, D. (2005). Subjectivity and Selfhood: Investigating the First-​Person Perspective. Cambridge,
MA: MIT Press.
 63

Chapter 4

The embodiment of time: How

interoception shapes the perception
of time
Marc Wittmann and Karin Meissner

4.1  Introduction: From phenomenology to neuroscience

Conscious awareness and subjective time are intricately connected. In phenomenolog-
ical analyses, self-​consciousness and time consciousness cannot be separated. In a phe-
nomenal description, consciousness can be seen as a world that is present, as an island
of presence in the continuous flow of time. This “window of presence” is concerned
with what is happening right now (Metzinger, 2004; Revonsuo, 2006). This description
complements the analysis of time experience by Edmund Husserl (1928) and William
James (1890), both of whom also discerned two complementary aspects of temporality.
Subjective time is described as (a)  a continuous flow and (b)  the feeling of a present
moment. The flow constitutes itself through the experience of passage created by the se-
quence of (a)  an expectation of what is going to happen, (b)  the actual experience of
what is happening right now, and (c) memory of what has happened. Besides the flow of
time, the unity of the present moment (or the feeling of “nowness”) is a basic property of
consciousness, which comprises the qualitative character of subjective experience. Felt
presence is not a duration-​less instant in time but is temporally extended and for which
neural correlates are assumed and investigated (Pöppel, 1997; Varela, 1999; Lloyd, 2012;
Northoff, 2016).
Conscious experience has a first-​person mode of “givenness.” Conscious states are in-
herently given to me as the experiencer, or phenomenal experience is mine (Nagel, 1974;
Metzinger, 2008; Kiverstein, 2009). This quality of “mineness” thus includes a minimal
sense of self within experience. The act of conscious perception includes a basic form
of self-​consciousness (Zahavi, 2005). Moreover, the self as part of present experience is
extended. This temporal extension of the presence stems from temporal properties, past
moments, present awareness, and expectations, all being interwoven in present experi-
ence. Husserl (1928) termed these facets of present awareness as retention, impression,
and protention, respectively, forming the implicit temporal structure of any conscious
experience. The phenomenological connection between “time” and the “self ” can, there-
fore, be expressed as follows: I become aware of what is happening now to me through
64

64 The embodiment of time

what just happened to me and expectations of what might happen to me. The realization
of a self (what happens to me) is created through this tripartite structure of conscious-
ness (Kiverstein, 2009). According to this conceptualization, time consciousness and self-​
consciousness are manifestations of the same underlying process.
The last connection needed in this short phenomenological introduction to our
neuroscientific inquiry on subjective time is the concept of embodiment, the notion that
subjective time emerges through bodily processes. According to the enactive/​embodied
cognition models of subjectivity (Varela, Thompson, & Rosch, 1991/​2016), the phenom-
enal first-​person perspective depends on the physical self. In other words, the mental
self is created by the continuous visceral and proprioceptive input from the body. The
physical self is the functional (bodily) anchor of subjective experience (Metzinger, 2008).
Subjective time emerges only through the existence of the self across time as an enduring
and embodied entity. In his phenomenological analysis, Maurice Merleau-​Ponty (1945)
was most explicit when he declared that every mental act is based on a bodily function.
Accordingly, one has to understand time as subject and the subject as time (“Il faut
comprendre le temps comme sujet et le sujet comme temps,” Merleau-​Ponty, 1945, p. 483);
the physical self and subjective time are inseparable.

4.2  Neural models of time perception

The dimension of time is undoubtedly essential for human cognitive functioning. There
are many temporal levels of organization in perception and action, ranging from the pro-
cessing of milliseconds of duration up to daily rhythms and beyond (Buhusi & Meck,
2005; Wittmann, 2016). Specifically, the feeling of time passing and the estimation of
duration in the seconds and minutes range are fundamental for orientation in the world
and for decision-​making. The neural underpinnings of subjective time are, however,
poorly understood; there is no consensus on how and where time is processed in the brain
(Wittmann & van Wassenhove, 2009).
Regarding functional principles, regardless of questions of neural implementation,
there are several conceptualizations concerning time perception which are summarized
here. For one, memory decay has repeatedly been suggested to govern the experience
of duration (Staddon, 2005). The underlying idea is that memory capacity decreases
over time; therefore, memory decay could function as a clock. An alternative model
of time perception encompassing a memory sub-​function is the dual klepsydra model
(Wackermann & Ehm, 2006), where subjective time is represented by the states of lossy
accumulators. The accumulator receives inflow during the presentation of a stimulus to
be timed (producing the representation of duration), but, at the same time, a continuous
outflow reflects the loss of representation over time. In further conceptualizations, energy
expenditure has been proposed to underlie subjective duration (Mach, 1911; Marchetti,
2009). It has been shown that a greater demand on mental activity while analyzing more
complex or novel situations expands subjective duration. In contrast, experienced repe-
tition leads to higher coding efficiency and to shorter duration estimates (Eagleman &
 65

Neural models of time perception 65

Pariyadath, 2009). Accordingly, the experience of duration would be a function of the

amount of energy expended during the timed interval.
The most influential cognitive model is based on the idea of a pacemaker-​accumulator
clock. A  pacemaker would produce a series of pulses like the ticks of a clock, and the
number of pulses recorded during an interval would lead to experienced duration
(Treisman, 2013; Gibbon, Church, & Meck, 1984). In a variant of that model, an atten-
tional gate is assumed to open only when attention is directed to time. Time units then
enter the accumulator. Accordingly, the more attention is paid to the passage of time,
the longer duration is experienced (Zakay & Block, 1997; Wearden, 2016). This notion
fits well with everyday experience. Time expands when we become aware of it, like in a
waiting situation. Its strong heuristic value probably explains why this model is predomi-
nantly used in psychology to account for empirical results.
Regarding the neuroanatomy and neurophysiology of time perception, several models
have been brought forward to provide an answer on which brain areas and which neural
mechanisms underlie duration judgments. The following summary is by no means ex-
haustive, but provides a rough sketch of the most prominent ideas. Related to the timing
of events ranging from the milliseconds to a few seconds at most, a general mechanism
for different magnitudes, such as time, space, and number, has been proposed in the
right posterior parietal cortex (Bueti & Walsh, 2009), as well as mechanisms related to
neural-​delay conduction pathways in the cerebellum (Ivry et  al., 2002). Alternatively,
it has been suggested that many neural networks have intrinsic temporal-​processing
properties. Accordingly, duration estimation for very brief events lasting less than half a
second are related to time-​dependent neural changes, such as short-​term synaptic plas-
ticity (Buonomano, Bramen, & Khodadadifar, 2009). Subjective duration is thus an emer-
gent property (or processed at a later stage) stemming from modality-​related processes
in the brain (van Wassenhove, 2009). One of the most prominent recent neural models
is the striatal beat-​frequency model, which assumes coincidence detection mechanisms
involving oscillatory signals with various frequencies in cortico-​striatal circuits (Matell &
Meck, 2004). Phases of different cortical oscillators are read out by neurons located in the
striatum and give rise to a unique activation pattern over time that distinguishes different
durations. Neural oscillations of various frequencies have been shown to be associated
with mechanisms, such as coincidence detection (neural phase timing), as well as re-
lated to different cognitive functions, including interval timing: the synchronous timing
of events and temporal expectation (van Wassenhove, 2016). Therefore, an oscillatory
model favouring synchronization and coincidence detection seems a promising avenue
for future research (Kononowicz & van Wassenhove, 2016). Although a specific model
employing oscillatory mechanisms has not received unanimous empirical support, EEG
data show the involvement of neural oscillations in time perception, at least with intervals
of up to a few seconds’ duration (Kononowicz & van Rijn, 2015).
The dopaminergic system plays a key role regarding the involvement of neurotransmitters
in the perception of time. Many studies in animals and humans have shown that dopamine-​
receptor agonists and antagonists provoke a relative overestimation and underestimation of
6

66 The embodiment of time

duration, respectively (Rammsayer, 2009; Coull, Cheng, & Meck, 2011). Complementing
these findings, carriers of genotypic variants modulating D2 receptor density in the stri-
atum were associated with duration-​discrimination variability (but not accuracy) in a
timing task with a stimulus duration of around 500 ms (Wiener, Lohoff, & Coslett, 2011).
In contrast, duration-​discrimination variability with visual stimuli lasting around two
seconds was associated with a polymorphism related to an enzyme modulating synaptic
dopaminergic metabolism in the prefrontal cortex. These outcomes clearly show how
different durations are processed by different neural systems. Other transmitter systems
also influence subjective time. One study using sub-​second visual stimuli showed that
resting-​state GABA concentration correlated with timing accuracy in humans (Terhune
et al., 2014). The neuromodulator oxytocin induces a subjective time dilation for happy
female faces and a time compression for happy male faces in heterosexual male viewers
(Colonnello, Domes, & Heinrichs, 2016), showing how transient physiological states in-
fluence duration estimates. Other studies have revealed the involvement of the seroto-
nergic (5-​HT) transmitter system in the estimation of longer time intervals. For example,
intake of sub-​threshold doses of psilocybin, a hallucinogenic substance and 5-​HT 2A/​
1A receptor agonist, interfered with a duration-​reproduction task with intervals up to 5
seconds (Wackermann et al., 2008). Even though participants did not notice any changes
in their states of consciousness, duration reproduction was less accurate compared to a
placebo condition. Gene polymorphisms associated with the serotonin system, but not
the dopamine system, were associated with duration discrimination of intervals lasting
between 3.2 and 6.4 seconds (Sysoeva, Tonevitsky, & Wackermann, 2010).
Complementing these findings on the duration-​dependent involvement of transmitter
systems, meta-​analyses of neuroimaging data point to at least two distinct neural timing
systems. A  timing system for sub-​second intervals is more strongly governed by sub-
cortical systems in the brain, and a more cognitively controlled system of supra-​second
timing is more strongly related to cortical areas in the brain (Lewis & Miall, 2003;
Wiener, Turkeltaub, & Coslett, 2010). Further neuroimaging evidence supports a dual
system separating intervals below and above two seconds (Morillon, Kell, & Giraud,
2009). Several systems probably function in coordination with each other, and individual
systems become dominant, depending on the time range involved (Petter et al., 2016).
Almost all of these models of human time perception are based on empirical findings
with intervals ranging from milliseconds to a few seconds. These time intervals are def-
initely important for the understanding of individual sensorimotor processing required
when performing sports, playing music, driving a car, or even for our survival as homo
sapiens in interpersonal communication and when hunting and gathering (Wittmann &
Pöppel, 1999). Brain-​based models of human time perception rarely encompass empir-
ical data from longer time intervals in the multiple-​second range. The few existing neu-
roimaging studies employing multiple-​second intervals show that a multitude of areas
in the brain are activated including the basal ganglia and several cortical areas (Hinton
& Meck, 2004; Wittmann et al., 2010). Subjective time as an experience and as related to
everyday decision-​making is crucially dependent upon the perception of the passage of
 67

Attention, affect, and interoception 67

time and duration in the multiple-​second range (Wittmann & Paulus, 2009). The feeling
of boredom, when time drags on and duration expands, or of time pressure, when there
does not seem to be enough time to finish something, are commonly encountered in eve-
ryday experience and strongly influence our decisions (Zakay, 2014). These perceptions
of time are related to the feeling of time passing and most likely cannot be explained by
mechanisms dealing with sensorimotor processes lasting between a fraction of a second
and up to three seconds.

4.3  Body time: Attention, affect, and interoception

A considerable body of evidence indicates an intricate interplay between affective states
and subjective time. Subjects overestimate the duration of emotionally arousing pictures
or short audio files with emotional content presented from several hundred milliseconds
to a few seconds (Droit-​Volet, Brunot, & Niedenthal, 2004; Noulhiane et  al., 2007;
Wackermann et al., 2014). This relative overestimation of duration is typically explained
by the above-​mentioned cognitive model of prospective time perception, the pacemaker-​
accumulator model (Zakay & Block, 1997; Wearden, 2016). Accordingly, increased
physiological arousal leads to a higher pacemaker rate which, in turn, leads to a larger
accumulation of temporal units over a given time interval (Droit-​Volet & Meck, 2007;
Wittmann & Paulus, 2008). Temporal dilation effects of emotionally arousing stimuli
seem to be related to the embodiment of emotions, at least when judging the duration
of presented photos with emotional faces. Participants relatively overestimated the dura-
tion of presented faces when they were free to spontaneously imitate perceived emotions
(Effron et al., 2006). When subjects had a pen in their mouths that prevented the mim-
icking of emotions, the duration of faces was judged to be relatively shorter. The effects
of emotions on duration estimation are also observed with longer time intervals. In a
retrospective duration-​judgment task with three video clips lasting 45 seconds each, a
fear-​inducing horror video led to a relative overestimation compared to a serious docu-
mentary and a funny cartoon, the latter being estimated as having lasted the shortest time
(Pollatos, Laubrock, & Wittmann, 2014). In other words, the highly arousing horror video
elicited time dilation, whereas the cartoon was experienced as “time flies when you are
having fun.” These relative over-​and underestimations were boosted when subjects were
instructed to pay attention to their bodily reactions to the film clips. In a further condi-
tion, when subjects were instructed to focus on physical sensations (interoceptive focus),
they over-​and underestimated the two emotional films even more strongly compared to
the neutral documentary.
The notion that feelings and emotions depend on visceral and somatosensory feed-
back from the peripheral nervous system goes back at least to the theories of James and
Lange in the late nineteenth century. Related modern theories postulate that body sig-
nals are integrated with perceptual, motivational, social, and cognitive information,
leading to the awareness of complex emotional states (Damasio, 1999; Singer, Critchley,
& Preuschoff, 2009). For example, one empirical study showed how those individuals who
were more sensitive in perceiving their heartbeat also felt more aroused while judging
68

68 The embodiment of time

photos with emotional content (Pollatos et al., 2007). The timing of the heartbeat is de-
cisive for generating affective experience. It was shown that fearful faces were detected
more accurately and were judged as more arousing when photos were presented at the
cardiac systole compared to when they were presented at the diastole. This behavior and
experience corresponded with fMRI-​recorded higher amygdala responses (Garfinkel
et al., 2014). Conscious emotional experiences are directly felt in the body, and people
can draw emotion-​specific activation maps of physical sensations (Nummenmaa et al.,
2014). It is argued that perception of these emotion-​triggered and topologically specific
body changes generates consciously felt emotions.
A direct link between the perception of time and physical processes was proposed by
Craig (2009, 2015). Craig suggested that the experience of time is related to emotional
and visceral processes because they share a common underlying neural system, the inter-
oceptive system, including the insular cortex. Through integrating the ascending somatic
signals, the insula would be involved in creating a series of conscious emotional moments
in time. As a consequence, the sense of duration would be created by these successive
moments of self-​realization, essentially formed by information originating within the
body. In this context, it is important to note that the underlying computational cogni-
tive architecture regarding discrete or continuous processing over time is still debated
(Madl et al., 2016). The aforementioned phenomenological insight of the dual embod-
iment of self and time corresponds to a recent conceptualization of time-​awareness in
neuroscience (Wittmann, 2016). Thus, everyday experience becomes understandable. In
situations such as when waiting, where one easily gets frustrated, one attends to time and
is comparably aroused, one is strongly aware of his/​her own emotional and body feelings,
and, as a consequence, subjective duration expands. The insular cortex is an integral part
of the saliency network involved in cognitive control and attentional processes (Menon
& Uddin, 2010). The same brain network is also involved in the functions of body aware-
ness, attentional control and the sense of time. In combining these three functions, one
can argue that “attending to time” is equivalent to “attending to bodily signals.” This at
least is the strong hypothesis.
An fMRI study by Critchley, Wiens, Rotshtein, and Dolan (2004) was probably the
first to link the insular cortex anatomy and activity, interoception, attention, and explicit
timing abilities. Participants took part in an interoceptive task where they had to judge
the timing of their own heartbeats in relation to a series of presented tones. Activation
in the right anterior insula predicted accuracy in the heartbeat-​detection task; that is,
whether the tones were synchronous with the own heartbeat or not. Grey-​matter volume
in that specific region was also correlated with interoceptive timing accuracy, as well as
with subjective ratings of heartbeat awareness. The striking novel information this influ-
ential research report provided is that the right anterior insula is involved in temporally
fine-​tuned conscious sensing of bodily processes. Empirical findings now identify the role
of the insular cortex in temporally integrating the heartbeat with exteroceptive signals
(Salomon et al., 2016). The awareness of visual stimuli depends on when they appear in
relation to the heartbeat. When stimuli occur close to the heartbeat, detection of external
 69

Climbing activation 69

stimuli is less accurate. More empirical research is now confirming the notion that visual
conscious experience is dependent upon neural events locked to the heartbeats (Park
et al., 2014). A recent line of research using fMRI technology and psychophysiological
measurements complements these findings on the timing of sub-​second events: insular-​
cortex activation and direct physiological responses, such as the heart rate and skin con-
ductance levels, are also related to the estimation of duration in the multiple-​second range
(Wittmann et al., 2010; Meissner & Wittmann, 2011).

4.4  Body time: Climbing activation

The initial motivation to use longer time intervals in our research stems from an applied
research agenda. Trait-​like impulsivity, such as that found in many psychiatric disorders
such as borderline personality disorder or drug dependence, is associated with an over-
estimation of duration in the multiple-​second to minute range (Berlin & Rolls, 2004;
Wittmann et al., 2007). In accordance with our embodiment concept of subjective time,
impulsivity can be understood as a strongly felt urge for immediate gratification, which is
generated through the interoceptive system, including insular-​cortex functioning (Turel &
Bechara, 2016). To test individuals later with higher levels of impulsivity, we began to test
healthy young students with a duration-​reproduction task while recording activation via
functional neuroimaging (fMRI) technology. The computerized duration-​reproduction
task became the standard experimental timing task for the following investigations re-
ported here (for details on the methods used, see Wittmann et al., 2010, 2011; Wittmann
& Meissner, 2011; Otten et al., 2015). In the duration-​reproduction task, participants were
instructed to reproduce the duration of tones with intervals of 3, 9, and 18 seconds. In
each trial a 1.2 kHz tone was presented for one of the three durations (encoding phase),
followed by a variable pause after which a second, 2 kHz tone was presented. After the
second tone had started (reproduction phase), subjects had to stop the tone by pressing a
key when they estimated that the duration of the first tone had elapsed. Although subjects
were requested not to count, to prevent any such tendency the participants had to com-
plete a secondary working-​memory task. At the beginning of each trial, four numbers
between zero and nine were presented to be memorized. After the second tone had been
switched off by the subject, one number appeared, and the subjects had to respond by
pressing one of two buttons for “yes” and “no” whether the presented number was one of
the four from the beginning of the trial.
In the first of two fMRI studies (Wittmann et  al., 2010), several areas of activation in
sub-​cortical and cortical areas were detected for the contrasts between the encoding and
reproduction phases and, very similarly, for the two timing phases contrasted with a con-
trol reaction-​time task. In the control task, subjects had to wait the same duration (3, 9,
18 seconds), but had to press a button as fast as possible at the end of the tone. Because
we employed comparably long intervals, we were able to analyze activation patterns in
the identified areas of activation over time. Time-​activity curves showed an increasing
(climbing) activation pattern over time in the left and right posterior insula during the
70

70 The embodiment of time

(a)
Encoding phase 9 s 0.15 tone duration *

ROI activation
L R SMA

0
0 2 4 6 8 10 12 14 16 18 20 22
SMA

–0.15

0.45
X=8
L p Ins, ST
ROI activation

0.3

0.15

R L
0
0 2 4 6 8 10 12 14 16 18 20 22
p Ins 0.45
p Ins
ROI activation

0.3 R p Ins, ST
ST ST

0.15
Z = 14

0
0 2 4 6 8 10 12 14 16 18 20 22
Duration [sec]

* Projected peak of hemodynamic response

Figure 4.1 Brain activity during the 9-​s (4.1a) and 18-​s (4.1b) encoding phase. A sagittal (x = 8)
and an axial slice (z = 14) show significant brain activity (p < 0.01, corrected) in three regions
encompassing a bilateral medial frontal area (SMA), left and right posterior insula (p Ins) as well
as superior temporal cortex (ST) as related to the encoding versus control contrast in the 9-​and
18-​s conditions. Individual time activity curves (set to zero at the onset of the stimulus) show
an inverted u-​shape function in the SMA and climbing brain activity that peaks at the end of
the stimulus (with a delay of around 6 s reflecting the hemodynamic response function) for left
(L) and right (R) p Ins, ST.
 71

Climbing activation 71

(b)
Encoding phase 18 s Tone duration *

ROI activation
0.15
L R SMA

0
SMA 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

–0.15 Duration (sec)

0.15
X=8
ROI activation
0.3
L p Ins
0.15

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
–0.15
R L Duration (sec)

0.35
ROI activation

p Ins R p Ins
p Ins 0.2

0.05
ST, IP
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
–0.1
Duration (sec)
z = 14 0.6
ROI activation

0.45
R ST, IP
0.3
0.15
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Duration (sec)
* Projected peak of hemodynamic response

Figure 4.1  Continued

encoding phase (Figure 4.1). Most notably, this region of activation encompassed the dorsal
posterior insula, which is the primary interoceptive area (Craig, 2015). This neural signature
was interpreted as being related to the accumulation of subjective duration. Time-​activity
curves in the reproduction phase revealed similar climbing activation which peaked shortly
before the motor response in the left and right anterior insula, the inferior frontal cortex,
and the pre-​SMA. These results were interpreted as an indication that the anterior insula
and related areas are involved in the comparison of the two presented intervals (encoding,
reproduction phase). In a following fMRI study assessing time perception with students
covering a range of self-​reported impulsivity levels, we were able to replicate the main
findings of time-​activity curves in the posterior and anterior insula (Wittmann et al., 2011).
72

72 The embodiment of time

Following the fMRI results of insular cortex activation, we next investigated whether
performance in the duration-​reproduction task was related to individual interoceptive
awareness and with measurable changes in autonomic activity during the task (Meissner
& Wittmann, 2011). Healthy volunteers participated in two tasks:  (a) the heartbeat-
perception task conducted before the timing task as a measure of interoceptive sensitivity
and (b) the auditory duration-​reproduction task. Temporal intervals of 8-​, 14-​and 20-​
second duration were used while skin-​conductance levels and cardiac and respiratory
periods were recorded. First of all, a relationship was found between the accuracy in
perceiving one’s resting-​state heartbeat and performance in the following timing task. The
heartbeat-​perception task was used as suggested originally by Schandry (1981). Subjects
are asked to count their own heartbeats and report the number at the end of a desig-
nated interval. Participants who had higher interoceptive accuracy (they deviated less
from the true number of heartbeats) performed more accurately (with less deviation) in
the duration-​reproduction task (for a discussion of different interoceptive dimensions,
see Garfinkel et al., 2015). This indicated that a generally better access to visceral signals—​
ascending signals from the heart—​enables a more accurate representation of duration.
Regarding psychophysiological indices during the duration-​reproduction task, cardiac
periods increased (the heart rate slowed down) and skin-​conductance levels decreased (a
sign of relaxation) progressively and almost linearly during the encoding of the 8-​, 14-​,
and 20-​second intervals (Figure 4.2). These analyses point to a possible relationship be-
tween changes in physiological signals from the body and time-​perception accuracy. It is
important to emphasize that it is not the mean heart rate during a target interval that is
predictive of timing behavior, as was assessed by Schwarz and colleagues (2013), but the
changing (or increase) in cardiac periods over time.
The findings of linearly increasing cardiac periods and decreasing skin-​conductance
levels during the auditory duration-​estimation task were replicated in a follow-​up study
where we compared the behavior of meditators to non-​meditators (Otten et al., 2015).
Moreover, these physiological changes over time were also observed during an additional
visual duration-​reproduction task. Results thus strongly support the modality independ­
ence of mechanisms. One finding from the initial study (Meissner & Wittmann, 2011) was
not replicated: the heartbeat perception scores in the Schandry task (1981) were not as-
sociated with timing performance in the auditory and visual duration-​estimation tasks.
The accuracy index of interoception was, therefore, not predictive of time-​perception ac-
curacy in the following duration-​reproduction tasks. A visible ceiling effect—​individuals
were overall better aware of their heartbeats than in the former study—​could potentially
explain the lack of replication. Methodological considerations also have to be discussed,
such as the use of more objective psychophysiological tasks when assessing interoceptive
accuracy (Garfinkel et  al. 2015). One association with the interoception task used was
that heartbeat-​perception scores were positively related to correct answers in a divided-​
attention test (Otten et al., 2015). This result confirms a previously found correlation be-
tween the Schandry task and divided-​attention performance, adding evidence for the
embodiment conception of mental functioning (Matthias et al., 2009).
 73

Climbing activation 73

8-sec Interval
Cardiac Periods (ms) 830
820
810
800
790
780
770
760
750
12345678 123 12345 123
Encode ISI Reprod. ISI

Interval (S)

14-sec Interval
830
Cardiac Periods (ms)

820
810
800
790
780
770
760
750
1 2 3 4 5 6 7 8 9 11 13 123 123456789 123
Encode ISI Reprod. ISI

Interval (S)

20-sec Interval
830
Cardiac Periods (ms)

820
810
800
790
780
770
760
750
1 2 3 4 5 6 7 8 9 11 13 15 17 19 123 1 2 3 4 5 6 7 8 9 11 123
Encode ISI Reprod. ISI

Interval (S)
Figure 4.2  Mean second-​to-​second changes of cardiac periods (s) during the encoding interval,
the reproduction interval, and the first 3 s of the subsequent inter-​stimulus intervals (ISI) for tones
of 8-​s, 14-​s, and 20-​s duration. Due to individual differences in the length of the reproduced
intervals, the reproduction intervals were restricted to the initial seconds that were available for
all subjects, namely the first 5 s, 9 s, and 12 s of the 8-​s, 14-​s, and 20-​s intervals, respectively.
74

74 The embodiment of time

Another potential component of the heartbeat related to time-​estimation behavior has

recently been proposed: heart rate variability as an indicator of vagal control was meas-
ured during a baseline resting period. Using different indices of heart rate variability,
a positive correlation was found between duration-​reproduction accuracy of multiple-​
second intervals and the root mean square of successive differences (RMSSD) of inter-​
beat intervals (Pollatos et  al., 2014). A  positive correlation was also reported between
cardiac vagal tone (HF) and errors in a millisecond timing task: the greater the variability,
the better the temporal judgment (Cellini et al., 2015). These results fit with the model
of neuro-​visceral integration, where a greater vagal tone (stronger parasympathetic ac-
tivation) is associated with more efficient attentional regulation and response flexibility
(Thayer & Brosschot, 2005). A higher general vagal tone as measured in the resting-​state
condition could thereby facilitate the allocation of attention involved in time perception
and lead to more accurate duration judgment. However, these first study results have to
be followed up by studies that clarify the role of autonomic cardiac regulation in relation
to respiratory rate for performance in time perception tasks.

4.6  Applied research on embodied time

The upsurge of studies on meditation techniques is relevant for conceptual issues of the
conscious self and subjective time. In mindfulness meditation, functional aspects of self-​
consciousness are modulated by specific processes during meditation induction, that is
by increased attentional focus on body states and emotion regulation (Hölzel et al., 2011).
An important aspect that is noticed by novices during a meditation session and by expe-
rienced individuals at least at the beginning of a session is that subjective time slows down
considerably as the physical self becomes the focus of attention (Kabat-​Zinn, 2005). This
phenomenon is understandable within the conceptualization of insular cortex function
underlying the subjective passage of time. fMRI studies show how the instruction to focus
on “being in the present—​here and now” and on breathing sensations results in an in-
crease in neural activation in the insular cortex in experienced meditators (Farb et al.,
2007, Farb, Segal, & Anderson, 2012).
An interoceptive view on subjective time might also be fruitful for an understanding
of psychiatric and neurologic syndromes related to distorted notions of the physical and
mental self, emotional states, and time perception (Hartocollis, 1983). Schizophrenia
can even be seen as a disturbance of the embodied self (de Haan & Fuchs, 2010; Seth,
2013). Disturbances in time perception are reported by individual patients as the feeling
of being “stuck in time” (Vogeley & Kupke, 2007; Giersch & Mishara, 2017). Collected
reports show that for some of those patients, time is not passing and that the experi-
enced presence feels expanded: “What is the future? One cannot reach it. . . Time stands
still . . . This is boring, stretched time without an end.” (Fischer, 1929). Recent experi-
mental approaches show impairments in the consciously felt passage of time, as well as
in implicit measures of time continuity (Giersch et al., 2009; Lalanne et al., 2012). Similar
distortions can also be seen in individual patients with schizotypal disorders, such as
 75

References 75

the depersonalization syndrome, which are accompanied by depressive symptoms. These

patients experience a detachment from their own bodies and, in case studies, exhibit
disturbances in the subjective sense of time that are complemented by impairments in
objective timing tasks (D’Allonnes, 1905; Zaytseva et al., 2015). These empirical findings
correspond with the notion that the underlying self and time disturbances are related to
disruptions in embodied perceptual experience.
The passage of time can also be studied in patients who suffer from physical symptoms.
Wittmann and colleagues (2006) showed that patients with a life-​threatening illness who
had a lower health-​related quality of life and higher anxiety levels reported time to pass
more slowly. This finding suggests that the burden of symptoms slows down the passage
of time. This corresponds with preliminary results from an experimental study in which
nausea was experimentally induced by optokinetic stimulation: Participants with higher
nausea levels at the end of the 20-​minute nausea-​stimulation period reported time to
have passed more slowly than participants with lower nausea levels (n = 69, Spearman’s
rho = –​0.31, p < 0.01) (Meissner et al., in preparation). Although the specific mediators
of the association between symptom severity and the passage of time have yet to be
identified, these findings indicate the close relationship between the affective bodily self
and subjective time.
To sum up our analysis of the literature, a variety of time-​perception models exist si-
multaneously. Recent research presented here has added the idea of an “embodiment of
subjective time” implying the relationship between emotional physical states and the pro-
cessing of time duration. The classification of human time perception into different time
ranges indicates that different neural mechanisms are involved, depending on the length
of the processed time interval. To find consensus on this issue, research on the neural
basis of time perception has to test more systematically for duration-​dependent effects by
expanding the range of tested intervals within one study.

References
Berlin, H. A. and Rolls, E. T. (2004). Time perception, impulsivity, emotionality, and personality in self-​
harming borderline personality disorder patients. Journal of Personality Disorders, 18, 358–​78.
Bueti, D. and Walsh, V. (2009). The parietal cortex and the representation of time, space, number and
other magnitudes. Philosophical Transactions of the Royal Society B, 364, 1831–​40.
Buhusi, C. V. and Meck, W. H. (2005). What makes us tick? Functional and neural mechanisms of
interval timing. Nature Reviews Neuroscience, 6, 755–​65.
Buonomano, D. V., Bramen, J., and Khodadadifar, M. (2009). Influence of the interstimulus interval
on temporal processing and learning: Testing the state-​dependent network model. Philosophical
Transactions of the Royal Society B, 364(1525), 1865–​73.
Cellini, N., Mioni, G., Levorato, I., Grondin, S., Stablum, F., and Sarlo, M. (2015). Heart rate variability
helps tracking time more accurately. Brain and Cognition, 101,  57–​63.
Colonnello, V., Domes, G., and Heinrichs, M. (2016). As time goes by: Oxytocin influences the
subjective perception of time in a social context. Psychoneuroendocrinology, 68,  69–​73.
Coull, J. T., Cheng, R.-​K., and Meck W. H. (2011). Neuroanatomical and neurochemical substrates of
timing. Neuropsychopharmacology Review, 36,  3–​25.
76

76 The embodiment of time

Craig, A. D. (2009). Emotional moments across time: A possible neural basis for time perception in the
anterior insula. Philosophical Transactions of the Royal Society B, 364, 1933–​42.
Craig, A. D. (2015). How Do You Feel? An Interoceptive Moment with your Neurobiological Self.
Princeton, NJ: Princeton University Press.
Critchley, H. D., Wiens, S., Rotshtein, P., and Dolan, R. J. (2004). Neural systems supporting
interoceptive awareness. Nature Neuroscience, 7, 189–​95.
Damasio, A. (1999). The feeling of What Happens: Body and Emotion in the Making of Consciousness.
San Diego, CA: Harcourt Inc.
D’Allonnes, G.-​R. (1905). Rôle des sensations internes dans les émotions et dans la perception de la
durée. Revue Philosophique de la France et de l’Étranger, 60, 592–​623.
de Haan, S. and Fuchs, T. (2010). The ghost in the machine: Disembodiment in schizophrenia—​two
case studies. Psychopathology, 43, 327–​33.
Droit-​Volet, S., Brunot, S., and Niedenthal, P. M. (2004). Perception of the duration of emotional
events. Cognition and Emotion, 18, 849–​58.
Droit-​Volet, S. and Meck, W. H. (2007). How emotions colour our perception of time. Trends in
Cognitive Sciences, 11, 504–​13.
Eagleman, D. and Pariyadath, V. (2009). Is subjective duration a signature for coding efficiency?
Philosophical Transactions of the Royal Society B, 364, 1841–​52.
Effron, D. A., Niedenthal, P. M., Gil, S., and Droit-​Volet, S. (2006). Embodied temporal perception of
emotion. Emotion, 6,  1–​9.
Farb, N. A., Segal, Z. V., Mayberg, H., Bean, J., McKeon, D., Fatima, Z., et al. (2007). Attending to the
present: Mindfulness meditation reveals distinct neural modes of self-​reference. Social Cognitive
and Affective Neuroscience, 2, 313–​22.
Farb, N. A., Segal, Z. V., and Anderson, A. K. (2012). Mindfulness meditation training alters cortical
representations of interoceptive attention. Social Cognitive and Affective Neuroscience, 8,  15–​26.
Fischer, F. (1929). Zeitstruktur und Schizophrenie. Zeitschrift für die gesamte Neurologie und Psychiatrie,
121, 544–​74.
Garfinkel, S. N., Minati, L., Gray, M. A., Seth, A. K., Dolan, R. J., and Critchley, H. D. (2014). Fear
from the heart: Sensitivity to fear stimuli depends on individual heartbeats. Journal of Neuroscience,
34, 6573–​82.
Garfinkel, S. N., Seth, A. K., Barrett, A. B., Suzuki, K., and Critchley, H. D. (2015). Knowing your own
heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology,
104,  65–​74.
Gibbon, J., Church, R. M., and Meck, W. H. (1984). Scalar timing in memory. In J. Gibbon and L. Allan
(eds), Annals of the New York Academy of Sciences: Vol. 423. Timing and Time Perception. New York,
NY: New York Academy of Sciences, pp. 52–​77.
Giersch, A., Lalanne, L., Corves, C., Seubert, J., Shi, Z., Foucher, J., et al. (2009). Extended visual
simultaneity thresholds in patients with schizophrenia. Schizophrenia Bulletin, 35, 816–​25.
Giersch, A. and Mishara, A. (2017). Disrupted continuity of subjective time in the milliseconds range
in the self-​disturbances of schizophrenia: Convergence of experimental, phenomenological, and
predictive coding accounts. Journal of Consciousness Studies, 24,  62–​87.
Hartocollis, P. (1983). Time and Timelessness, Or, The Varieties of Temporal Experience (A Psychoanalytic
Inquiry). New York, NY: International Universities Press.
Hinton, S. C. and Meck, W. H. (2004). Frontal–​striatal circuitry activated by human peak-​interval
timing in the supra-​seconds range. Cognitive Brain Research, 21, 171–​82.
Hölzel, B. K., Lazar, S. W., Gard, T., Schuman-​Olivier, Z., Vago, D. R., and Ott, U. (2011). How does
mindfulness work? Proposing mechanisms of action from a conceptual and neural perspective.
Perspectives on Psychological Science, 6, 537–​59.
 7

References 77

Husserl, E. (1928). Vorlesungen zur Phänomenologie des inneren Zeitbewußtseins. Halle: Max Niemeyer
Verlag.
Ivry, R. B., Spencer, R. M., Zelaznik, H. N., and Diedrichsen, J. (2002). The cerebellum and event
timing. Annals of the New York Academy of Sciences, 978, 302–​17.
James, W. (1890). The Principles of Psychology. London: MacMillan.
Kabat-​Zinn, J. (2005). Coming to Our Senses. New York, NY: Hyperion.
Kiverstein, J. (2009). The minimal sense of self, temporality and the brain. Psyche, 15,  59–​74.
Kononowicz, T.W. and van Rijn, H. (2015). Single trial beta oscillations index time estimation.
Neuropsychologia, 75,  381–​9.
Kononowicz, T.W. and van Wassenhove, V. (2016). In search of oscillatory traces of the internal clock.
Frontiers in Psychology, 7(224),  1–​5.
Lalanne, L., van Assche, M., Wang, W., and Giersch, A. (2012). Looking forward: An impaired ability
in patients with schizophrenia? Neuropsychologia, 50, 2736–​44.
Lewis, P. A. and Miall, R. C. (2003). Distinct systems for automatic and cognitively controlled time
measurement: Evidence from neuroimaging. Current Opinion in Neurobiology, 13,  250–​5.
Lloyd, D. (2012). Neural correlates of temporality: Default mode variability and temporal awareness.
Consciousness and Cognition, 21, 695–​703.
Mach, E. (1911). Die Analyse der Empfindungen und das Verhältnis des Physischen zum Psychischen.
Jena: Verlag von Gustav Fischer.
Madl, T., Franklin, S., Snaider, J., and Faghihi, U. (2016). Continuity and the flow of time: A cognitive
science perspective. In: B. Mölder, V. Arstila, and P. Øhrstrøm (eds), Philosophy and Psychology of
Time. Cham: Springer International Publishing, pp. 135–​60.
Marchetti, G. (2009). Studies on time: A proposal on how to get out of circularity. Cognitive Processes,
10,  7–​40.
Matell, M. S. and Meck, W. H. (2004). Cortico-​striatal circuits and interval timing: Coincidence
detection of oscillatory processes. Cognitive Brain Research, 21, 139–​70.
Matthias, E., Schandry, R., Duschek, S., and Pollatos, O. (2009). On the relationship between
interoceptive awareness and the attentional processing of visual stimuli. International Journal of
Psychophysiology, 72,  154–​9.
Meissner, K. and Wittmann, M. (2011). Body signals, cardiac awareness, and the perception of time.
Biological Psychology, 86, 289–​97.
Menon, V. and Uddin, L.Q. (2010). Saliency, switching, attention and control: A network model of
insula function. Brain Structure and Function, 214, 655–​67.
Merleau-​Ponty, M. (1945). Phénoménologie de la perception. Paris: La Librairie Gallimard.
Metzinger, T. (2004). Being No One. The Self-​Model Theory of Subjectivity. Cambridge MA: MIT Press.
Metzinger, T. (2008). Empirical perspectives from the self-​model theory of subjectivity: A brief
summary with examples. In: R. Banerjee and B. K. Chakrabarti (eds), Progress in Brain Research,
168. Amsterdam: Elsevier, pp. 215–​45.
Morillon, B., Kell, C. A., and Giraud, A. L. (2009). Three stages and four neural systems in time
estimation. Journal of Neuroscience, 29, 14803–​11.
Nagel, T. (1974). What it is like to be a bat? The Philosophical Review, 83, 435–​50.
Northoff, G. (2016). Slow cortical potentials and “inner time consciousness”—​A neuro-​phenomenal
hypothesis about the “width of present”. International Journal of Psychophysiology, 103, 174–​84.
Noulhiane, M., Mella, N., Samson, S., Ragot, R., and Pouthas, V. (2007). How emotional auditory
stimuli modulate time perception. Emotion, 7, 697–​704.
Nummenmaa, L., Glerean, E., Hari, R., and Hietanen, J.K. (2014) Bodily maps of emotions.
Proceedings of the National Academy of Sciences of the USA, 111, 646–​51.
78

78 The embodiment of time

Otten, S., Schötz, E., Wittmann, M., Kohls, N., Schmidt, S., and Meissner, K. (2015). Psychophysiology
of duration estimation in experienced mindfulness meditators and matched controls. Frontiers in
Psychology, 6, 1215.
Park, H. D., Correia, S., Ducorps, A., and Tallon-​Baudry, C. (2014). Spontaneous fluctuations in neural
responses to heartbeats predict visual detection. Nature Neuroscience, 17, 612–​18.
Petter, E. A., Lusk, N. A., Hesslow, G., and Meck, W. H. (2016). Interactive roles of the cerebellum
and striatum in sub-​second and supra-​second timing: Support for an initiation, continuation,
adjustment, and termination (ICAT) model of temporal processing. Neuroscience & Biobehavioral
Reviews, 71, 739–​55.
Pöppel, E. (1997). A hierarchical model of temporal perception. Trends in Cognitive Sciences, 1,  56–​61.
Pollatos, O., Herbert, B. M., Matthias, E., and Schandry, R. (2007). Heart rate response after
emotional picture presentation is modulated by interoceptive awareness. International Journal of
Psychophysiology, 63, 117–​24.
Pollatos, O., Laubrock, J., and Wittmann, M. (2014). Interoceptive focus shapes the experience of time.
PloS One, 9, e86934.
Pollatos, O., Yeldesbay, A., Pikovsky, A., and Rosenblum, M. (2014). How much time has passed? Ask
your heart. Frontiers in Neurorobotics, 8, 15.
Rammsayer, T. (2009). Effects of pharmacological induced dopamine-​receptor stimulation on human
temporal information processing. NeuroQuantology, 7, 103–​13.
Revonsuo, A. (2006). Inner Presence. Consciousness as a Biological Phenomenon. Cambridge,
MA: MIT Press.
Salomon, R., Ronchi, R., Dönz, J., Bello-​Ruiz, J., Herbelin, B., Martet, R., et al. (2016). The insula
mediates access to awareness of visual stimuli presented synchronously to the heartbeat. Journal of
Neuroscience, 36, 5115–​27.
Schandry, R. (1981). Heart beat perception and emotional experience. Psychophysiology, 18,  483–​8.
Schwarz, M. A., Winkler, I., and Sedlmeier, P. (2013). The heart beat does not make us tick: The impacts
of heart rate and arousal on time perception. Attention, Perception & Psychophysics, 75, 182–​93.
Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive
Sciences, 17, 565–​73.
Singer, T., Critchley, H. D., and Preuschoff, K. (2009). A common role of insula in feelings, empathy
and uncertainty. Trends in Cognitive Sciences, 13, 334–​40.
Staddon, J. E. R. (2005). Interval timing: Memory, not a clock. Trends in Cognitive Sciences, 9, 312–​14.
Sysoeva, O. V., Tonevitsky, A., and Wackermann, J. (2010). Genetic determinants of time perception
mediated by the serotonergic system. PLoS One 5, e12650.
Thayer, J. F. and Brosschot, J. F. (2005). Psychosomatics and psychopathology: Looking up and down
from the brain. Psychoneuroendocrinology, 30, 1050–​8.
Terhune, D. B., Russo, S., Near, J., Stagg, C. J., and Kadosh, R. C. (2014). GABA predicts time
perception. Journal of Neuroscience, 34, 4364–​70.
Treisman, M. (2013). The information-​processing model of timing (Treisman, 1963): Its sources and
further development. Timing & Time Perception, 1, 131–​58.
Turel, O. and Bechara, A. (2016) A triadic reflective-​impulsive-​interoceptive awareness model of
general and impulsive information system use: Behavioral tests of neuro-​cognitive theory. Frontiers
in Psychology, 7, 601.
van Wassenhove, V. (2009). Minding time in an amodal representational space. Philosophical
Transactions of the Royal Society of London B, 364, 1815–​30.
Wittmann, M., Vollmer, T., Schweiger, C., and Hiddemann, W. (2006). The relation between the
experience of time and psychological distress in patients with hematological malignancies. Palliative
& Supportive Care, 4, 357–63.
 79

References 79

van Wassenhove, V. (2016). Temporal cognition and neural oscillations. Current Opinion in Behavioral
Sciences, 8, 124–​30.
Varela, F. J., Thompson, E., and Rosch, E. (1991/​2016). The Embodied Mind: Cognitive Science and
Human Experience. Cambridge, MA: MIT Press.
Varela, F. J. (1999). Present-​time consciousness. Journal of Consciousness Studies, 6, 111–​40.
Vogeley, K. and Kupke, C. (2007). Disturbances of time consciousness from a phenomenological and a
neuroscientific perspective. Schizophrenia Bulletin, 33, 157–​65.
Wackermann, J. and Ehm, W. (2006). The dual klepsydra model of internal time representation and
time reproduction. Journal of Theoretical Biology, 239, 482–​93.
Wackermann, J., Wittmann, M., Hasler, F., and Vollenweider, F. X. (2008). Effects of varied doses of
psilocybin on time interval reproduction in human subjects. Neuroscience Letters, 435,  51–​5.
Wackermann, J., Meissner, K., Tankersley, D., and Wittmann, M. (2014). Effects of emotional valence
and arousal on acoustic duration reproduction assessed via the “dual klepsydra model”. Frontiers in
Neurorobotics, 8, 11.
Wearden, J. (2016). Psychology of Time Perception. London: Palgrave Macmillan.
Wiener, M., Turkeltaub, P., and Coslett, H. B. (2010). The image of time: A voxel-​wise meta-​analysis.
NeuroImage, 49, 1728–​40.
Wiener, M., Lohoff, F. W., and Coslett, H. B. (2011). Double dissociation of dopamine genes and timing
in humans. Journal of Cognitive Neuroscience, 23, 2811–​21.
Wittmann, M. and Pöppel, E. (1999). Temporal mechanisms of the brain as fundamentals of
communication—​with special reference to music perception and performance. Musicae Scientiae, 3
(Suppl.),  13–​28.
Wittmann, M., Leland, D. S., Churan, J., and Paulus, M. P. (2007). Impaired time perception and
motor timing in stimulant-​dependent subjects. Drug and Alcohol Dependence, 90, 183–​192.
Wittmann, M. and Paulus, M. P. (2008). Decision making, impulsivity and time perception. Trends in
Cognitive Sciences, 12,  7–​12.
Wittmann, M. and Paulus, M. P. (2009). Temporal horizons in decision making. Journal of
Neuroscience, Psychology, and Economics, 2,  1–​11.
Wittmann, M. and van Wassenhove, V. (2009). The experience of time: Neural mechanisms and the
interplay of emotion, cognition and embodiment. Philosophical Transactions of the Royal Society B,
364, 1809–​13.
Wittmann, M., Simmons, A. N., Aron, J. L., and Paulus, M. P. (2010). Accumulation of neural activity
in the posterior insula encodes the passage of time. Neuropsychologia, 48, 3110–​20.
Wittmann, M., Simmons, A. N., Flagan, T., Lane, S. D., Wackermann, J., and Paulus, M. P. (2011).
Neural substrates of time perception and impulsivity. Brain Research, 1406,  43–​58.
Wittmann, M. (2016). Felt Time. The Science of How We Experience Time. Cambridge MA: MIT Press.
Zahavi, D. (2005). Subjectivity and Selfhood: Investigating the First-​Person Perspective. Cambridge,
MA: MIT Press.
Zakay, D. and Block, R. A. (1997). Temporal cognition. Current Directions in Psychological Science,
6(1),  12–​16.
Zakay, D. (2014). Psychological time as information: The case of boredom. Frontiers in Psychology,
5, 917.
Zaytseva, Y., Szymanski, C., Gutyrchik, E., Pechenkova, E., Vlasova, R., and Wittmann, M. (2015).
A disembodied man: A case of somatopsychic depersonalization in schizotypal disorder. PsyCh
Journal, 4, 186–​98.
80

Chapter 5

The neurobiology of gut feelings

Qasim Aziz and James K. Ruffle

5.1 Introduction
Interoception is defined as a “sensitivity to stimuli arising inside of the body,” in con-
trast to exteroception which can be defined as a “sensitivity to stimuli arising out-
side of the body.” There are a myriad of “gut feelings” which, taken together, form
the construct of gastrointestinal (GI) interoception. The GI tract is the second most
innervated bodily organ, second only to the brain, and specific modalities span from
interoceptive awareness percepts including visceral sensation and pain signalling,
hunger, or nausea, but also those outside of conscious perception including the in-
teroceptive impact of sub-​threshold sensory stimuli. These modalities are processed
by means of neuroanatomical pathways ascending to the level of the brainstem, sub-
cortical and cortical brain regions which may invoke a cognitive or bodily response.
Additionally, humoral mechanisms play a critical role in visceral interoception, as does
the autonomic nervous system (ANS). Moreover, these intricate signals are influenced
by various individual factors, including core physiology such as gender or genetics
and ANS tone, but also complex psyche components such as personality traits. In fact,
some research suggests that the intestinal bacterial flora composition may even influ-
ence how we “feel.”
Visceral interoceptive information reaches the brain through an array of neural, hu-
moral, and immune pathways, amalgamating in an abundance of ascending signalling
data which ensures homeostasis within the GI tract, many of which do not reach con-
scious awareness of interoceptive signals, for example in regulating GI motility and secre-
tion. However, those that engender conscious perception include sensations of hunger,
satiety, nausea, the urge to defecate, and, probably the most studied interoceptive signal
in GI physiology, pain (Derbyshire, 2003).
To begin, the neuroanatomical pathways of sensory signalling in the GI tract will be
described. Since the core emphasis in visceral interoception research has consisted of
understanding the sensory spectrum of the gut, with a vast research body investigating
pain, this therefore will form a key aspect in this chapter. When describing the neurobi-
ology and neuroanatomy of visceral sensory signalling this chapter will, for clarity, use
visceral pain as the sensory example. Additional conscious GI interoceptive signals will be
discussed later throughout the chapter, including that of hunger and nausea. Sequentially,
the humoral pathways underpinning the neurobiology of gut feelings will be described,
 81

Interoceptive signalling pathways 81

including the brain–​gut axis, and tie in how inter-​individual factors hold an influential
role in the regulation of these interoceptive signals.

5.2  Interoceptive signalling pathways

5.2.1  Ascending sensory pathways of visceral pain
Following a visceral insult that would instigate pain, nociceptive GI afferents localized
in the visceral tissue are activated. The ascending neural pathways from the GI tract are
encompassed in two main afferent routes: (a) the spinal pathway and (b) the tenth cranial
nerve (CNX), the vagus nerve (Coen, Hobson, & Aziz, 2012). These two afferent types
converge at multiple supra-​spinal levels, with the vagus projecting to the nucleus tractus
solitarii (NTS), a group of brainstem nuclei that receives numerous inputs including that
originating from baroreceptors, GI and pulmonary afferents (Altschuler et al., 1989). For
lamina I neurons, these relay principally to the parabrachial nucleus, before subsequently
ascending to higher centers, whilst spinothalamic tract fibers may project directly to tha-
lamic regions (Craig, 2003).
Both spinal and vagal afferents also indirectly project to subcortical and cortical areas,
including the thalamus (ventral posterior lateral, medial dorsal, and ventral medial pos-
terior nuclei), insula, amygdala, prefrontal cortex (PFC), primary somatosensory cortex
(SI), secondary somatosensory cortex (SII), and the anterior cingulate cortex (ACC). By
large these regions are thought to exhibit a degree of “viscerotopic” organization, meaning
that specific anatomical parts of the nuclei correspond to different visceral components
(Altschuler et al., 1989; Aziz et al., 2000). That being said, some exceptions do exist, with
the ventral posterior lateral aspect of the thalamus as an example of a region that does not
display viscerotopic organization (Coen et al., 2012). Overall, by these elaborate nuclei
interconnections, a visceral sensory “neuromatrix” exists both for innocuous and noci-
ceptive (painful) stimuli (Melzack & Wall, 1965).

5.2.2  Humoral signalling

In addition to the vagal and spinal homeostatic afferent pathways, an intricate humoral
pathway also aids in homeostatic regulation of visceral interoceptive signalling, reaching
the central nervous system by circulating substances. There are at least three disparate hu-
moral pathways permitting information transfer. First, the ventricular humoral pathway
(also known as the classical pathway) detects changes in substances at the level of the third
and fourth ventricle, engaging the adjacent circumventricular organs including the area
postrema, organum vasculosum of lamina terminalis, and subfornical organ (Figure 5.1)
(Ceunen, Vlaeyen, & Van Diest, 2016). In turn, processed humoral information projects
to the NTS, hypothalamus, PB nucleus, dorsal motor nucleus, nucleus ambiguous, insula,
ACC, sympathetic medullary nuclei, and midline thalamic regions.
Second, the blood−brain humoral pathway (also known as the non-​classical pathway)
detects changes in substances passing the blood−brain barrier. This pathway, which
invokes processing at the NTS, hypothalamus, amygdala, and monoamine systems, aims
82

82 The neurobiology of gut feelings

Pineal
Subfornical
Organ

Organum Vasculosum
of the Lamina Terminalis
Median
Eminence

Posterior Pituitary

Area Postrema

Figure 5.1 Location of circumventricular organs in the rat brain. AP = area postrema,

ME = median eminence, OVLT = organum vasculosum of the lamina terminalis, P = pineal gland,
PP = posterior pituitary, SFO = subfornical organ. (Lechan & Toni, 2016).

to influence information relay between insula, cingulate cortices, and ventral striatum.
Third, the microglial humoral pathway (also known as the extraneuronal pathway) refers
to the response engendered by residing microglia within circumventricular organs, lep-
tomeninges, and choroid plexus in the presence of both pathogen and/​or inflammation.
There exists a complex intercommunication between interoceptive signalling modalities.
Whilst for comprehensibility academic literature will describe neural, humoral, and in-
deed immune interoceptive signalling mechanisms in a reductionist and disparate fashion,
there exists significant interconnection and influence from one modality to the next,
which convolutes further with the influence of inter-​individual variability. As an example
highlighted by (Browning & Travagli, 2011), the visceral feeling state of dyspepsia can be
influenced by both psychological factors such as stress but also food indigestion. With
both stress and indigestion, glucagon-​like peptide 1 (GLP-​1), cholecystokinin (CCK),
and corticotrophin releasing factor (CRF) are released, which cause increase in cyclic
AMP levels at brainstem regions. By these humoral factors increasing a localized increase
in cAMP at vago-​vagal neurocircuitry, the dampening effect of GABAergic synapses is
decreased. Therefore, signals from normally innoxious events such as meal ingestion are
processed inappropriately and gastric function disrupted (Browning & Travagli, 2011;
Tack, Bisschops, & Sarnelli, 2004).

5.3  The “brain–​gut  axis”

The “brain–​gut axis” refers to the bi-​directional signalling processes between the brain
and the gut (Aziz and Thompson, 1998). Although physiological axes such as the
 83

Brain processing of GI sensory signalling 83

“hypothalamic pituitary axis” (HPA) have been historically described for some time, the
construct of physiological data transfer between the brain and the gut has become a rap-
idly developing area of interoceptive research particularly in the last few decades (Aziz
and Thompson, 1998). A  variety of research disciplines, albeit with a particular em-
phasis on functional neuroimaging, have provided a wealth of evidence to support the
proposal of a brain–​gut axis such that it has become an adopted term and central com-
ponent to contemporaneous research in visceral sensation (Omran and Aziz, 2014). By
large, a “visceral sensory neuromatrix” has been largely determined, including not only
aforementioned sensory neural components but also the autonomic nervous system
(ANS), neuroendocrine HPA, and neuroimmune systems (Van Oudenhove, Coen, &
Aziz, 2007).
The brain–​gut axis has served as an answer to the proposal that an intricate com-
munication system between the GI tract and the brain must exist in order to permit
interoception, but also to regulate GI function by means of brain to gut signalling. In
addition, the physiological, anatomical, and psychological factors that influence this bi-​
directional signalling pathway have become an intriguing area of brain–​gut axis research,
not least about how its function may be perturbed in various disorders, a poignant ex-
ample being irritable bowel syndrome (IBS) (Farmer & Ruffle, 2015).

5.4  The autonomic nervous system

The autonomic nervous system (ANS) is a key aspect of the bi-​directional nature of the
brain–​gut axis. Comprised of two largely opposing arms, the sympathetic (SNS) and par-
asympathetic nervous system (PNS), its function is to integrate changes in the external
environment with the internal self so as to maintain homeostasis. The functions of the
ANS are numerous and essential throughout the body, including regulation of metabo-
lism and critical components of the cardiorespiratory system. Importantly, it also plays a
key role in visceral interoception, including pain processing via the vagal nerve, the main
branch of the PNS (Ruffle et al., 2018; Ruffle et al., 2017a). Whilst the vagal nerve is largely
comprised of afferent fibers, the efferent arm is posited to have a key role in modulation
of visceral nociceptive signals (Botha et al., 2015), GI motility (Frokjaer et al., 2016), car-
diorespiratory regulation, and inflammation (Bonaz et al., 2013).

5.5  Brain processing of gastrointestinal sensory signalling

5.5.1  Visceral pain processing
Although research in visceral sensory representation in the brain is continually evolving,
the brain regions arguably best implicated thus far in pain processing are those of the SI,
SII, cingulate cortex, insula, PFC, thalamus, and amygdala (Aziz et al., 1997; Coen et al.,
2007; Derbyshire, 2003; Van Oudenhove et al., 2007). These regions have been elucidated
by means of stimulation of multiple visceral sites including the esophagus, stomach, and
rectum, but also varying stimulation methods such as mechanical, electrical, or acid-​
induced stimulation.
84

84 The neurobiology of gut feelings

Neuroanatomical and behavioral studies suggest that pain processing in the brain is
divided into medial and lateral pain systems. The somatosensory cortices, SI and SII, act
to encode the intensity and localization of the painful stimulus, referred to as the “lateral
pain system” (Aziz et  al., 2000; Van Oudenhove et  al., 2007). The multifaceted region
of the cingulate, including the ACC and PFC, as well as interconnection to the anterior
insula, amygdala, hippocampus, PAG, and brainstem, forms the “medial pain system,”
which plays an important role in in both the affective-​motivational (experience of pain
unpleasantness and related anxiety) and cognitive-​evaluative (both anticipation of and
attention to pain) (Gregory et al., 2003; Kulkarni et al., 2005).
The insula cortex holds a key role in interoception (Craig, 2002). Indeed, for visceral
sensation, the right anterior insula has been regarded by some groups as the “interocep-
tive cortex,” playing a key role in the awareness of the bodily self as a feeling entity (Craig,
2002). Although multimodal in its role, the insula is thought to have roles in processing
the affective dimension of pain whereby it may integrate the experience of pain with emo-
tional information (Ploner et al., 2011). The insula additionally has efferent outputs to
other key brain regions including the amygdala, hypothalamus, and periaqueductal gray
(PAG), which have additional roles in the processing of visceral pain (Carrive & Bandler,
1991; Gregory et al., 2003). The amygdala has an important role in the affective dimen-
sion of pain, and is regarded as the brain’s “fear center” (Stein et al., 2007). Additionally,
both the amygdala and PAG play key roles in the descending modulation of pain (Tracey
et al., 2002).
The PFC is understood to aid in the cognitive influence on visceral pain (Apkarian
et al., 2005). The PFC is a cluster of various sub-​regions, of which the orbitofrontal cortex
(OFC) processes visceral sensory information and encodes the affective, motivational,
and hedonic aspects (Kringelbach, 2005). Furthermore, the OFC assists in the decisions
pertaining to the autonomic and behavioral response to the stimulus, including the in-
tricate interaction between cognition (e.g. anticipation to pain) and emotion to pain
(Bantick et  al., 2002). The dorsolateral PFC (dlPFC) is another important sub-​region,
which is involved in cognition, specifically the attention and anticipation of visceral pain
(Aziz et al., 2000).
The thalamus is an important subcortical cluster of nuclei located in the center of the
brain. Its many roles include pain processing. Specifically, it transfers information be-
tween cortical regions via the relay nuclei, but it also has projections to higher cortical
regions. It is implicated in both pain sensation and the arousal response, yielding con-
nectivity to the aforementioned insula cortex, SI, and PFC (Aziz et al., 1997; Craig et al.,
1994). In the context of neuroimaging however, its activation is not always reported in the
context of visceral pain, compared to somatosensory where it appears to be more consist-
ently activated (Coen et al., 2012).

5.5.2  Brain networks in visceral sensation

Advances in functional neuroimaging over the last decade has permitted the ability to de-
termine proposed brain networks for visceral sensory and pain processing (Labus et al.,
 85

Descending analgesia in visceral pain 85

Table 5.1  Networks of visceral sensation

Homeostatic Afferent Emotional Arousal Cognitive-​Modulatory

Thalamus Amygdala ACC

Insula LC Amygdala
OFC ACC Insula
Dorsal ACC OFC
Three key networks for visceral sensation. Abbreviations: ACC, anterior cingulate cortex; LC, locus
coeruleus; OFC, orbitofrontal cortex.
From Omran & Aziz, 2014

2008; Tillisch & Labus, 2011). These networks are determined by means of correlating
brain activity during functional neuroimaging of a painful visceral stimulus, thus
inferring the interaction or “networking” between several brain regions. At present, it has
been suggested that three main brain networks may be key in the visceral pain matrix, as
follows: (a) the homeostatic-​afferent network, comprising the thalamus, insula, OFC, and
dorsal ACC; (b) the emotional arousal network, comprising the amygdala, LC, and ACC
sub-​regions; and (c) the cortical-​modulatory network, comprising the ACC, amygdala,
insula, and medial aspect of the OFC (Table 5.1, Figure 5.2) (Labus et al., 2008; Stein et al.,
2007; Tillisch & Labus, 2011).
The roles of these networks in processing a complex visceral interoceptive signal such
as pain differ. The homeostatic-​afferent network is understood to be responsible for the
processing of the interoceptive physiological input via the PB nucleus, whilst the emo-
tional arousal network is responsible for how the brain would interpret and perceive the
stimulus (Pezawas et al., 2005; Tillisch & Labus, 2011). Lastly, the cortical-​modulatory
network aids in the modulation of the pain experience, tying in to the aforementioned
process of descending analgesia (Coen et al., 2012).

5.6  Descending analgesia in visceral pain

A key mantra of the brain–​gut axis is the bi-​directionality of neural signalling. Signals
transmit caudally from brain to gut, and one such example is that of “descending an-
algesia,” a phenomenon that permits the brain to modulate the sensation of visceral
pain (Melzack & Wall, 1965). Most central brain structures that receive a visceral sen-
sory input, whether noxious or innocuous, additionally relay information caudally as a
means to modulate the sensory transmission of visceral afferents, in particular to the
dorsal horn of the spinal cord. One key region in descending analgesia is the ACC which,
prior to further caudal transmission, projects to the amygdala and PAG which may allow
both affective and cognitive factors to modulate the visceral pain experience (Dunckley
et  al., 2005; Tracey et  al., 2002). The amygdala and PAG caudally project to the locus
coeruleus (LC), and onwards to the dorsal horn of the spinal cord. It should be noted
that the construct of descending analgesia is not specific to visceral pain but engenders
86

86 The neurobiology of gut feelings

(a) Homeostatic afferent network (b) Emotional arousal network

aMCC aMCC
spgACC

pgACC
Insula

Thalamus Amygdala
Brain stem

Figure 5.2 Networks of brain activation in visceral stimulation studies. (a) The homeostatic–​

afferent network encompasses areas that are thought to resemble those of pain when stimulated
in visceral pain studies both in participants with a FGID and in participants who are otherwise
healthy. The core regions are shown.(b) The emotional arousal network is depicted. Cognitive
(e.g. expectation and anticipation), emotional (e.g. sadness), and psychological aspects have
all been shown to be involved in visceral perceptions and this progress has established what is
known as the “emotional arousal network.” The central components of this network are the
amygdala and parts of the ACC. Abbreviations: ACC, anterior cingulate cortex; aMCC, anterior
midcingulate cortex; FGID, functional gastrointestinal disorder; pgACC, perigenual ACC; sgACC,
subgenual ACC.
Reprinted by permission from Nature Reviews Gastroenterology & Hepatology, 11 (9), Functional brain imaging
in gastroenterology: to new beginnings, Yasser Al Omran and Qasim Aziz, pp. 565–​76, Figure 1, doi:10.1038/​
nrgastro.2014.89, Copyright © 2014, Springer Nature.

somatosensory nociception also, overall referred to famously as Melzack and Wall’s pain
“gate control theory” (see Melzack & Wall, 1965). Interestingly however, this has formed
an important area of visceral pain research regarding how individual factors such as the
psyche, including personality or anxiety, also influence descending analgesia (Farmer
et al., 2014a; Farmer et al., 2013; Tracey, 2011). Such roles for individual factors will be
further described later.

5.7  Additional interoceptive modalities

Visceral states include numerous modalities other than core sensory and pain pathways.
These range from sensations under the umbrella of conscious interoceptive awareness,
such as nausea and hunger, to far more complex visceral states such as anticipation of a
visceral threat, or sub-​threshold sensation to reach conscious awareness. These differing
modalities will be discussed in the following sections.

5.7.1 Nausea
Nausea is a common and distressing symptom, often preceding vomiting. Numerous
evidence has purported that the bi-​directional interactions between brain–​gut axis, in
addition to both autonomic and endocrine systems (in particular arginine vasopressin,
 87

Additional interoceptive modalities 87

ghrelin, and cortisol), play important roles in the experience of nausea. Indeed, the sen-
sory experience of nausea is often accompanied by numerous physiological changes, in-
cluding pallor, sweating, and gastric dysrhythmia (Sanger & Andrews, 2006).
The neurobiology of nausea has been implicated the ANS, in particular the vagal
pathway to cortical brain regions. Through mechanism similar to that of pain as
described earlier, abdominal vagal afferent fibers reach the NTS and area postrema
which are thought to play a major role in integrating both the emetic and nausea re-
sponse. Our understanding of the neurobiology of nausea is further complicated by
studies illustrating that abdominal vagal afferents are not essential for its sensation,
given that humans with bilateral abdominal vagotomy can still experience nausea
(Sanger & Andrews, 2006). It is important to note, however, the difference between
emesis, the act of vomiting, and nausea, the physiological sense of feeling sick. Whilst
the two often occur together, they are not equivalent. Because there are no clear an-
imal models of nausea (whilst there are for emesis), the neurobiology of emesis has
been clearly demonstrated, with a particular emphasis at the brainstem level, given
that decerebrate animal models still can exhibit emetic behavior (Miller, Nonaka, &
Jakus, 1994).
Functional neuroimaging studies suggest that “higher” cortical regions and the ves-
tibular system are particularly important in the genesis of nausea. In a recent functional
magnetic resonance imaging (fMRI) study, it has been shown that nausea (induced
by visual stimulus) is associated with decreased plasma ghrelin and vasopressin, with
increased ACC, inferior frontal, and middle occipital gyri activity (Farmer et al., 2015).
Meanwhile, cerebellar areas showed decreased activity, including the declive nucleus and
parahippocampal gyrus (Farmer et al., 2015).

5.7.2 Hunger
Hunger or appetite is an additional visceral interoceptive state. Comprising
interconnections between behavior and core physiology, it has been purported that
four main brain areas play key role in the control of appetitive behavior. These are as
follows:  (a) amygdala/​hippocampus, (b)  insula, (c)  OFC, and (d)  striatum, together
formulating a complex interoceptive phenomena of learning about reward (food), alloca-
tion of attention and resource/​effort (to food), and the integration between bodily home-
ostasis regarding energy stores, GI contents, and extrinsic signals such as food availability
(Figure 5.3) (Dagher, 2009).
Homeostatic and GI information is largely relayed by circulating gut peptide hormones
and nutrients, which act at the hypothalamus. Additional brain structures are also in-
volved, as is the vagal nerve, in regulating appetite and the behaviors as described earlier,
which are perturbed if their interconnections are disrupted. For example, one study has
shown that lesion-​induced disruption of the amygdala–​OFC interconnection abolishes
sensory-​specific satiety—​that is, a normal response whereby food fed to satiety loses its
incentive properties (Holland & Gallagher, 2004). The PFC additionally adds cognitive
influence over the apparent brain network of appetite regulation.
8

88 The neurobiology of gut feelings

Sensation Lateral PFC Interoception

(taste, olfaction, vision) (hunger, nausea)

Insula

Amygdala/
OFC
hippocampus

Striatum

VTA (DA)

Hypothalamus

Figure 5.3 A brain network for appetitive behavior. (Not all connections are depicted)
PFC, prefrontal cortex; OFC, orbitofrontal cortex; VTA, ventral tegmental area; DA, dopamine.
Reprinted by permission from International Journal of Obesity, 33 (S2), The neurobiology of appetite: hunger as
addiction, A. Dagher, pp. S30–​S33, Figure 1, doi: 10.1038/​ijo.2009.69, Copyright © 2009, Springer Nature.

5.7.3  Interoceptive awareness of sensory signals

Under normal physiological conditions, stimulation of intestinal afferents does not typi-
cally reach conscious awareness. However, these neural signals likely still reach the brain-
stem in order to subsequently trigger GI reflexes to maintain bodily homeostasis, and
thus permit interoceptive awareness (Kern & Shaker, 2002). Without this process, there
would be no feedback loop to maintain the internal GI milieu. Until relatively recently, it
was also unknown whether viscerosensory signals of interoceptive awareness even reach
higher cortical regions.
In an early study, by means of rectal distension to a threshold below conscious per-
ception in healthy volunteers, it was shown that these afferent signals, whilst not strong
enough to reach conscious perception, are still registered in the cortex, and are detect-
able by functional neuroimaging. However, the degree of magnitude of activity of these
cortical areas was much smaller than a visceral interoception sensory afferent signal that
would reach the level of conscious perception (Kern & Shaker, 2002). The areas activated
following stimulation that was not consciously perceived were fourfold:  sensorimotor,
parietal-​occipital, ACC, and insula (Kern & Shaker, 2002). The group posited that these
areas of activation came as a consequence of vagal or spinal afferents as discussed earlier
in this chapter. Additionally, given that sensorimotor cortical regions are supplied by
spinal afferents alone, it was suggested that visceral interoception sensory signals that do
not reach conscious perception may be relayed to the cortex by spinal afferents, whilst the
vagal nerve be more implicated in conscious sensation or pain proper (Kern & Shaker,
2002). In follow-​up studies from the same group, the effect of esophageal acid stimulation
 89

Inter-individual factors in visceral interoception 89

to below-​conscious perception thresholds were additionally explored, revealing sim-

ilar findings of activity in the cingulate, insula and sensorimotor regions (Kern, Chai, &
Lawal, 2009). Indeed, more recent studies have also shown how the functional connec-
tivity of interoceptive regions such as the insula are modulated by esophageal acid stimu-
lation below a threshold of conscious perception (Babaei et al., 2013).

5.7.4  Anticipation of visceral threat

An additional visceral interocept that has gained attention over the last decade is that
of anticipation to a bodily threat. Largely achieved by means of the threat of pain (e.g.
to be told one would soon receive visceral pain), it has been shown that anticipation is
processed at the brain level quite differently to that of pain proper. When expecting a
threat which merits emotional arousal, one’s anticipation modulates the sequential cog-
nitive response to that stimuli, such as the stress and anxiety that might accompany it.
Numerous groups have studied pain anticipation in an effort to characterize brain re-
gions implicated, and perhaps even that an anticipatory brain network should such exist.
One frequently proposed brain region in anticipation is the anterior insula, given its well-​
described roles in conscious interoceptive awareness. However, additionally proposed
regions include the ACC, PAG, PFC, medial frontal lobe, amygdala, and OFC. These re-
gions are implicated in the processing of visceral pain also, and thus their documented
activation prior to actual pain proper is interesting in the context of modulation, certainly
warranting further study (Bishop et  al., 2004; Coen et  al., 2011; Fairhurst et  al., 2007;
Ploghaus et al., 2001; Ruffle et al., 2015b). In the context of disease, anticipation of pain
in IBS is different from healthy controls as they inactivate certain brain regions, including
the ACC and amygdala, to a lesser degree than IBS patients, possibly representing a mal-
adaptive coping strategy to pain in IBS patients (Berman et al., 2008).

5.8  Inter-​individual factors in visceral interoception

A major focus into advancing our understanding to the “neurobiology of gut feelings”
has been to characterize the factors of an individual which influence visceral interocep-
tive processing, referred to as the “inter-​individual factors.” In gastroenterology, further-
more, the Rome working collaborative group has previously highlighted how lack of
understanding of these factors at present significantly limits advancements in the field
(Drossman & Hasler, 2016). To date, there are an array of factors which have been discov-
ered, and span genetics, physiology, neuroanatomy, and psychophysiology. Taking this a
step further, some groups have suggested the concept of visceral pain “endophenotypes”
(Farmer & Aziz, 2014; Ruffle et al., 2017b; Tracey, 2011). In understanding the neurobi-
ology of visceral interoception, it is important to take these factors into account, and thus
will be discussed later in this chapter.

5.8.1 Gender
One frequently studied difference in visceral interoception is that of gender. It has been
often documented that women demonstrate both higher pain sensitivity and greater
90

90 The neurobiology of gut feelings

prevalence of chronic visceral pain conditions than the male counterparts (Kano et al.,
2013), IBS being an example discussed in section 5.9.2. In fact, some contemporaneous
studies have even moved to studying only males or females, so as to eliminate the pos-
sibility of a gender confound in their results. These gender differences have been and
continue to be investigated with the ever-​advancing tools in neuroimaging. As one
example, a previous study utilizing acute esophageal pain, by mechanical distension,
showed that female subjects display greater brain activity in the mid-​cingulate cortex,
anterior insula, and premotor cortex regions well associated with the emotional arousal
constituent of visceral pain processing which led the group to suggest that females may
attribute greater emotional importance to an acute visceral pain stimulus than males
(Kano et  al., 2013). These findings of increased activity in emotional arousal brain
regions have also been reported by other groups using connectivity analysis (Labus
et al., 2008).

5.8.2 Genetic
Genetic differences have additionally been investigated in the brain processing of vis-
ceral signals. For example, polymorphisms in the 5-​hydroxytryptamine (5-​HT) signalling
system have been shown to play a role in processing of interoceptive signals, the subse-
quent stress response a signal may invoke and furthermore the emotional regulation that
accompanies it (Kilpatrick et  al., 2015). Specific to visceral pain, the 5-​HT (serotonin)
transporter gene-​linked polymorphic region (5-​HTTLPR) has been shown to affect the
brain response. In a neuroimaging study of brain connectivity in healthy males exposed
to painful mechanical rectal distension, the S/​S genotype of 5-​HTTLPR was associated
with significantly greater hippocampal-​amygdala strength, an important central pro-
cessing component of the stress response and emotion regulation, compared to alter-
native genotypic carriers (Kilpatrick et  al., 2015). Indeed, additional studies have also
suggested that the 5-​HTTLPR genotype is associated with a differing extent of emotional
regulation in visceral pain, whereby the S/​S genotype has been shown to correspond to
greater cerebral blood flow (as a surrogate to brain activity) in the ACC, hippocampus,
and OFC in one study (Fukudo & Kanazawa, 2011; Fukudo et al., 2009), while to the in-
sula, inferior frontal gyrus, supplementary motor area, and precentral gyrus in another
(Schaub et al., 2013). Furthermore, it has been suggested that the functional gene poly-
morphism may predict the efficacy of a selective serotonin re-​uptake inhibitor (SSRI) in
the use of visceral pain, a finding particular poignant in current GI practice where SSRIs
have been investigated for their utility in relieving IBS symptoms (Farmer & Ruffle, 2015;
Tack et al., 2006).

5.8.3  Autonomic physiology

Abnormalities in the activity of the ANS are apparent in numerous GI disorders, many
of which are associated with perturbation in visceral interoception processes (fre-
quently pain). These clinical disorders specific to the GI tract include functional chest
pain, IBS, inflammatory bowel disease, and Ehlers–​Danlos syndrome (Bonaz et  al.,
 91

Inter-individual factors in visceral interoception 91

2013; De Wandele et al., 2014; Hoff et al., 2016; Ruffle et al., 2015a; Ruffle et al., 2018;
Spaziani et  al., 2008). Over the last decade, accumulating evidence has suggested an
anti-​nociceptive role for the PNS whereby many of these aforementioned disorders ex-
hibit a paucity of baseline vagal tone when compared to healthy controls (Botha et al.,
2015; Farmer et al., 2014a). The exact mechanism by which abnormal function of the
vagus may lead to anti-​nociception is not yet known, although it has been proposed to
have an anti-​nociceptive and anti-​inflammatory function (Farmer et  al., 2016; Ruffle
et al., 2017a).
Baseline autonomic function is an additional important variable to account for in un-
derstanding the brain processing of visceral interoceptive stimuli. With structural neuro-
imaging, it has been shown that cortical thickness of the mid-​cingulate cortex correlates
with degree of resting parasympathetic activity, quantified by heart rate variability (HRV)
(Winkelmann et al., 2016). Meanwhile, using fMRI it has additionally been shown that
activity in the ACC and cerebellum correlates with vagal tone (Kano et al., 2014). Using
vertex analysis, it has recently been reported that the shapes of subcortical nuclei are
influenced by both the resting SNS and PNS tone of an individual (Ruffle et al., 2018).
Specifically, SNS was positively correlated to outward shape changes of the brainstem,
nucleus accumbens, amygdala, and pallidum, whilst PNS vagal tone was negatively
correlated to inward shape changes of the amygdala and pallidum. Furthermore, vagal
tone was correlated to total volume of the brainstem and the putamen. Needless to say, it
is clear that baseline autonomic function in an individual influences their processing of
visceral neural signals.

5.8.4 Psychophysiological
The emotional and affective dimension of a visceral signal, especially a threat such as
pain, is a key aspect to its cognitive interpretation. Psychophysiological factors such
as personality traits have historically been thought highly important in the cognitive
processes underpinning processing a threat such as pain (Harkins, Price, & Braith,
1989). However, in the context of the viscera, the influence of personality has only rela-
tively recently been studied. Using mechanical esophageal distension in induce visceral
pain, it has been shown with fMRI that personality, anxiety, negative emotion, and so-
ciability influence central brain processing of anticipation and actual visceral pain (Coen
et al., 2008; Coen et al., 2011; Kumari et al., 2007; Paine et al., 2009a; Paine et al., 2009b;
Ruffle et al., 2015b).
During a study by Coen and colleagues, in anticipation to visceral pain, higher neu-
roticism scores were associated with greater activity in the thalamus, parahippocampal
gyrus, and ACC, but these regions had lower activity during visceral pain proper. These
regions are previously attributed to the emotional and cognitive appraisal of pain, but
it is thought that these findings may reflect the neurotic individual’s tendency to a
heightened arousal during pain anticipation yet an avoidance coping response during
pain (Coen et  al., 2011). Notably, neuroticism has strong ties to IBS also and is a
risk factor for unexplained chronic abdominal pain in the disorder (Hazlett-​Stevens
92

92 The neurobiology of gut feelings

et al., 2003). Referring to extraversion, in a follow-​up study it was shown that during
both pain anticipation and visceral pain, a higher degree of extraversion was asso-
ciated with greater activity to the right insula, which may link to the sympathetic-​
predominant response to threat which higher extraversion individuals display (Ruffle
et al., 2015b).

5.8.5 Microbiota
The human microbiota is a diverse and dynamic ecosystem within the GI tract, which
has evolved to be symbiotic to the host. An estimated 1014 microorganisms populate the
adult gut (Ley, Peterson, & Gordon, 2006). Whilst sterile prenatally, inoculation occurs
vertically during birth, and over the course of the first year of life it is fully established.
Throughout an individual’s life, this ecosystem is determined by genetics, diet, and en-
vironment (Dominguez-​Bello et  al., 2011). The microbiota protects the body from ex-
ternal pathogens, aids in digestion and metabolism, and additionally acts as a source
for neurochemicals in numerous bodily processes, including neural transmission.
As an example, the microbiota is responsible for 95% of bodily serotonin (Sommer &
Backhed, 2013).
The microbiota has formed an interesting area of GI research in recent years,
culminating in the suggestion that it forms an aspect in bi-​directional brain–​gut axis
communication; the brain influences the microbiota, and the microbiota influences the
brain (Figure 5.4) (De Vadder et al., 2014). The central nervous system can influence the
GI microbiota composition (a) directly, for example by means of neurotransmitter release
(e.g. serotonin) from enterochromaffin cells, neurones, and immune cells, or (b)  indi-
rectly, by fluctuations in GI motility and secretions by autonomic tone (this subsequently
affecting blood flow and nutrient availability for said microbiota) (Rhee, Pothoulakis, &
Mayer, 2009).
To investigate the effect of microbiota on the brain, studies have largely relied on the
premise of sterility in prenatal animals and maintenance in sterile environments postna-
tally (see Cryan & Dinan, 2012; Cryan & O’Mahony, 2011). Using these methods, groups
have shown that the microbiota affect the central nervous system by affecting behavioral
traits in animals, accompanied by changes in neurotrophic factors such as brain-​derived
neurotrophic factor (BDNF) in regions such as the hippocampus (Bercik et  al., 2011).
Numerous animal examples are apparent (Cryan & Dinan, 2012; Cryan & O’Mahony,
2011; Farmer, Randall, & Aziz, 2014b).
In GI disease, it is also thought that changes to the gut microbiota may play an im-
portant role. The frequent comorbidity of mood disorders in GI disorders, including
both IBS and coeliac disease, has been proposed in part due to changes in the micro-
biota. Furthermore, microbiota may perturb the brain–​gut axis, one poignant example
being that of post-​infection IBS (Spiller & Lam, 2012). The concept of the microbiota
influencing the neurobiology of gut feelings is an exciting and developing field expected
to reveal significant developments for our understanding of the GI microbiota and its im-
pact on interoception (Farmer et al., 2014b; Rhee et al., 2009).
 93

Inter-individual factors in visceral interoception 93

Mood, cognition
& emotion

Hypothalamus
CRF
Pituitary

ACTH Vagus nerve

Adrenals
Systemic
circulation
Cytokines

Cortisol

Gastrointestinal lumen

Gastrointestinal microbiota Neurotransmitters

Figure 5.4 Bidirectional brain–​gut–​microbiota pathways. Multiple pathways, including but not

limited to neural, endocrine, and immune, exist in which the gastrointestinal microbiota may
modulate the brain. Abbreviations: ACTH, adrenocorticotrophic hormone; CRF, corticotrophin
releasing hormone.

5.8.6  Combining factors to form complex endophenotypes

One future direction of this research area is the possibility of combining these factors
to form visceral “endophenotypes” as a form of statistical dimension reduction of mul-
tiple data arrays into a simpler construct (Tracey, 2011). Endophenotypes are measur-
able components of a disease/​condition, which may include neuroanatomical or cognitive
characteristics that have simpler ties to underpinning factors such as genetics. Clustering
individuals into endophenotypes has been suggested as a means to aid identification of
patients at risk for developing chronic pain.
Using a group of healthy controls, it has been previously shown that more pain-​sensitive
and pain-​resistant phenotypes do indeed exist, using arrays of physiological, psychophys-
iological, and genetic data (Farmer et al., 2014a; Farmer et al., 2013). With data coalesced
94

94 The neurobiology of gut feelings

into two main clusters, individual’s in pain cluster 1, at baseline, show higher neuroti-
cism and anxiety, a preponderance for the SS/​LS 5-​HTTLPR genotype, higher baseline
sympathetic tone, lower parasympathetic tone, and higher serum cortisol. During pain
however, this group tolerated it less, habituated less to the stimulus, showed a greater
parasympathetic nervous system activation, a sympathetic nervous system withdrawal,
and had an elevated serum cortisol. In contrast, pain cluster 2 at baseline showed higher
extraversion but lower anxiety, a preponderance for the L/​L genotype, lower serum cor-
tisol, a higher parasympathetic tone with low sympathetic tone. During pain however, the
group tolerated the pain stimulus to a greater extent, displayed a greater habituation effect
and a parasympathetic withdrawal coupled to a sympathetic activation. Using fMRI,
individuals with high extraversion and low anxiety have been shown preferentially to
activate the frontal cortex, insula, and left thalamus during pain in comparison to cluster
1 individuals, see (Farmer et al., 2014a; Farmer et al., 2013). In a follow-​up study of this
cohort, it has also recently been shown that visceral pain endophenotypes can be accu-
rately predicted using machine learning with whole-​brain connectivity data with over
85% accuracy (Ruffle et al., 2017b).

5.9  Comparing gut feelings in health and disease

5.9.1  Chronic pain
Acute and chronic pain both have a distinct neurobiology, and indeed visceral pain is no
exception. Some authorities have even called for chronic pain to be considered an indi-
vidual disease entity (Tracey & Bushnell, 2009). Herein two common causes of chronic
visceral pain will be discussed, IBS and inflammatory bowel disease (IBD). Both IBS and
IBD are associated with altered interoception, and thus these conditions also demonstrate
altered brain processing of acute visceral painful compared to healthy controls.

5.9.2  Irritable bowel syndrome

IBS is one of the most often studied “functional” GI disorders with regards to neuroim-
aging of visceral interoception in disease, in part attributable to its incomplete under-
standing and high prevalence of between 5% and 20% (Farmer & Ruffle, 2015). IBS is
more prevalent in women, and is associated with a significant socioeconomic burden by
means of a reduction in health-​related quality of life, not least large healthcare costs from
recurrent consultation and over-​investigation. It has been described as one of the most
common persistent “pain syndromes,” where dysregulation of the aforementioned brain–​
gut axis is a core aspect of the underpinning disorder. This in turn leads to the develop-
ment of a visceral hypersensitivity state. Denoted by the Rome working group criteria,
now in its fourth iteration, IBS patients by definition experience recurrent abdominal
pain which improves with defecation (Drossman & Hasler, 2016).
As discussed earlier in the chapter, there are three key networks in the processing
of a visceral pain, the homeostatic-​afferent, the emotional arousal, and the cortical-​
modulatory. In IBS, a disorder for which the presence of recurrent visceral pain is part
 95

Comparing gut feelings in health and disease 95

of its very definition, the homeostatic-​afferent and emotional arousal networks dem-
onstrate increased engagement (Tillisch & Labus, 2011). Because the pathophysiology
of IBS has been difficult to understand, multiple groups have turned to neuroimaging
as a means to investigate it. These studies have demonstrated that the majority of the
visceral pain neuromatrix is activated as described earlier (Mayer et al., 2005), however,
differences from healthy controls are apparent. A meta-​analysis has shown that patients
with IBS display more consistent brain activation in emotional arousal network regions,
including the pregenual ACC and amygdala, as well as brainstem regions implicated
in the cortical-​modulatory aspects of descending analgesia (Tillisch, Mayer, & Labus,
2011). Structural brain differences in IBS patients and controls have also been studied,
revealing that IBS patients exhibit a decreased grey matter density (GMD) in many re-
gions involved in visceral pain processing, including the thalamus, PFC, posterior pa-
rietal cortex, and ventral striatum, but an increase in GMD in the ACC and OFC, areas
involved in the cognitive and attentional aspects of visceral pain (Seminowicz et  al.,
2010). Additionally, it is possible that increased anxiety in the IBS group also affects
these findings (Seminowicz et al., 2010), not least the change in baseline autonomic tone
(Ruffle et al., 2015a; Ruffle et al., 2018). It is possible that duration of disease (Blankstein
et al., 2010) or indeed important inter-​individual factors influence such results. Further
research is certainly warranted.
It is also possible to investigate white matter tract differences between chronic vis-
ceral pain conditions and healthy controls, by means of diffusion tensor imaging. Using
probalistic tractography, a reduced structural connectivity was demonstrated in IBS
patients between the amygdala and the dorsolateral PFC, possibly related to a decreased
or insufficient inhibition of emotional arousal circuitry leading to amplification of the
visceral pain experience (Labus et al., 2010). Notably, healthy control studies investigating
brain white matter have shown a complex neural network for visceral pain including the
insula, thalamus, ACC, PFC, SI, and SII (Moisset et  al., 2010). However, the degree of
white matter connectivity between these regions has been shown to be highly variable,
most likely due to inter-​individual variability factors (Moisset et  al., 2010). Lastly, the
influence of genetic factors on the processing of visceral interoceptive signals in healthy
individuals was discussed in section 5.8.6 with specific reference to the 5-​HTTLPR. This
polymorphic region has also been studied in the context of IBS, whereby it has been
shown that the S/​S genotype corresponded to decreased IBS symptoms, whilst the LS/​SS
genotype was associated both with increased pain ratings in one study (Camilleri et al.,
2008), but also the severity of symptoms as a whole in another (Colucci et al., 2013).

5.9.3  Inflammatory bowel disease

IBD refers to both ulcerative colitis (UC) and Crohn’s disease (CD) which are chronic
GI disorders frequently accompanied by abdominal pain. The neurobiology of this pain
has been investigated by some groups. Using positron emission tomography and painful
mechanical rectal distension, it has been shown that UC patients, whilst showing brain
region activity comparable to the aforementioned pain neuromatrix including the ACC,
96

96 The neurobiology of gut feelings

insula, and PAG, also show greater activity in the frontal cortex and PAG circuits (Mayer
et al., 2005). Interestingly, when comparing IBD to IBS cohorts, IBD patients additionally
show greater activity in the pons and PAG in comparison to IBS patients. Conversely, IBS
patients display greater activity in the ACC, PFC, and amygdala in comparison to UC
patients (Tillisch & Labus, 2011). These findings may suggest that in IBS there is greater
involvement of emotional arousal circuits to visceral pain, whilst UC patients appear to
show greater activity of regions involved in descending analgesia (such as the PAG).

5.10 Conclusion
The neurobiology of gut feelings is a complex and ever-​evolving field of interoception.
Ascending signalling pathways include neural, humoral, neuroimmune, and endocrine,
and culminate with intricate brain networks specific to a specific visceral state, including
those of visceral pain, hunger, and nausea. With the brain–​gut axis, a bi-​directional com-
munication between gut and brain is permitted to ensure gastrointestinal homeostasis,
and to that end, inter-​individual factors including genetics, demographic, the psycho-
physiological, and the neurophysiological influence its regulation at an individual level.
Furthermore, this neurobiology is perturbed in common GI disorders such as IBS and
IBD. Future avenues of visceral interoception research should further interrogate these
inter-​individual factors thus far identified, including autonomic neurophysiology and gut
microbiota, and indeed how these interact with one another.

References
Altschuler, S. M., Bao, X. M., Bieger, D., Hopkins, D. A., and Miselis, R. R. (1989). Viscerotopic
representation of the upper alimentary tract in the rat: Sensory ganglia and nuclei of the solitary and
spinal trigeminal tracts. Journal of Comparative Neurology, 283(2), 248–​68.
Apkarian, A. V., Bushnell, M. C., Treede, R. D., and Zubieta, J. K. (2005). Human brain mechanisms of
pain perception and regulation in health and disease. European Journal of Pain, 9(4), 463–​84.
Aziz, Q., Andersson, J. L., Valind, S., Sundin, A., Hamdy, S., Jones, A. K., et al. (1997). Identification
of human brain loci processing esophageal sensation using positron emission tomography.
Gastroenterology, 113,  50–​59.
Aziz, Q. and Thompson, D. G. (1998). Brain–​gut axis in health and disease. Gastroenterology, 114(3),
559–​78.
Aziz, Q., Thompson, D. G., Ng, V. W., Hamdy, S., Sarkar, S., Brammer, M. J., et al. (2000). Cortical
processing of human somatic and visceral sensation. Journal of Neuroscience, 20(7), 2657–​63.
Babaei, A., Siwiec, R. M., Kern, M., Douglas Ward, B., Li, S. J., and Shaker, R. (2013). Intrinsic
functional connectivity of the brain swallowing network during subliminal esophageal acid
stimulation. Neurogastroenterology and Motility, 25(12), 992–​e779.
Bantick, S. J., Wise, R. G., Ploghaus, A., Clare, S., Smith, S. M., and Tracey, I. (2002). Imaging how
attention modulates pain in humans using functional MRI. Brain, 125(Pt 2), 310–​19.
Bercik, P., Denou, E., Collins, J., Jackson, W., Lu, J., Jury, J., et al. (2011). The intestinal microbiota
affect central levels of brain-​derived neurotropic factor and behavior in mice. Gastroenterology,
141(2), 599–​609, 609 e1–​3.
Berman, S. M., Naliboff, B. D., Suyenobu, B., Labus, J. S., Stains, J., Ohning, G., et al. (2008). Reduced
brainstem inhibition during anticipated pelvic visceral pain correlates with enhanced brain response
 97

References 97

to the visceral stimulus in women with irritable bowel syndrome. Journal of Neuroscience, 28(2),
349–​59.
Bishop, S., Duncan, J., Brett, M., and Lawrence, A. D. (2004). Prefrontal cortical function and
anxiety: Controlling attention to threat-​related stimuli. Nature Neuroscience, 7(2),  184–​8.
Blankstein, U., Chen, J., Diamant, N. E., and Davis, K. D. (2010). Altered brain structure in
irritable bowel syndrome: Potential contributions of pre-​existing and disease-​driven factors.
Gastroenterology, 138(5), 1783–​9.
Bonaz, B., Picq, C., Sinniger, V., Mayol, J. F., and Clarencon, D. (2013). Vagus nerve stimulation: From
epilepsy to the cholinergic anti-​inflammatory pathway. Neurogastroenterology and Motility, 25(3),
208–​21.
Botha, C., Farmer, A. D., Nilsson, M., Brock, C., Gavrila, A. D., Drewes, A. M., et al. (2015).
Preliminary report: Modulation of parasympathetic nervous system tone influences oesophageal
pain hypersensitivity. Gut, 64(4), 611–​17.
Browning, K. N. and Travagli, R. A. (2011). Plasticity of vagal brainstem circuits in the control of
gastrointestinal function. Autonomic Neuroscience, 161(1–​2),  6–​13.
Camilleri, M., Busciglio, I., Carlson, P., McKinzie, S., Burton, D., Baxter, K., et al. (2008). Candidate
genes and sensory functions in health and irritable bowel syndrome. American Journal of
Physiology–​Gastrointestinal and Liver Physiology, 295(2), G219–​25.
Carrive, P. and Bandler, R. (1991). Viscerotopic organization of neurons subserving hypotensive
reactions within the midbrain periaqueductal grey: A correlative functional and anatomical study.
Brain Research, 541(2), 206–​15.
Ceunen, E., Vlaeyen, J. W., and Van Diest, I. (2016). On the origin of interoception. Froniers in
Psychology, 7, 743.
Coen, S. J., Gregory, L. J., Yaguez, L., Amaro Jr., E., Brammer, M., Williams, S. C., et al. (2007).
Reproducibility of human brain activity evoked by esophageal stimulation using functional magnetic
resonance imaging. American Journal of Physiology, Gastrointestinal and Liver Physiology, 293, G188–​97.
Coen, S. J., Aziz, Q., Yaguez, L., Brammer, M., Williams, S. C., and Gregory, L. J. (2008). Effects of
attention on visceral stimulus intensity encoding in the male human brain. Gastroenterology, 135(6),
2065–​74, 2074 e1.
Coen, S. J., Kano, M., Farmer, A. D., Kumari, V., Giampietro, V., Brammer, M., et al. (2011).
Neuroticism influences brain activity during the experience of visceral pain. Gastroenterology,
141(3), 909–​17 e1.
Coen, S. J., Hobson, A. R. and Aziz, Q. (2012). Processing of Gastrointestinal Sensory Signals in the
Brain. In: L. R. Johnson (ed.), Physiology of the Gastronintestinal Tract, 5th edn. Oxford: Academic
Press, pp. 689–​702.
Colucci, R., Gambaccini, D., Ghisu, N., Rossi, G., Costa, F., Tuccori, M., et al. (2013). Influence of the
serotonin transporter 5HTTLPR polymorphism on symptom severity in irritable bowel syndrome.
PLoS One, 8(2), e54831.
Craig, A. D., Bushnell, M. C., Zhang, E. T., and Blomqvist, A. (1994). A thalamic nucleus specific for
pain and temperature sensation. Nature, 372(6508),  770–​3.
Craig, A. D. (2002). How do you feel? Interoception: The sense of the physiological condition of the
body. Nature Reviews Neuroscience, 3(8), 655–​66.
Craig, A. D. (2003). Pain mechanisms: Labeled lines versus convergence in central processing. Annual
Review of Neuroscience, 26,  1–​30.
Cryan, J. F. and O’Mahony, S. M. (2011). The microbiome–​gut–​brain axis: From bowel to behavior.
Neurogastroenterology and Motility, 23(3), 187–​92.
Cryan, J. F. and Dinan, T. G. (2012). Mind-​altering microorganisms: The impact of the gut microbiota
on brain and behaviour. Nature Reviews Neuroscience, 13(10), 701–​12.
98

98 The neurobiology of gut feelings

Dagher, A. (2009). The neurobiology of appetite: Hunger as addiction. International Journal of Obesity

(London), 33 Suppl. 2, S30–​3.
De Vadder, F., Kovatcheva-​Datchary, P., Goncalves, D., Vinera, J., Zitoun, C., Duchampt, A., et al.
(2014). Microbiota-​generated metabolites promote metabolic benefits via gut-​brain neural circuits.
Cell, 156(1–​2),  84–​96.
De Wandele, I., Rombaut, L., Leybaert, L., Van de Borne, P., De Backer, T., Malfait, F., et al. (2014).
Dysautonomia and its underlying mechanisms in the hypermobility type of Ehlers–​Danlos
syndrome. Seminars in Arthritis and Rheumatism, 44(1), 93–​100.
Derbyshire, S. W. (2003). A systematic review of neuroimaging data during visceral stimulation.
American Journal of Gastroenterology, 98(1),  12–​20.
Dominguez-​Bello, M. G., Blaser, M. J., Ley, R. E., and Knight, R. (2011). Development of the human
gastrointestinal microbiota and insights from high-​throughput sequencing. Gastroenterology,
140(6), 1713–​19.
Drossman, D. A. and Hasler, W. L. (2016). Rome IV-​functional GI disorders: Disorders of gut–​brain
interaction. Gastroenterology, 150(6), 1257–​61.
Dunckley, P., Wise, R. G., Fairhurst, M., Hobden, P., Aziz, Q., Chang, L., et al, . (2005). A comparison
of visceral and somatic pain processing in the human brainstem using functional magnetic
resonance imaging. Journal of Neuroscience, 25(32), 7333–​41.
Fairhurst, M., Wiech, K., Dunckley, P., and Tracey, I. (2007). Anticipatory brainstem activity predicts
neural processing of pain in humans. Pain, 128(1–​2), 101–​10.
Farmer, A. D., Coen, S. J., Kano, M., Paine, P. A., Shwahdi, M., Jafari, J., et al. (2013).
Psychophysiological responses to pain identify reproducible human clusters. Pain, 154(11), 2266–​76.
Farmer, A. D. and Aziz, Q. (2014). Mechanisms of visceral pain in health and functional
gastrointestinal disorders. Scandinavian Journal of Pain, 5(2),  51–​60.
Farmer, A. D., Coen, S. J., Kano, M., Naqvi, H., Paine, P. A., Scott, S. M., et al. (2014a).
Psychophysiological responses to visceral and somatic pain in functional chest pain identify
clinically relevant pain clusters. Neurogastroenterology and Motility, 26(1), 139–​48.
Farmer, A. D., Randall, H. A., and Aziz, Q. (2014b). It’s a gut feeling: How the gut microbiota affects
the state of mind. Journal of Physiology, 592(14), 2981–​8.
Farmer, A. D. and Ruffle, J. K. (2015). Irritable bowel syndrome. Hamdan Medical Journal, 8(3).
Farmer, A. D., Ban, V. F., Coen, S. J., Sanger, G. J., Barker, G. J., Gresty, M. A., et al. (2015). Visually
induced nausea causes characteristic changes in cerebral, autonomic and endocrine function in
humans. Journal of Physiology, 593(5), 1183–​96.
Farmer, A. D., Albu-​Soda, A., and Aziz, Q. (2016). Vagus nerve stimulation in clinical practice. British
Journal of Hospital Medicine (London), 77(11), 645–​51.
Frokjaer, J. B., Bergmann, S., Brock, C., Madzak, A., Farmer, A. D., Ellrich, J., et al. (2016).
Modulation of vagal tone enhances gastroduodenal motility and reduces somatic pain sensitivity.
Neurogastroenterology and Motility, 28,  592–​8.
Fukudo, S., Kanazawa, M., Mizuno, T., Hamaguchi, T., Kano, M., Watanabe, S., et al. (2009). Impact of
serotonin transporter gene polymorphism on brain activation by colorectal distention. Neuroimage,
47(3), 946–​51.
Fukudo, S. and Kanazawa, M. (2011). Gene, environment, and brain–​gut interactions in irritable bowel
syndrome. Journal of Gastroenterology and Hepatology, 26 Suppl 3, 110–​15.
Gregory, L. J., Yaguez, L., Williams, S. C., Altmann, C., Coen, S. J., Ng, V., et al. (2003). Cognitive
modulation of the cerebral processing of human oesophageal sensation using functional magnetic
resonance imaging. Gut, 52(12), 1671–​7.
Harkins, S. W., Price, D. D., and Braith, J. (1989). Effects of extraversion and neuroticism on
experimental pain, clinical pain, and illness behavior. Pain, 36(2), 209–​18.
 9

References 99

Hazlett-​Stevens, H., Craske, M. G., Mayer, E. A., Chang, L., and Naliboff, B. D. (2003). Prevalence
of irritable bowel syndrome among university students: The roles of worry, neuroticism, anxiety
sensitivity and visceral anxiety. Journal of Psychosomatic Research, 55(6),  501–​5.
Hoff, D. A., Brock, C., Farmer, A. D., Dickman, R., Ruffle, J. K., Shaker, A., and Drewes, A. M. (2016).
Pharmacological and other treatment modalities for esophageal pain. Annals of the New York
Academy of Sciences, 1380(1),  58–​66.
Holland, P. C. and Gallagher, M. (2004). Amygdala-​frontal interactions and reward expectancy. Current
Opinions in Neurobiology, 14(2), 148–​55.
Kano, M., Farmer, A. D., Aziz, Q., Giampietro, V., Brammer, M. J., Williams, S. C., et al. (2013).
Sex differences in brain activity to anticipated and experienced visceral pain in healthy subjects.
American Journal of Physiology–​Gastrointestinal and Liver Physiology, 304(8), G687–​99.
Kano, M., Coen, S. J., Farmer, A. D., Aziz, Q., Williams, S. C., Alsop, D. C., et al. (2014). Physiological
and psychological individual differences influence resting brain function measured by ASL
perfusion. Brain Structure and Function, 219(5), 1673–​84.
Kern, M. K. and Shaker, R. (2002). Cerebral cortical registration of subliminal visceral stimulation.
Gastroenterology, 122(2),  290–​8.
Kern, M., Chai, K., Lawal, A., and Shaker, R. (2009). Effect of esophageal acid exposure on the cortical
swallowing network in healthy human subjects. American Journal of Physiology–​Gastrointestinal and
Liver Physiology, 297(1), G152–​8.
Kilpatrick, L. A., Mayer, E. A., Labus, J. S., Gupta, A., Hamaguchi, T., Mizuno, T., et al. (2015).
Serotonin transporter gene polymorphism modulates activity and connectivity within an emotional
arousal network of healthy men during an aversive visceral stimulus. PLoS One, 10(4), e0123183.
Kringelbach, M. L. (2005). The human orbitofrontal cortex: Linking reward to hedonic experience.
Nature Reviews Neuroscience, 6(9), 691–​702.
Kulkarni, B., Bentley, D. E., Elliott, R., Youell, P., Watson, A., Derbyshire, S. W., et al. (2005). Attention
to pain localization and unpleasantness discriminates the functions of the medial and lateral pain
systems. European Journal of Neuroscience, 21(11), 3133–​42.
Kumari, V., ffytche, D. H., Das, M., Wilson, G. D., Goswami, S., and Sharma, T. (2007). Neuroticism
and brain responses to anticipatory fear. Behavioral Neurosciences, 121(4), 643–​52.
Labus, J. S., Naliboff, B. N., Fallon, J., Berman, S. M., Suyenobu, B., Bueller, J. A., et al. (2008). Sex
differences in brain activity during aversive visceral stimulation and its expectation in patients with
chronic abdominal pain: A network analysis. Neuroimage, 41(3), 1032–​43.
Labus, J. S., Vianna, E., Jarcho, J. M., Tillisch, K., Bueller, J. A., and Mayer, E. A. (2010). 858 reduced
structural connectivity between amygdala and prefrontal cortex in patients with irritable bowel
syndrome: A diffuse tensor imaging study. Gastroenterology, 138(5),  S-​118.
Lechan, R. M. and Toni, R. (2016). Functional Anatomy of the Hypothalamus and Pituitary. South
Dartmouth, MA: MDText.com.
Ley, R. E., Peterson, D. A., and Gordon, J. I. (2006). Ecological and evolutionary forces shaping
microbial diversity in the human intestine. Cell, 124(4), 837–​48.
Mayer, E. A., Berman, S., Suyenobu, B., Labus, J., Mandelkern, M. A., Naliboff, B. D., and Chang,
L. (2005). Differences in brain responses to visceral pain between patients with irritable bowel
syndrome and ulcerative colitis. Pain, 115(3), 398–​409.
Melzack, R. and Wall, P. D. (1965). Pain mechanisms: A new theory. Science, 150(3699),  971–​9.
Miller, A. D., Nonaka, S., and Jakus, J. (1994). Brain areas essential or non-​essential for emesis. Brain
Research, 647(2), 255–​64.
Moisset, X., Bouhassira, D., Denis, D., Dominique, G., Benoit, C., and Sabate, J. M. (2010).
Anatomical connections between brain areas activated during rectal distension in healthy
volunteers: A visceral pain network. European Journal of Pain, 14(2),  142–​8.
10

100 The neurobiology of gut feelings

Omran, Y. A. and Aziz, Q. (2014). Functional brain imaging in gastroenterology: To new beginnings.
Nature Reviews Gastroenterology and Hepatology, 11(9), 565–​76.
Paine, P., Kishor, J., Worthen, S. F., Gregory, L. J., and Aziz, Q. (2009a). Exploring relationships for
visceral and somatic pain with autonomic control and personality. Pain, 144(3), 236–​44.
Paine, P., Worthen, S. F., Gregory, L. J., Thompson, D. G., and Aziz, Q. (2009b). Personality differences
affect brainstem autonomic responses to visceral pain. Neurogastroenterology and Motility, 21(11),
1155–​e98.
Pezawas, L., Meyer-​Lindenberg, A., Drabant, E. M., Verchinski, B. A., Munoz, K. E., Kolachana, B.
S., et al. R. (2005). 5-​HTTLPR polymorphism impacts human cingulate-​amygdala interactions: A
genetic susceptibility mechanism for depression. Nature Neurosciences, 8(6), 828–​34.
Ploghaus, A., Narain, C., Beckmann, C. F., Clare, S., Bantick, S., Wise, R., et al. (2001). Exacerbation
of pain by anxiety is associated with activity in a hippocampal network. Journal of Neuroscience,
21(24), 9896–​903.
Ploner, M., Lee, M. C., Wiech, K., Bingel, U., and Tracey, I. (2011). Flexible cerebral connectivity
patterns subserve contextual modulations of pain. Cerebral Cortex, 21(3), 719–​26.
Rhee, S. H., Pothoulakis, C., and Mayer, E. A. (2009). Principles and clinical implications of the brain–​
gut -​enteric microbiota axis. Nature Reviews Gastroenterology and Hepatology, 6(5), 306–​14.
Ruffle, J., Shah, M., Monro, J., and Julu, P. (2015a). Pattern of dysautonomia in patients with functional
gastrointestinal disorders. Autonomic Neuroscience: Basic and Clinical, 192(Complete), 119.
Ruffle, J. K., Farmer, A. D., Kano, M., Giampietro, V., Aziz, Q., and Coen, S. J. (2015b). The influence
of extraversion on brain activity at baseline and during the experience and expectation of visceral
pain. Personality and Individual Differences, 74(0), 248–​53.
Ruffle, J. K., Coen, S. J., Giampietro, V., Williams, S. C. R., Farmer, A. D., and Aziz, Q. (2017a).
Higher baseline cardiac vagal tone implicates a subcortical functional brain network during acute
oesophageal pain. Neurogastroenterology and Motility, 29((Suppl. 2)),  47–​8.
Ruffle, J. K., Coen, S. J., Giampietro, V., Williams, S. C. R., Farmer, A. D., and Aziz, Q. (2017b). OC-​
054 Pain endophenotypes can be accurately predicted by whole brain connectivity during painful
oesophageal stimulation. Gut, 66(Suppl. 2), A28.
Ruffle, J. K., Coen, S. J., Giampietro, V., Williams, S. C. R., Apkarian, A. V., Farmer, A. D., et al.
(2018). Morphology of subcortical brain nuclei is associated with autonomic function in healthy
humans. Human Brain Mapping, January 39(1), 381–​92.
Sanger, G. J. and Andrews, P. L. (2006). Treatment of nausea and vomiting: Gaps in our knowledge.
Autonomic Neuroscience, 129(1–​2),  3–​16.
Schaub, N., Kano, M., Farmer, A. D., Aziz, Q., and Coen, S. J. (2013). Su2109 The Influence of the 5-​
HTTLPR Genotype on the cerebral processing of esophageal pain—​A pilot study. Gastroenterology,
144(5),  S-​560.
Seminowicz, D. A., Labus, J. S., Bueller, J. A., Tillisch, K., Naliboff, B. D., Bushnell, M. C., et al.
(2010). Regional gray matter density changes in brains of patients with irritable bowel syndrome.
Gastroenterology, 139(1), 48–​57 e2.
Sommer, F. and Backhed, F. (2013). The gut microbiota—​Masters of host development and physiology.
Nature Reviews Microbiology, 11(4), 227–​38.
Spaziani, R., Bayati, A., Redmond, K., Bajaj, H., Mazzadi, S., Bienenstock, J., et al. (2008). Vagal
dysfunction in irritable bowel syndrome assessed by rectal distension and baroreceptor sensitivity.
Neurogastroenterology and Motility, 20(4), 336–​42.
Spiller, R. and Lam, C. (2012). An update on post-​infectious irritable bowel syndrome: Role of genetics,
immune activation, serotonin and altered microbiome. Journal of Neurogastroenterology and
Motility, 18(3), 258–​68.
 10

References 101

Stein, J. L., Wiedholz, L. M., Bassett, D. S., Weinberger, D. R., Zink, C. F., Mattay, V. S., et al. (2007). A
validated network of effective amygdala connectivity. Neuroimage, 36(3), 736–​45.
Tack, J., Bisschops, R., and Sarnelli, G. (2004). Pathophysiology and treatment of functional dyspepsia.
Gastroenterology, 127(4), 1239–​55.
Tack, J., Broekaert, D., Fischler, B., Van Oudenhove, L., Gevers, A. M., and Janssens, J. (2006). A
controlled crossover study of the selective serotonin reuptake inhibitor citalopram in irritable bowel
syndrome. Gut, 55(8), 1095–​103.
Tillisch, K. and Labus, J. S. (2011). Advances in imaging the brain–​gut axis: Functional gastrointestinal
disorders. Gastroenterology, 140(2), 407–​11 e1.
Tillisch, K., Mayer, E. A., and Labus, J. S. (2011). Quantitative meta-​analysis identifies brain regions
activated during rectal distension in irritable bowel syndrome. Gastroenterology, 140(1), 91–​100.
Tracey, I., Ploghaus, A., Gati, J. S., Clare, S., Smith, S., Menon, R. S., et al. (2002). Imaging attentional
modulation of pain in the periaqueductal gray in humans. Journal of Neuroscience, 22(7), 2748–​52.
Tracey, I. and Bushnell, M. C. (2009). How neuroimaging studies have challenged us to rethink: Is
chronic pain a disease? Journal of Pain, 10(11), 1113–​20.
Tracey, I. (2011). Can neuroimaging studies identify pain endophenotypes in humans? Nature Reviews
Neurology, 7(3), 173–​81.
Van Oudenhove, L., Coen, S. J., and Aziz, Q. (2007). Functional brain imaging of gastrointestinal
sensation in health and disease. World Journal of Gastroenterology, 13(25), 3438–​45.
Winkelmann, T., Thayer, J. F., Pohlack, S., Nees, F., Grimm, O., and Flor, H. (2016). Structural brain
correlates of heart rate variability in a healthy young adult population. Brain Structure and Function,
222(2), 1061–​8.
102

Chapter 6

The cutaneous borders

of interoception: Active and social
inference of pain and pleasure on
the skin
Mariana von Mohr and Aikaterini Fotopoulou

6.1 Introduction
Humans are capable of conscious feelings that concern the state of the body, such as
pain, itch, muscular and visceral sensations, hunger, thirst, sexual desire, and air need.
The classification of such feelings, and particularly their relation to the more classical
sensory systems for vision, audition, and touch, as well as to emotions such as anger and
happiness, has been a matter of ongoing debate. Unlike sight, smell, and hearing that
have dedicated sensory organs, there are no dedicated bodily organs for position and
movement sense, pain, and many other modalities. Instead, developments in physics,
anatomy, and physiology since the nineteenth century have given rise to a wide interest
in mapping and classifying the senses with reference to criteria such as the nature of the
stimulus, anatomy and location of receptors across body parts, the pathways to and the
representation of the signal at the central nervous system (CNS), as well as the quality of
the experience. This interest led to a number of classifications of the senses; for example,
in exteroceptive (their receptive field “lies freely open to the numberless vicissitudes
and agencies of the environment” Sherrington, 1910, p.  132), interoceptive (sensory
receptors located within the body and primarily in the viscera), and proprioceptive
sensations (receptors in muscles, tendons, and joints detecting position and movement
of the body). Since this influential classification (see Ceunen, Vlaeyen, & Van Diest, 2016
for a review), exteroceptive and proprioceptive systems have received far more attention
than interoceptive modalities. However, as this volume exemplifies, this has changed
in the two last decades. On the one hand, theories and studies in affective neurosci-
ence (e.g. Damasio, 2010)  have brought to the foreground William James’s older idea
that interoceptive sensations may lie at the heart of our emotions and self-​awareness.
On the other hand, progress in anatomy and physiology has urged certain researchers
(e.g. Craig, 2002) to propose alternative classifications of the senses that include a more
encompassing definition of interoception as the sense of the physiological condition of
the entire body, not just the viscera.
 103

Introduction 103

Bud Craig’s proposal relies in synthesizing findings regarding the functional anatomy
of a lamina I spinothalamocortical pathway that is portrayed as the long-​missing afferent
complement of the efferent autonomic nervous system, underlying distinct, conscious,
affective bodily feelings such as cool, warm, itch, first (pricking) pain, second (burning)
pain, pleasant or sensual touch, muscle burn, joint ache, visceral fullness, flush, nausea,
cramps, hunger, thirst, and visceral taste (Craig, 2002). Specifically, he proposes that
the primate brain has evolved a direct sensory pathway to the thalamus that provides a
modality-​specific representation of various individual aspects of the physiological con-
dition of the body (interoception redefined; Craig 2002, 2003a). This pathway is thought
to originate in lamina I of the spinal dorsal horn and in the nucleus of the solitary tract
in the caudal medulla, and to represent the afferent inputs from sympathetic (somatic)
and parasympathetic nerves, respectively, and to terminate with a posterior-​to-​anterior
somatotopic organization in a specific thalamic structure (the posterior and basal parts of
the ventral medial nucleus, Craig 2002). He has further proposed that the functional role
of this pathway is to represent the sensory aspects of homeostatic emotions (Craig 2003a,
2008)  and their accompanying motivations (represented in anterior cingulate cortex)
that serve to maintain the body in relative stability despite ongoing internal and external
changes (e.g. variabilities in metabolic energy levels and the availability of food). This
proposal brings the concept of interoception into a tight relation to the notion of home-
ostasis (Cannon, 1929, see also Chapter 1 and Chapter 15 by Corcoran and Jakob Hohwy
in the present volume), so that interoception is the sensory representation of the phys-
iological condition of the whole body allowing homeostatic, and ultimately ‘allostatic’
control (i.e. self-​initiated temporary change in homeostatic imperatives to prepare for
a predicted external change). In other words, interoceptive signals provide information
regarding current homeostatic levels (e.g. reduced glucose levels in the blood), which
are used as motivations to steer action (e.g. ingest food to restore glucose levels). This
definition of interoception, which subsumes cutaneous pain, itch, and pleasant touch,
differs greatly from the classic association of these modalities with exteroception and
particularly discriminatory touch. Moreover, in addition to this “spinal pathway,” there
are also other proposed interoceptive pathways (Critchley & Harrison, 2013), and more
broad proposals regarding the role of higher order processing in interoception (Ceunen
et al., 2016).
We will here focus on pleasant touch and cutaneous pain, which are two interoceptive
(Craig, 2003a, 2003b) sub-​modalities of touch that have contrasting affective qualities
(pleasantness/​unpleasantness) and social meanings (care/​harm). Although the source of
the skin stimulation lies outside the body and the resulting sensations can be used to gain
information about how, where, and by what one is touched, we also assume that these
modalities are of fundamental homeostatic importance, signalling physiological safety
(i.e. the pleasantness of touch signifies a homeostatically safe environment in is contact
with the body) or threat (i.e. pain signifies the reverse) to the organism and leading to
certain behavioral and physiological reactions of homeostatic and allostatic significance.
In the present chapter, we first briefly outline the current literature on the peripheral
104

104 The cutaneous borders of interoception

and central neurophysiology of unpleasant, cutaneous pain and affective, pleasant touch.
Subsequently, we make use of recent neurocomputational theories of perception and
action, as applied to both exteroceptive and interoceptive modalities, to put forward a
unifying model of how bottom-​up and top-​down signals can be integrated to give rise
to these modalities. We speculate that the understanding of these modalities within the
Bayesian predictive coding framework of “active inference” (Friston, 2010) offers a unique
opportunity to unify various insights into a common framework that emphasizes partic-
ularly: (a) the deep interdependence between bottom-​up and top-​down mechanisms in
any modality; (b) the deep interdependence of perception and action in any modality;
(c) the special role of these modalities in homeostatic and allostatic control; and (d) the
particular relevance of social developmental factors in determining the salience of inter-
oceptive modalities such as pain and pleasant touch.

6.2  The peripheral and central neurophysiology

of cutaneous pain and pleasant touch
Surrogate animal models and human studies have revealed that nociceptors are distinc-
tive afferent units rather than the extremes of a single class of receptors with a continuum
of features (reviewed by Marks et al., 2006). While low-​threshold, mechanoreceptive, or
thermoreceptive afferent neurons cannot discriminate reliably between noxious and non-​
noxious (innocuous) stimulation, nociceptors can (Bessou et al., 1971). These two classes
of fibers also differ in their termination patterns in the spinal cord (Sugiura, Lee, & Perl,
1986), their membrane constituents (Caterina et al., 1997), and properties, including their
action potential shape (Ritter & Mendell, 1992). Broadly, nociceptors can be divided into
two types: A-​(most in the Aδ-​range) and C-​fibers, which are mediated by myelinated fast
(5–​30 m/​s) and unmyelinated slow (0.4–​1.4m/​s) conductive axons (Dubin & Patapoutian,
2010), corresponding to initial fast-​onset pain (sharp pain sensation) and slow second
pain (pervasive burning pain sensation), respectively. Nociceptors, particularly in mus-
culoskeletal tissue, have been mostly thought to be electrically ‘silent’, transmitting all or
no action potentials only when excited, and thus give rise to pain (Marks et al., 2006).
Different pathways on how the nociceptor is conveyed to the CNS have been suggested,
including different spinal neural features and their functional role. First, it is thought
that an afferent volley is produced upon activation of the nociceptor. The nociceptive
volley travels along the periphery and enters the dorsal horn of the spinal cord (Brooks &
Tracey, 2005) and mostly terminates in laminae I where they synapse with relay neurons
and local interneurons important for signal modification (see Dubin & Patapoutian, 2010
for the specific role of laminae I, IV, and V in relation to A-​and C-​fibers). Via spinal
ascending pathways, the relay neurons project to the thalamus and brainstem, which in
turn project to large distributed brain networks (Dubin & Patapoutian, 2010). However, a
different type of multimodal spinal neurons located deeper in the dorsal horn, namely the
wide dynamic (WDR) neurons, has also been implicated in nociceptive and pain-​related
mechanisms (Perl, 2007).
 105

The peripheral and central neurophysiology of cutaneous pain and pleasant touch 105

More generally, peripheral neurophysiological specificity does not seem to lead to a

direct relation between nociception and conscious pain perception. While the activa-
tion of nociceptors and nociceptive pathways can lead to pain (Marks et al., 2006), it is
also known that nociceptors can be active in the absence of pain perception and pain
can occur without known nociceptive activity. Indeed, there have been observations of
a lack of reported pain by soldiers during battle, despite severe injuries, as well as ex-
perimental evidence suggesting that pain perception varies with psychological state and
context (Head & Holmes, 1911; Melzack, Wall, & Ty, 1982). In fact, since the proposal of
the influential “gate control theory” (inhibition of nociceptive excitatory signalling at the
level of the spinal cord), and more recent insights regarding the heightened sensitivity of
afferent signals at the same level, known as “central sensitization,” it is widely accepted
that although much pain is a consequence of stimulation of peripheral nociceptors, the
CNS plays a major role in the processing of noxious sensations (Melzack & Wall, 1965).
Furthermore, more potent neuroscientific methods in recent decades have provided
corroborating evidence for the role of the brain in pain (Rainville et al., 1997; Ploghaus
et al., 1999; Ploner, Freund, & Schnitzler, 1999). For example, novel cortical stimulation
studies have qualified Penfield’s inability to detect ‘pain cortical areas’ (Mazzola et  al.,
2011). Moreover, functional neuroimaging studies indicate that noxious stimulation
involves large distributed brain networks (Brooks & Tracey, 2005; Talbot et al., 1991). The
so-​called pain matrix has been subdivided into a medial and lateral pain system, based
on their respective projection sites from the thalamic structures to the cortex. The lateral
pain system involves the S1 and secondary somatosensory cortex (S2) and is thought to
play a role in the sensory-​discriminative aspect of pain (i.e. where is the stimulus and how
intense it is), whereas the medial pain system, including areas such as the AAC, the in-
sula, and the amygdala, is thought to be involved in the affective-​cognitive aspect of pain.
However, the insular cortex may play a role in facilitating the integration of information
between the lateral and medial pain systems (Brooks & Tracey, 2005) and some studies
suggest that the functional role of these areas may not be pain-​specific but rather relating
to the processing of all sensory-​salient events (see Legrain et al., 2011). Nevertheless, it
is assumed that a top-​down descending circuitry modulates ascending nociceptive infor-
mation and consequently, influences pain perception. Hence, scientific and health organ-
izations such as the International Association for the Study of Pain stress the difference
between nociception and pain. However, debates regarding the bottom-​up versus the top-​
down contributions to pain and their corresponding definitions remain. Similar debates
surround the study of pleasant touch.
Recent research suggests that slow conducting unmyelinated (non-​ nociceptive)
afferent CT fibers mediate the affective, pleasant component of touch. These C-​fiber tac-
tile afferents were first identified in a cat in 1939 by showing low spike heights using
the skin-​nerve preparation technique (Zotterman, 1939). More recently, low threshold
mechanosensitive C-​fibers (C-​LTMs; detected by cutaneous sensory neurons, i.e. C-​low
threshold mechanoreceptors, C-​LTMRs) have been found in the hairy skin of rodents and
primates (Bessou et al., 1971). C-​LTMs are now acknowledged also to exist in human skin,
106

106 The cutaneous borders of interoception

termed CT afferents (Vallbo, Olausson, & Wessberg, 1999). CTs have different character-
istics than myelinated fast conducting Aβ-​fibers associated with discriminative touch,
including their conduction axon velocity (0.6–​1.3 m/​s) and skin location (i.e. found in
hairy but not glabrous skin).
Microneurography studies have shown that CTs are highly sensitive mechanoreceptors
responding to stimuli that are clearly innocuous, and their firing rate seems to be dis-
tinct from myelinated afferents, reflecting an inverted U-​shaped relationship between the
stroking velocity and mean firing rate with the most vigorous responses being at 1–​10 cm/​
s (Löken et  al., 2009). Moreover, subjective responses of perceived pleasantness in re-
sponse to stroking also showed an inverted U-​shape relationship, with the highest pleas-
antness responses found at 1–​10 cm/​s stroking velocities (Löken et al., 2009), indicating
that CT afferents may carry a positive hedonic quality. Furthermore, CTs are also tem-
perature sensitive (i.e. preferentially discharged ≈ 32°C, the typical skin temperature;
Ackerley et  al., 2014). However, one main difficulty in our understanding of selective
CT stimulation is related to the fact that to date, we cannot stimulate CT fibers without
stimulating Aβ-​fibers in healthy subjects. Nevertheless, insights have been provided from
patients with sensory neuropathy, as these patients are thought to lack Aβ afferents while
their CTs afferents remain intact (Olausson et al., 2002, 2008). Research has shown that
CT stimulation in these patients activates the insula (i.e. the preferential cortical target for
CT afferents), but not somatosensory regions associated with the sensory discriminative
processing of touch (Olausson et al., 2002). Moreover, these patients were able to detect,
although poorly, slow brushing on the forearm (where CTs are abundant; Olausson et al.,
2008). Given the sensory discriminative properties associated with Aβ-​fibers and the lack
thereof in these patients, it is possible to presume that CT afferents may follow a separate
neurophysiological route than Aβ mediated discriminative touch (Olausson et al., 2008).
Unfortunately, our knowledge of how CTs peripheral information reaches spinal, brain-
stem, and cortical areas in humans remains scarce, yet meaningful insights regarding the
spinal processing of CTs have been obtained from animal studies. Mice studies suggest
that C-​LTMS enter the laminae II of the dorsal horn, with axons arborizing in lamina
I, where they would synapse with secondary afferent neurons (reviewed by McGlone,
Wessberg, & Olausson, 2014). Secondary afferent neurons then project to higher centers
such as the insula via spinal pathways (Andrew, 2010). Furthermore, as with pain, there
could be different classes of spinal neurons responsive to gentle touch, including WDR
neurons (Andrew, 2010). Finally, yet controversially, recent findings using mice ge-
netic tools indicate the dorsal horn as the key initial focus for integration of Aβ and C-​
LTMRs (Abraira & Ginty, 2013). Together, these lines of inquiry suggest that there may
be different pathways through which CT peripheral information is conveyed to higher
centers, although these pathways may likely vary across species.
Similar to pain, neuroimaging research has shown that gentle stroking activates the
posterior superior temporal sulcus, medial prefrontal cortex, orbitofrontal cortex (OFC),
and ACC, which are typically implicated in the cognitive-​affective aspects of pleasant
touch (Gordon et al., 2013). Further, while investigating the cortical areas that represent
 107

An integration: The predictive, active and social components of pain and pleasant touch 107

pleasant touch, painful touch, and neutral touch, studies have also found increased ac-
tivity in the OFC in response to pleasant and painful touch, highlighting the role of the
OFC on the affective aspects of the touch. In contrast, the somatosensory cortex was
less activated by pleasant and painful touch, relative to neutral touch (Rolls et al., 2003;
see also Gordon et al., 2013; Olausson et al., 2002). These studies suggest that CT-​based
touch may not be involved in the discriminative aspects of touch. Importantly, slow gentle
touch on CT skin has also been shown to preferentially activate the insula (Olausson
et al., 2002; Gordon et al., 2013), although the insula also plays a critical role in integrating
sensory-​discriminative and affective-​cognitive aspects of the touch (McCabe et al., 2008;
Rolls, 2010).

6.3  An integration: The predictive, active, and social

components of pain and pleasant touch
The history of the study of pain, and more recently of pleasant touch can be said to be
steeped in the debates between bottom-​up, neurophysiological specificity at the periphery
versus top-​down convergence and gating at the spinal and brain levels. In this section, we
make use of a Bayesian, predictive coding framework, namely the Free Energy Principle,
also referred to as “active inference” (Friston, 2010) to put forward a unifying model of
how bottom-​up and top-​down signals can be integrated to give rise to affective, pleasant
touch and unpleasant, cutaneous pain.

6.3.1  Action–​perception loops and the control

of physiological states
Recent neurocomputational theories of perception and action assume that the brain is an
organ that learns and self-​improves a generative model of the organism and its environ-
ment based on sensory signals and action (Friston, 2010). A basic tenet of such accounts
is that perception is an active process, whereby top-​down mechanisms are activated to
make predictions about the upcoming bottom-​up sensory signals. Thus, perception is an
inferential process, whose aim is to minimize prediction errors or the difference between
top-​down hypotheses about the most likely causes of sensations (termed “empirical prior
beliefs”) and current sensations. Recurrent message passing among several levels of the
sensorimotor hierarchy allows the suppression of (small or irrelevant) prediction errors by
priors, or the adjustment of (empirical) prior expectations by (large or highly salient) pre-
diction errors. Furthermore, the relative influence of predictions versus prediction errors
across several layers in this hierarchical organization is determined by the weighting (pre-
cision) of predictions versus prediction errors at each level. Precision can be regarded
as a measure of signal-​to-​noise ratio or confidence, or mathematically, as the inverse
variance, uncertainty, or reliability of a signal (Feldman & Friston, 2010; Friston et al.,
2012). Uncertainty is thought of as encoded mainly by neuromodulations of synaptic gain
(such as dopamine and acetylcholine) that encode the precision of random fluctuations
about predicted states —​the context in which sensory data is encountered (Quattrocki &
108

108 The cutaneous borders of interoception

Friston, 2014). For example, cholinergic or dopaminergic neuromodulatory mechanisms

can optimize the attentional gain of populations encoding prediction errors, so that
greater attention is allocated to certain salient events in the environment, influencing the
relative weighting or importance of prediction errors.
Importantly, prediction errors can also be minimized through action. At the simplest
control loop level, peripheral reflexes are engaged to suppress proprioceptive predic-
tion errors (Feldman & Friston, 2010), generated by comparing primary afferents from
receptors in muscles, tendons, and joints with proprioceptive predictions regarding body
position that descend to alpha motor neurons in the spinal cord and cranial nerve nu-
clei. Thus, action is driven by such predictions rather than descending motor commands.
Ultimately, action is seen as a prediction-​driven tendency to re-​sample the world to gen-
erate more sensory evidence for one’s predictions (active inference). Critically, the or-
ganism could solve a discrepancy between prediction and error (e.g. unexpected noxious
stimulation) by either changing its predictions (effectively convincing oneself that one is
not in pain) or by generating protecting action (moving to avoid the noxious source of the
prediction error). Both of these can be adaptive depending on the magnitude as well as
the context of the noxious stimulation and hence, their relation needs to be optimized by
weighting in each case. This framework emphasizes the tight interconnection of percep-
tion and action as well as the fundamental integration of bottom-​up and top-​down factors
in all perceptual and active inference.
Recently several proposals have applied this framework to interoception (Paulus &
Stein, 2006; Barrett & Simmons, 2015; Gu et al., 2013; Pezzulo, Rigoli, & Friston, 2015;
Seth, 2013; Seth, Suzuki, & Critchley, 2011), and by extension to the concepts of “homeo-
static” and “allostatic” control (see also Chapter 15 in the present volume). “Homeostasis”
(Cannon, 1929)  refers to the maintenance of a relative stability in one’s physiological
states despite ongoing internal and external changes. “Allostasis” refers to the idea that
physiological changes need to be anticipated by adaptive changes and choices across
different spatial and temporal scales, for example adjusting one’s metabolic needs in cer-
tain environments where foraging is dangerous (Sterling & Eyer, 1988). In predictive
coding frameworks, both homeostatic and allostatic control can be cast formally as active
inference (e.g. Pezzulo et  al., 2015; Stephan et  al., 2016). Homeostatic control enslaves
reflexes to produce corrective actions that fulfill beliefs about bodily states, and allostatic
control entails changing homeostatic beliefs under guidance by higher predictive models
about future perturbations of bodily states.
Moreover, as in the case of exteroceptive perception and action, the balance between
homeostatic and allostatic regulation rests upon the precision (i.e. weighting) placed
in deeper expectations about the organism and its environment. For example, during
conditions of bodily threat or psychological stress, such as the anticipated pain from
a sharp object approaching one’s face, noxious signals on one’s body may induce low-​
level proprioceptive predictions that mobilize withdrawal movements away from the
source of the stimulation. However, high-​precision predictions at a higher level of the
neurocognitive hierarchy may indicate that the source of the noxious stimulation is
 109

An integration: The predictive, active and social components of pain and pleasant touch 109

actually our dentist, and then predictions of tooth pain can be fulfilled without engaging
low-​level motor reflexes and instead engage allostatic changes in the form of updated
beliefs about the “safety” and tolerance (i.e. attenuated pain) of nociceptive signals in this
context, in order to ensure future pain-​free and healthy teeth.

6.3.2  Active, interoceptive inference, and feelings on the skin

Despite these proposals of interoceptive predictive coding, there is currently no direct
evidence for the proposal that interoceptive predictions, prediction errors, and their re-
lation rest on a common neurocomputational framework (for a first step, see Kleckner
et al., 2017). There are, however, ample circumstantial findings in the pain and pleasant
touch literature that can be cast in this light and importantly, the framework can allow
some specific predictions regarding the nature of pain and pleasant touch, and their mod-
ulation by cognitive and social factors, that we will focus on here.
First, this framework suggests that peripheral signals, such as nociceptive and CT
tactile channels, do not cause homeostatic perceptions or emotions (e.g. pain or the
affectivity of touch), or vice versa. Instead, there is a circular and multi-​layered cau-
sality, where on one end of the neural hierarchy, neuronally encoded predictions about
bodily states, including in this case states of the skin, engage autonomic, somatic, and
motor reflexes in a top-​down fashion. On the other end of the hierarchy, specialized skin
organs and their spinal cord circuitry carry interoceptive signals in a bottom-​up way that
informs and updates predictions at the levels above. These aspects can be linked to the
cognitive and sensory aspects of pain, respectively. Moreover, the affective component
of pain or touch can be seen as an attribute of the weighting (precision, see earlier) of
any representation that generates predictions and prediction errors about the physio-
logical state of the skin (see also Ainley et al., 2016; Fotopoulou, 2013). In other terms,
the subjective feelings of pain or pleasant touch can be linked to the neuromodulatory
weighting of the corresponding sensory prediction errors in relation to higher-​order
predictions regarding these sensory states. Typically, the optimization of precision is
linked with the function of neuromodulators in the brain (see section 6.3.1) but similar
processes of synaptic gain modulation have long been described in the spinal cord, par-
ticularly in the context of pain (see section 6.2.). We have previously proposed that in
interoceptive modalities, optimizing the precision of internal body signals can be seen
as optimizing interoceptive sensitivity and related feelings in perceptual inference (see
also Ainley et al., 2016; Fotopoulou, 2013). We propose here that concepts such as “pre-
cision” and its reverse, uncertainty, relate to the affective, conscious components of pain
and pleasant touch. The intensity of painful or pleasurable aspects of touch can be thus
understood as our sensitivity to such tactile, interoceptive signals in a given context (e.g.
a measurement of our subjective pain threshold in the lab) and our corresponding be-
havioral tendency to approach the world to gather more information (uncertainty) or to
avoid resampling (the certainty of pain and pleasant touch). This view can offer a new
integration of previous theories of pain and hence potentially also pleasant touch, as we
specify in the following.
10

110 The cutaneous borders of interoception

Specifically, classic theories may view cutaneous pain and the affectivity of touch as
signals of danger or safety to the organism respectively, starting in the periphery and
reaching consciousness if a “threshold” is surpassed at the spinal cord level, allowing the
brain to “read” them as pain or pleasure. For example, this threshold may be equated to
the “gate control theory” (Melzack & Wall, 1965) or more modern “central sensitization”
theories, where the gain of the spinal cord nociceptive synapse is amplified and hence
“travels up” the hierarchy to elicit conscious pain (Woolf & Salter, 2000). On the contrary,
more “active,” alternative theories of pain suggest that acute pain is not a warning signal
but rather is the failure of the “aversion” machinery (nociceptor activity) designed to op-
erate unconsciously in order to avoid harm and ultimately also conscious pain (Baliki &
Apkarian, 2015). In such accounts, most nociceptive activity is designed to remain “sub-
conscious” and protect the organism from harm without necessarily eliciting conscious
pain. Conscious pain instead only emerges when subconscious pain is converted to con-
scious pain in subcortical areas in the brain. In such accounts, it is conscious pain that has
the capacity to modulate spinal nociceptive sensitivity and thus actively determine “gate
control” and/​or “central sensitization” spinal nociceptive processes, mediated through de-
scending pathways (Vera-​Portocarrero et al., 2006).
From the point of view of active inference models, these are not competing but supple-
mentary views. Allostatic control is an extension of homeostatic control and they both
work to minimize prediction errors. Thus, these two perspectives can be integrated in
the following way, illustrated here with specific reference to cutaneous pain and pleasant
touch. For homeostatic control purposes, the organism entails (in an embodied manner)
a set of inherited prior expectations of the state of the skin. Any stimulation of the skin
that deviates from the range of such predicted states generates a prediction error. This
prediction error is corrected in simple, unconscious loops, by reflexive motor or auto-
nomic reactions that fulfil the initial beliefs about the state of the skin. If, however, these
“homeostatic corrections” fail (i.e. the prediction error persists), then the prediction error
travels up the hierarchy to generate posterior beliefs (updated predictions) at the above
hierarchical level. These updated beliefs act as priors towards future positive or nega-
tive events, thus attempting to anticipate and avoid danger, or anticipate and approach
pleasure, before these occur (allostasis). Specifically, more complex, generative, predictive
models of the organism’s needs are better able to predict stimuli at the levels below and
at different time-​scales and hence ‘suppress’ any future, anticipated prediction errors at
the level below by guiding autonomic function and action more effectively and under
the control of higher-​order predictive models. Please note that these homeostatic and
allostatic control operations are understood to be processes of unconscious inference for
the most part, so conscious feelings of skin pain and pleasure are not necessary for such
processes. This conclusion, however, raises the question of why should we have conscious
feelings such as pain and pleasure, if we can predict and control our sensations uncon-
sciously? We speculatively propose that it is important that the organism registers the
core feelings that relate to the specificity of innate, homeostatic needs (in this case safe or
dangerous contact on the skin), so that the cognitive resources available for scanning the
 1

An integration: The predictive, active and social components of pain and pleasant touch 111

world and the body for novelty and salience are always constrained by, and in competi-
tion with, the high precision of our innate expectations. In other terms, conscious pain
and pleasant touch are there to ensure that we do not habitually update, or ignore our
predictions about what is safe versus dangerous for the skin.
Interestingly, although these two modalities, pain and unpleasant touch, appear oppo-
site in hedonic content and behavior tendencies towards their particular sensory stimulus
(i.e. avoidance versus approach), from the point of view of the certainty–​uncertainty axis
described here, they are of similar characteristics. The greater the pain, or the felt pleasure
of touch, the more one’s attention and behavior is captured in the experience and the less
one is likely to engage in active, exploration of new sensations. Instead, the organism’s
resources are focused on controlling or escaping pain, and enjoying or prolonging the
feelings of pleasant touch. This view goes against the intuitive, long-​standing view of core
affective consciousness, pain and pleasure, as monitoring hedonic quality. Instead, the
core quality of affective consciousness is a kind of certainty–​uncertainty, or disambigu-
ation principle (Fotopoulou, 2013). Pain and pleasant touch therefore are a measure of
how important is for a given organism, in a given context, to be “certain” about what was
predicted versus what occurred.
This view of the conscious feelings of pain and pleasant touch tallies with long-​standing
insights regarding the dissociation between sensory and affective aspects of pain and,
more recently, pleasant touch, as well as with the fact that the physiology of nociception
has a well-​known specificity at the periphery which is not mirrored at the brain (see
section 6.2). The unique feeling qualities of painful or pleasant touch may be associated
with the CNS’s capacity for synaptic gain modulation and large-​scale integration of in-
formation arising from the body and the world in different time-​scales. This is consistent
with the fact that no single area or network in the brain has been reliably associated with
the conscious perception of pain (Baliki & Apkarian, 2015). Instead, the various networks
that have been associated with pain and its modulation, and with pleasant touch and
its modulation, are not only common to these two modalities, but seem relevant to the
processing of the salience of any sensory modality (Legrain et al., 2011). Indeed, several
recent neuroimaging studies have included such areas and their observed functional con-
nectivity in various hypothesized “salience networks” (Legrain et al., 2011; Medford &
Critchley, 2010; Wiech et al., 2010). For instance, predictive signals from such a “salience
network” process and integrate information about the significance of an impending nox-
ious stimulus and determine whether or not such a stimulus will be consciously perceived
as painful (Wiech et al., 2010).
More generally, a plethora of neuroimaging studies have shown that cognitive, affective,
and social factors modulate our perception of cutaneous pain, with emerging evidence
also making a case for these factors modulating the pleasantness of CT-​optimal touch. For
example, expectations may help an individual to adjust sensory, cognitive, and motor sys-
tems in order to process the noxious stimuli in terms of neural and behavioral responses
optimally (Wiech, Ploner, & Tracey, 2008; see also Villemure & Brushnell, 2002 for re-
view). Most consistently with the present proposal, expectations in which there is a high
12

112 The cutaneous borders of interoception

level of certainty regarding the stimulus may activate descending control systems to at-
tenuate pain, whereas in contrast, uncertainty may increase pain (Ploghaus et al., 2003).
Although there is less evidence on the neural mechanisms underlying the cognitive and
social factors that modulate pleasant CT-​optimal touch, studies suggest that a person’s
beliefs about the stimulus (McCabe et al., 2008) or the person (Ellingsen et al., 2015) pro-
viding the pleasant touch influences the perceived pleasantness of the touch. We use the
example of the social modulation of pain in the following to unpack and illustrate these
ideas further.

6.3.3  The
mentalization of nociception and CT stimulation:
homeostatic and allostatic control by proxy
The long-​observed fact that conscious pain is modulated by social context has received
experimental support in recent years (see Krahé et al., 2013 for review). In the last decade,
similar observations have also been made regarding the modulation of pleasant touch
by social context. In this section, we will apply these insights from the active inference
framework to propose some mechanisms by which this social modulation takes place.
This application has the advantage that it can provide a mechanistic, unified account of
the relation between bottom-​up (e.g. neurophysiological) and top-​down (e.g. psychoso-
cial) influences on homeostasis and allostasis. Existing biopsychosocial models of pain
offer similar insights but the current model has the advantage of offering direct links be-
tween these different bottom-​up and top-​down determinants of pain and pleasant touch
instead of treating them as merely additive variables.
Specifically, we propose that the perception of the social environment of pain or pleasant
touch can affect inferential processes about the perception of these modalities, as well
as related active tendencies, by influencing the certainty or precision of an individual’s
predictions about an impending stimulus versus the certainty or precision of related
prediction errors. As mentioned earlier, top-​down predictions do not represent just the
content of lower level representations but also predict their context, defined in mathemat-
ical terminology as the precision of a probability distribution (inverse variance or uncer-
tainty). For example, the allocation of attention toward specific events can optimize their
salience and ultimately influence the relative weighting or importance of prediction errors
against predictions. This kind of top-​down prediction in sensory cortices is thought to be
mediated by cholinergic neuromodulatory mechanisms that optimize the attentional gain
of populations encoding prediction errors (Feldman & Friston, 2010), as well as by dopa-
mine in fronto-​striatal circuits (Fiorillo, Tobler, & Schultz, 2003) and by neuropeptides
such as oxytocin in social contexts (Quatrokki & Friston, 2014). In interoception, precision
may relate to attention to signals from the body or interoceptive sensitivity (Ainley et al.,
2016; Fotopoulou, 2013) and may be modulated by several contextual factors. Therefore,
factors such as active social support or empathy may modulate pain or pleasant touch by
changing the precision of top-​down predictions about nociception or CT stimulation. In
such social contexts, individuals have learned to anticipate social support and thus the
optimization of the weight allocated to bottom-​up signals versus top-​down predictions
 13

An integration: The predictive, active and social components of pain and pleasant touch 113

maybe different than in conditions of experiencing similar stimuli alone, or in hostile

environments. For example, in previous studies we have shown that the administration of
intranasal oxytocin versus placebo, or the provision of high versus low empathy, or social
support may modulate the subjective, behavioral and neural responses to noxious stimu-
lation (Hurter et al., 2014; Krahé et al., 2015; Paloyelis et al., 2016). More generally, based
on a systematic review of the experimental pain literature (Krahé et al., 2013), we have
concluded that precision modulation by interpersonal interactions takes several forms,
including two main categories: (a) social signals about the safety or threat and thus the
salience of the impending stimulus itself, and (b) social signals about the threat or safety
and thus the salience of the environment in which the stimulus occurs. In turn, the per-
ception and interpretation of such interpersonal variables themselves may in turn depend
on (a) their own salience, as well as (b) an individual’s prior beliefs about interpersonal
relating and associated behaviors (see following for further details).
This notion of social modulation as precision modulation is compatible with pre-
vious theories such as the social baseline theory, which proposes that the presence of
other people helps individuals to conserve metabolically costly somatic and neural re-
sources through the social regulation of emotion (Beckes & Coan, 2011; see also Decety
& Fotopoulou, 2015). Integrating such notions within a predictive coding model has the
advantage of placing them in a wider and neurobiologically plausible framework and
hence integrating findings across many fields, as well as generating novel hypotheses, as
we outline here.

6.3.3.1  Developmental considerations: The social origins

of interoceptive inference
The active inference framework we propose allows us to observe that the social modu-
lation of pain and pleasant touch is not a simple “add-​on” in our understanding of such
modalities. Rather, it appears that interpersonal interactions are necessary in shaping all
interoceptive modalities from the onset. This claim is supported by several observations
(presented in detail elsewhere, Fotopoulou & Tsakiris, 2017), the most important of which
we outline here: namely, in early infancy, when the human motor system is not yet de-
veloped, interoceptive function and homeostasis are wholly dependent on embodied
interactions with other bodies. Action and perception do not mature at the same time. As
human infants are born without a fully matured motor system, and hence they cannot reg-
ulate their own homeostasis unaided, the actions of their caregivers necessarily determine
how they come to update their beliefs by active inference and ultimately how they experi-
ence all their sensations and particularly those requiring purposeful actions. For example,
young infants cannot position, balance, feed, thermoregulate, or protect themselves from
accidentally cutting or burning their skin (beyond some reflexive avoidance movements).
Thus, in the case of these interoceptive modalities, no available movement on the part of
the infant alone can change certain key neurophysiological states relating to homeostasis and
allostasis. As such, the young infant cannot use action to collect evidence about the causes
of interoceptive experience to test its interoceptive predictions against the world.
14

114 The cutaneous borders of interoception

Instead, infants use autonomic and motor reflexes in response to unpredicted physi-
ological states (e.g. crying when hypothalamic function detect that glucose level are not
within the predicted viable range) to elicit caregivers’ actions that can change the infant’s
physiological state (e.g. by feeding it) until the homeostatic needs are met (i.e. glucose
levels are within the predicted range). Thus, updating interoceptive predictions in infants
(close the action–​perception loop) includes multisensory signals regarding the reaction of
caregivers to infants’ initial autonomic and proprioceptive predictions; a process we have
termed as the “mentalization” of physiological states elsewhere; Fotopoulou & Tsakiris,
2017). In other words, the origins of interoceptive active inference are always by necessity
social, and thus core subjective feelings such as hunger and satiation, pain and relief, cold
or warmth, have actually social origins.

6.3.3.2  Adult predictions about the role of others in pain and

pleasant touch
These conclusions about the social origins of interoceptive feelings such as pain and
pleasant touch are also consistent with the literature on the relation between these
modalities and social attachment (see also Panksepp, 1998). As we mentioned earlier, pain
and pleasant touch may be modulated by social factors. In turn, the perception and inter-
pretation of social variables themselves may depend on individual prior beliefs, or gener-
ative models about interpersonal relating and associated behaviors. One influential way
of conceptualizing prior beliefs about relating to others is attachment theory. Attachment
theory posits that from early in life, attachment partners can serve as a “secure base”
from which the infant explores the world (Bowlby, 1969). If a secure attachment bond is
formed over repeated instances of responsive caregiving, the “secure base” signals safety
to the infant, while insecure bonds lead to more ambivalent or even threatening signals
from others. These bonds lead to the formation of attachment styles, which remain rela-
tively stable into adulthood.
Individual differences in attachment style have been linked directly with the perception
of pain and related reactions (e.g. Hurter et al., 2014; Meredith, Ownsworth, & Strong,
2008; Sambo et al., 2010). Moreover, in the clinical pain literature, insecure attachment
has been proposed as a vulnerability factor for developing chronic pain (Meredith et al.,
2008), supporting its importance as a pain-​relevant prior. In a series of pain studies by
our lab, we have shown that differences in attachment style influence the effects of inter-
personal variables on subjective, behavioral, physiological and neural responses to pain.
For example, social contextual factors and individual differences in attachment style de-
termine the amount of subjective report, facial expressions, heart-​rate, skin conductance
and neural responses people show in response to experimental pain (Sambo et al., 2010;
Hurter et al., 2014; Krahe et al., 2015, 2016).
Using laser-​evoked potentials (LEPs) we further found that pleasant touch, as a form
of active social support, can reduce both subjective and neural pain-​related outcomes
(Krahe et al., 2016; von Mohr et al., under review). However, contrary to other neuro-
imaging studies on passive forms of social support between couples (e.g. Coan, Schaefer,
 15

Conclusion 115

& Davidson, 2006; Eisenberger et al., 2011), our neural effects indicate that the effects of
active support by one’s romantic partner may begin at earlier stages of cortical nociceptive
processing, as reflected by changes in the N1 local peak amplitude (von Mohr et al., under
review). The N1 component is thought to reflect pre-​perceptual sensory response (out-
side of conscious awareness), with activation in the operculoinsular and primary somato-
sensory cortex (Garcia-​Larrea, Frot, & Valeriani, 2003; Valentini et al., 2012). Given that
LEPs have been recently proposed to detect environmental threat to the body in response
to sensory salient events (Legrain et al., 2011), we speculate that pleasant touch by one’s
romantic partner seems to reduce the sensory salience of impending noxious stimulation.
Similar findings have been reported in relation to the perception of pleasantness
and attachment style in response to CT-​optimal touch (Krahé et  al., under review).
Nevertheless, the field of pleasant touch is still at its infancy and further neuroscientific
studies are needed in both humans and other animals before firm conclusions can be
drawn about the social nature of this modality.

6.4 Conclusion
Cutaneous pain and pleasant touch have been recently classified by some researchers as
interoceptive modalities, even if their stimulation site lies outside the body. This reclassi-
fication is the basis of a more encompassing definition of interoception itself as the sense
of the physiological condition of the entire body, not just the viscera. However, this reclas-
sification lies at the heart of long-​standing debates regarding the nature of such modalities
and particularly the question of whether they should be defined as sensations on their
basis of their bottom-​up, neurophysiological specificity at the periphery or as homeo-
static emotions on the basis of top-​down convergence and modulation at the spinal and
brain levels. In the present chapter, we speculatively recast this current state of knowl­
edge within a recent Bayesian predictive coding framework of brain function, namely the
active inference model. This framework suggests that peripheral signals, such as nocicep-
tive and CT tactile channels, do not cause homeostatic perceptions or emotions (e.g. pain
or pleasant touch), or vice versa. Instead, there is a circular and multilayered causality,
where on one end of the neural hierarchy, neuronally encoded predictions about bodily
states, including in this case states of the skin, engage autonomic, somatic, and motor
reflexes in a top-​down fashion. At the other end of the hierarchy, specialized skin organs
and their spinal cord circuity carry interoceptive signals in a bottom-​up way that informs
and updates predictions at the levels above. These aspects can be linked to the cognitive
and sensory aspects of pain, respectively. The affective component of pleasant or painful
touch is a third component of this circular causality. Such affects are an attribute of the
optimisation of the weighting (precision) of any representation that generates predictions
and prediction errors about the physiological state of the skin (see also Ainley et al., 2016;
Fotopoulou, 2013). This weighting is further not only determined by such specialized
modalities and pathways but also necessarily contextualized by concurrent propriocep-
tive signals, as well as by concurrent exteroceptive cues about the body itself and about
16

116 The cutaneous borders of interoception

the physical, material, and social environment currently and across different time-​scales
(for allostatic control purposes). The painful or pleasurable aspects of touch can be thus
understood as our sensitivity to bottom-​up signals in given interoceptive, exteroceptive,
cognitive, social, and time contexts and our corresponding behavioral and anticipatory
tendencies.
These assumptions have received some empirical support in adult studies from our
lab, as well as many other labs, that show that “on-​line” social factors such as active social
support or empathy, as well as “off-​line” predictions about the availability of social help
(e.g. individual differences in attachment style), may modulate pain or pleasant touch by
changing the precision of top-​down predictions versus prediction errors from nociception
or CT stimulation. Finally, such claims are supported by the developmental observation
that in early infancy, when the human motor system is not yet developed, interoceptive
function and homeostasis are dependent on embodied interactions with other bodies. It
is the adult’s actions that will generate changes in interoceptive states and hence ultimately
close the action–​perception loop. Thus, the origins of interoceptive active inference are
always, by necessity social, and core subjective feelings, such as hunger and satiation, pain
and relief, cold or warmth, actually have social origins.

Acknowledgments
This work was supported by the European Research Council (ERC) Starting Grant ERC-​
2012-​STG GA313755 (to AF) and the National Council on Science and Technology
(CONACyT-​538843) scholarship (to MVM).

References
Abraira, V. E. and Ginty, D. D. (2013). The sensory neurons of touch. Neuron, 79(4), 618–​39.
Ackerley, R., Wasling, H. B., Liljencrantz, J., Olausson, H., Johnson, R. D., and Wessberg, J. (2014).
Human C-​tactile afferents are tuned to the temperature of a skin-​stroking caress. Journal of
Neuroscience, 34(8), 2879–​83.
Ainley, V., Apps, M. A., Fotopoulou, A., and Tsakiris, M. (2016). “Bodily precision”: A predictive
coding account of individual differences in interoceptive accuracy. Philosophical Transactions of the
Royal Society B, 371(1708), 20160003.
Andrew, D. (2010). Quantitative characterization of low‐threshold mechanoreceptor inputs to lamina
I spinoparabrachial neurons in the rat. Journal of Physiology, 588(1), 117–​24.
Baliki, M. N. and Apkarian, A. V. (2015). Nociception, pain, negative moods, and behavior selection.
Neuron, 87(3), 474–​91.
Barrett, L. F. and Simmons, W. K. (2015). Interoceptive predictions in the brain. Nature Reviews.
Neuroscience, 16(7), 419.
Beckes, L. and Coan, J. A. (2011). Social baseline theory: The role of social proximity in emotion and