Sensory Integration and the Unity of Consciousness 1st Edition David Bennett

Sensory Integration and the Unity of
Consciousness 1st Edition David Bennett download
https://ebookgate.com/product/sensory-integration-and-the-unity-
of-consciousness-1st-edition-david-bennett/
Get Instant Ebook Downloads – Browse at https://ebookgate.com

Get Your Digital Files Instantly: PDF, ePub, MOBI and More
Quick Digital Downloads: PDF, ePub, MOBI and Other Formats
The unity of consciousness binding integration and
dissociation Axel Cleeremans
https://ebookgate.com/product/the-unity-of-consciousness-binding-
integration-and-dissociation-axel-cleeremans/
Sensory Integration Theory and Practice 3rd Edition
Shelly Lane
https://ebookgate.com/product/sensory-integration-theory-and-
practice-3rd-edition-shelly-lane/
The Character of Consciousness 1st Edition David J.
Chalmers
https://ebookgate.com/product/the-character-of-consciousness-1st-
edition-david-j-chalmers/
Bennett s Cardiac Arrhythmias Practical Notes on
Interpretation and Treatment Eighth Edition David H.
Bennett(Auth.)
https://ebookgate.com/product/bennett-s-cardiac-arrhythmias-
practical-notes-on-interpretation-and-treatment-eighth-edition-
david-h-bennettauth/

Sentence Comprehension The Integration of Habits and
Rules Language Speech and Communication David J.
Townsend
https://ebookgate.com/product/sentence-comprehension-the-
integration-of-habits-and-rules-language-speech-and-
communication-david-j-townsend/
Ethics Emotion and the Unity of the Self Oliver Letwin
https://ebookgate.com/product/ethics-emotion-and-the-unity-of-
the-self-oliver-letwin/
An Evidence Based Guide to Combining Interventions with
Sensory Integration in Pediatric Practice 1st Edition
Erna Imperator Blanche
https://ebookgate.com/product/an-evidence-based-guide-to-
combining-interventions-with-sensory-integration-in-pediatric-
practice-1st-edition-erna-imperator-blanche/
Using Intensive Interaction and Sensory Integration A
Handbook for Those Who Support People With Severe
Autistic Spectrum Disorder 1st Edition Phoebe Caldwell
https://ebookgate.com/product/using-intensive-interaction-and-
sensory-integration-a-handbook-for-those-who-support-people-with-
severe-autistic-spectrum-disorder-1st-edition-phoebe-caldwell/
Biology The Unity and Diversity of Life 12th Edition
Cecie Starr
https://ebookgate.com/product/biology-the-unity-and-diversity-of-
life-12th-edition-cecie-starr/

Sensory Integration and the Unity of Consciousness

Sensory Integration and the Unity of Consciousness
edited by David J. Bennett and Christopher S. Hill
The MIT Press
Cambridge, Massachusetts
London, England

© 2014 Massachusetts Institute of Technology
All rights reserved. No part of this book may be reproduced in any form by any
electronic or mechanical means (including photocopying, recording, or information
storage and retrieval) without permission in writing from the publisher.
MIT Press books may be purchased at special quantity discounts for business or sales
promotional use. For information, please email special_sales@mitpress.mit.edu.
This book was set in ITC Stone Serif Std by Toppan Best-set Premedia Limited, Hong
Kong. Printed and bound in the United States of America.
Library of Congress Cataloging-in-Publication Data
Sensory integration and the unity of consciousness / edited by David J. Bennett
and Christopher S. Hill.
pages cm
Includes bibliographical references and index.
ISBN 978-0-262-02778-6 (hardcover : alk. paper)
1. Consciousness. 2. Sensorimotor integration. 3. Senses and sensation. I. Bennett,
David J., 1957–, editor. II. Hill, Christopher S., editor.
B808.9.S46 2014
128—dc23
2014003872
10 9 8 7 6 5 4 3 2 1

For DJB’s mom and dad, Elizabeth and Jacob

Contents
Preface ix
I Mainly on Sensory Integration 1
1 Bayesian Modeling of Perceiving: A Guide to Basic Principles 3
David J. Bennett, Julia Trommershäuser, and Loes C. J. van Dam
2 The Multisensory Nature of Perceptual Consciousness 15
Tim Bayne
3 The Long-Term Potentiation Model for Grapheme-Color Binding in
Synesthesia 37
Berit Brogaard, Kristian Marlow, and Kevin Rice
4 Intermodal Binding Awareness 73
Casey O’Callaghan
5 The Unity Assumption and the Many Unities of Consciousness 105
Ophelia Deroy
6 Multimodal Unity and Multimodal Binding 125
Frédérique de Vignemont
7 Can Blue Mean Four? 151
Jennifer Matey
8 Establishing Cross-Modal Mappings: Empirical and Computational
Investigations 171
Pawan Sinha, Jonas Wulff, and Richard Held
9 Berkeley, Reid, and Sinha on Molyneux’s Question 193
James Van Cleve
10 Modeling Multisensory Integration 209
Loes C. J. van Dam, Cesare V. Parise, and Marc O. Ernst

viii Contents
II Primarily on the Unity of Consciousness 231
11 A Unity Pluralist Account of the Unity of Experience 233
David J. Bennett and Christopher S. Hill
12 Unity, Synchrony, and Subjects 255
Barry Dainton
13 Experiences and Their Parts 287
Geoffrey Lee
14 Unity of Consciousness: Advertisement for a Leibnizian View 323
Farid Masrour
15 Partial Unity of Consciousness: A Preliminary Defense 347
Elizabeth Schechter
16 E pluribus unum: Rethinking the Unity of Consciousness 375
Robert Van Gulick
17 Counting Minds and Mental States 393
Jonathan Vogel
Contributors 401
Index 403

Preface
The papers in the present volume examine two closely related aspects of
mind and mental functioning—the relationships among the various senses,
and the bonds that tie conscious experiences together to form unified
wholes. A flood of recent work in both philosophy and perception science
has challenged widespread conceptions of the traditional sensory systems—
sight, touch, and so on—as operating in isolation. The contributors to Part
I discuss the various ways and senses in which perceptual contact with the
world is or may be “multisensory.” Recent years have also seen a surge of
interest in unity of consciousness. The papers in Part II explore a range
of questions about this topic, including the nature of unity, the degree to
which conscious experiences are unified, and the relationship between uni-
fied consciousness and the self. Like the papers in Part I, those in Part II seek
to integrate scientific and philosophical concerns.
Work on the present volume began with a conference that occurred at
Brown University in November 2011. David Bennett and Susanna Siegel
were the organizers of the conference, and the funding came from the
Brown Department of Philosophy and the Network for Sensory Research.
The editors wish to thank Bernard Reginster, the Chair of Philosophy at
Brown, for help in securing support for the conference, and also Mohan
Matthen, Principal Investigator of the Network project. We also gratefully
acknowledge help of various kinds from Kevin Connolly, Philip Laughlin,
Aimee McDermott, and Joanna Rolfes.
There are some grouping and dependency relationships among the
papers that should be noted. This is especially true of the papers in Part I,
but there is also an important dependency relationship between two papers
in Part II. (1) The de Vignemont, Deroy, O’Callaghan, and Bayne papers
all concern various aspects of perceptual binding, addressing philosophi-
cal concerns and issues in deeply empirically informed ways. One ques-
tion addressed is how perceptual systems link disparate sources of sensory

x Preface
information to the same object or event. (2) Sinha and Van Cleve dis-
cuss “Molyneux’s Problem.” Sinha presents and discusses relevant recent
empirical results from his Project Prakash. Van Cleve places these recent
results in a philosophical-historical setting. (3) In their paper on synesthe-
sia, Brogaard, Marlow, and Rice survey recent empirical results, along with
different empirical-scientific explanatory accounts on offer—including
their own. Matey addresses the kind of content carried by certain kinds of
synesthetic experiences—connecting her analysis to contemporary debates
about the possible “high-level” content of everyday experiences. (4) Ben-
nett, Trommershäuser, and van Dam, in their contribution, guide readers
through basic principles of recent computational modeling of cue combi-
nation. In their chapter, van Dam, Parise, and Ernst discuss recent com-
putational modeling of cue combination—modeling that is tied closely to
recent empirical results, also described. (5) Finally, in Part II, while Vogel’s
contribution is to some extent a freestanding paper, it is largely a response
to Schechter’s contribution.
The volume represents substantial commitments of time and energy by
both editors, but CSH wishes to acknowledge that DJB is solely responsible
for the original idea for the project and also that DJB has made a larger con-
tribution to the editorial process.
We conclude by noting a work that is not cited in any of the following
papers nor elsewhere in the contemporary literature on unity of conscious-
ness, but that we believe to merit attention. Ivan Fox’s Ph.D. dissertation
The Unity of Consciousness and Other Minds (Harvard, 1984) argues forcefully
for the view that whole states of consciousness are metaphysically more
fundamental than the individual experiences that are thought to constitute
them.

I Mainly on Sensory Integration

1 Bayesian Modeling of Perceiving: A Guide to Basic
Principles
David J. Bennett, Julia Trommershäuser, and Loes C. J. van Dam
1 The Single Cue Case. Example: Perceiving Slant from Texture
We begin with a simplified example concerning the perception of sur-
face slant from sensitivity to texture information (cf. Knill, 2003, 2007).
The basic, perception-science version of Bayes will fall out naturally and
intuitively.
In our toy model, suppose that it is assumed that surface texture elements
are circular (see figure 1.1).1
Suppose a perceiver views a surface head on,
looking straight at a circular texture element. We’ll say that at “upright,” the
slant of the surface is 90 degrees. That circular element will project to a circle
on a flat projection plane, which approximates a sensory surface.
The height-to-width ratio of a projected circular texture element is called
the “aspect ratio” of the projected image. So, the aspect ratio of the image
when looking straight on at the (90-degree, upright) surface is 1 because the
image is itself circular.
Now imagine the surface to be slanted back in depth. The projections
of the same circular texture element will now be progressively “thinner”
ellipses with progressively smaller aspect ratios (all fractions of 1, but ever
smaller). There is a simple relation between such image aspect ratios and
the slant of the surface:
A = sin (S). (1)
(A stands for the aspect ratio. S is the slant of the surface, measured from the
“straight on” upright, where S = 90. If the surface is viewed straight on, the
projection of a circular texture element will be a circle and the aspect ratio
will be 1. As the surface is slanted back in depth away from upright—to 70
degrees, to 50 degrees, and so on—the projections will be successively thin-
ner ellipses, with smaller aspect ratios.)

4 D. J. Bennett, J. Trommershäuser, and L. C. J. van Dam
It follows that if the aspect ratio of a projection from the circular element
were measured accurately, surface slant could be immediately and directly
recovered under the assumption that surface texture elements are circular.
Thus, a measured aspect ratio of 0.707 would correspond to a surface slant
of 45 degrees.
Suppose reasonably, though, that there is noise in the measuring process
in recording aspect ratios. As a result, on one occasion, a surface slanted
at 45 degrees might give rise to an aspect ratio “measurement reading”
of 0.643—“best corresponding” to a slant of 40 degrees by application of
equation 1. On another occasion, that same surface (at 45 degrees) might
lead to a measured aspect ratio of 0.719—“best corresponding” to a surface
slant of 46 degrees by equation 1. And so on. If percipients operate by the
measured aspect (possibly in error due to noise), they’ll often be a bit off
under repeated viewing of the same 45-degree slanted surface. The mean or
center of the “aspect ratio” readings might correspond (by equation 1) to
a 45-degree slant, but there will be a spread or distribution. The amount of
“bouncing” or variability will depend on how noisy the measuring process
is. The greater the spread, the less precise or reliable the measurement is said
to be. (Note: accuracy is a different matter, determined by whether the mean
of the distribution aligns with the actual quantity being measured.)
Suppose, though, that one also had a decent prior guess or hypothesis
that surfaces tended to be slanted at 45 degrees in the surrounding organ-
ism environment.2
A basic Bayesian idea is that drawing on such a prior/
working assessment of the distribution of slants in the world can help
leaven the possibly detrimental effects of measurement noise; such a prior
slant (°)
slant = 90°
h/w = 1
slant = 45°
h/w = 0.707
slanted surface
texture element aspect ratio
lateral view
frontal view
Slant from texture Bayesian cue integration
prior p(s)
likelihood p(d|s)
posterior p(s|d)
Slant (°)
perceived slant
expected slant
measured slant
Figure 1.1

Bayesian Modeling of Perceiving 5
hypothesis might also counter error in the operative working model of
how the stimulus is generated or measured. See figure 1.1 for specific illus-
tration. Intuitively, assuming such a prior distribution of slants will tug
upshot estimates toward slant values in line with the prior probability
distribution.
With that as background, here is the perception-science version of Bayes
applied to the slant perception case:
P(s/d) = (P(d/s) * P(s)) / P(d) (2)
Think of d as the measured value of an aspect ratio. Therefore, on a given
perceptual encounter there will be a specific value of d—a measured aspect
ratio—plugged into equation 2 in inferring a perceptual estimate of what
slant is present.
The P(s) is the prior probability distribution, with the basic meaning just
discussed. Here the s ranges over slant values that are assumed to be distrib-
uted in the world in a certain way (see figure 1.1b). In our example, surface
slants are assumed to be distributed around a mean of 45 degrees.
The P(d/s) is the likelihood function (see figure 1.1b). This reflects a work-
ing model of: (i) how the world projects to a viewpoint or sensory surface,
and (ii) how those projections are measured—especially the noise/variabil-
ity in the measuring process. In our case, the likelihoods reflect the geom-
etry of projection of a circular texture element given by equation 1; for a
given measured aspect ratio, d, the mean of the likelihood function will
correspond to the slant given by plugging that measured value into equa-
tion 1. In our case, the likelihoods will also reflect the model of noise or
variability in the measuring process. This corresponds to the spread or vari-
ance of the likelihood function (i.e., how peaked it is or how spread out).
The P(s/d) is the posterior probability distribution (figure 1.1b). Intuitively:3
this gives the probability of surface slants given the measured value d of
(here) an aspect ratio. The idea is that this is what the organism “works
from.”
The most likely slant, given the aspect ratio, is the point where the poste-
rior distribution is maximal. Estimating the slant by choosing this point is
called maximum a posteriori estimation (MAP). If the distributions are Gauss-
ian, this will correspond to choosing the mean of the posterior. If the prior
distribution is flat (and therefore uninformative), the estimate of the slant
will be based only on the likelihood. This is called maximum likelihood esti-
mation (MLE).
To summarize, the upshot perceptual estimate of (here) slant will result
from interpreting the incoming stimulus in terms of two kinds of prior

assumptions about the world. Such “prior assumptions” are reflected in the
priors, and also in the likelihood functions. The latter (likelihood) assump-
tions may be embodied in a fairly complex model of stimulus/measure-
ment generation, often called a “generative model.”
2 Perceptual Estimation: Multiple Sources of Information
So, we’ll assume that one source of slant information is the visual slant
from texture source, as described above. Suppose on some occasions such
a visual-texture-based path yields an estimate of slant, vs. We will assume
that the other perceptual route at work derives slant from haptic informa-
tion, perhaps by gauging wrist flexion or angle. Suppose that, on this occa-
sion, this haptic route leads to an estimate of slant, hs. One way to arrive
at an upshot estimate of slant based on these two estimates from different
sources would be to take a weighted average:
Slant = w1 *vs + w2*hs. (3)
Here it is assumed that w1 + w2 = 1 (i.e., the weights are fractions that add
up to 1).4
The obvious question then is how to choose the weights.
The basic idea of linear cue integration is that sources are weighted pro-
portionally less if more noisy/variable (that is, less precise or less reliable).
Under some reasonable assumptions, the resulting upshot estimates of
(here) slant will be as precise—the least variable—as possible (see Landy,
Banks, & Knill, 2011, 7). That is, if the weights are chosen in this way to
reflect variability/precision, averaging will minimize the variability in the
upshot slant estimates over other choices of weights. In this circumscribed
sense, the slant estimates arrived at in the way described are guaranteed to
be “optimal.”5
Empirically it has often (though not invariably) been found that percep-
tual estimates that combine information from different sources are indeed
optimal in the sense described (cf. Ernst & Banks, 2002; see Landy, Banks,
& Knill, 2011 for an overview). Adjustments of weights to align with source
noisiness/variability can indeed be rapid, sometimes within a few trials (cf.
Seydell, Knill, & Trommershäuser 2010, 2011).
It turns out that this sort of weighted averaging is a special case of a
Bayesian model of cue integration that generalizes equation (2) above to
the multicue case (Landy, Banks, & Knill, 2011, 8–10).6
The case of estima-
tion by (3) would correspond on the Bayesian formulation to a form of
MLE (see section 1), only here there are two likelihood functions, one for

each source of information. If a nonflat prior distribution is used, then—
on the Bayesian formulation—the idea is that perceivers arrive at an esti-
mate via a MAP estimate by taking the maximum of the upshot posterior
distribution.
3 Perceptual Integration and Fusion
A challenge faced by perceptual systems is to determine when and how to
combine information. Here we will describe a few of the puzzles involved
in meeting such challenges. The first two subsections, (A) and (B), concern
“decisions” confronting perceptual systems about whether to combine or
integrate information in estimation. Subsection (C) concerns whether or
when perceptual systems retain separate estimates from different sources
(for example, visual and haptic) after integrating those estimates in estimat-
ing a worldly property.
We do not say much here about the details of the computational model-
ing designed to explain perceptual system behavior in these cases, which
can be complex—especially for cases (B) and (C). Our aim is mainly to iso-
late different perceptual system challenges. For a more complete summary
of the modeling details, see van Dam, Parise, and Ernst (this volume), who
also provide numerous references to relevant recent work on these topics in
the modeling literature.
(A) In the example of slant perception developed in the preceding
sections, the assumption made was that texture elements are circular.
As a result, surface slant can be determined by equation (1) above. But
suppose the additional use of stereo to gauge slant yields a widely discrep-
ant estimate of slant consistent with the unusual (but possible) situation
where the texture elements are ellipses and not circles. As a result, the
most sensible perceptual system strategy may well be to veto the slant-
from-texture assessment and instead rely only on the stereo-derived esti-
mate of slant. However, the cue combination schemes described thus far
would lead to an upshot estimate between these two estimates, thereby
skewing the upshot estimate in the direction of the divergent texture-based
estimate.
In a scheme explored by Knill (2003, 2007), veto behavior is modeled
using a mixture likelihood. Such a mixture likelihood is derived from a
likelihood reflecting the assumption that texture elements are circular and
from a likelihood reflecting the assumption that the texture elements are
elliptical. Such a mixture likelihood has long, flat, nonnegligible “tails”
that essentially reflect the possibility that the texture elements are ellipses.

This leads to veto-like behavior when the estimates are widely discrepant
(Knill, 2003, 2007).7
Note that in this case it is granted as somehow known or assumed by the
perceptual system that the separate (stereo and texture) estimates pertain to
the same (here) slant property. The source of the discrepancy of estimates
results from the falsity of the assumption that the texture elements are cir-
cular, which throws off the texture-based estimate of slant. The long-tailed
mixture likelihood, in effect, allows the estimate that is discrepant (due to
the failed assumption) to be vetoed or discarded.
(B) Section (A) presents an example in which additional information
from a second cue serves to rule out as discrepant the estimate resulting
from a false assumption about the world. However, more typically a dis-
crepancy between sensory information cannot so easily be blamed on
one information source or the other. Cue combination modeling should
also explain how perceptual systems make a reasonable determination of
whether sensory signals from different sources—say, haptic and stereo-
vision—pertain to the same worldly property (slant, size, etc.).
Consider, for instance, observers who are determining object size or
width by both gripping the object and determining size visually via ste-
reo. Due to measurement noise, there will inevitably be some discrepancy
between the haptic size signal or estimate and the visual-stereo size signal
even when the estimates are derived from the same width or expanse. The
perceptual-system challenge is to decide whether the estimates are indeed
of the same size or expanse and so should be integrated or combined in
arriving at an estimate of size. This correspondence problem (or problem of
causal inference) is an important challenge facing perceptual systems that
has only recently begun to draw the sustained attention of modelers.
The modeling details in, for instance, the work by Ernst and collabora-
tors are somewhat complex (see Ernst & Di Luca, 2011; Ernst, 2012; for a
summary of the Ernst work, see van Dam, Parise, & Ernst, this volume;
for a survey of related work, see Shams & Beierholm, 2010). However, the
basic idea is that the perceptual system determines the extent to which
any discrepancy in (here) haptic and visual estimates of size is likely due
to (i) noise (ii) a bias in the measuring of the worldly properties or (iii) a
difference in the worldly properties measured. To do this, prior knowledge
is needed specifying how likely it is that the combination of the relevant
kinds of sensory estimates—haptic and stereo, for example—co-occur or
yield estimates that align with each other. This is captured in what is called
the coupling prior encoding the assumption of how tightly the two sig-
nals are generally linked. An extremely tight linkage would mean that the

system assumes that these sensory sources are always providing informa-
tion about the same world property; a flat coupling prior means that their
co-occurrence is governed by chance. Using this prior, the separate sensory
estimates are each adjusted according to the estimate obtained from the
other source and the strength of the assumed association. If after this step
a discrepancy between the updated estimates is still sensed, this means it is
likely that there is either a measurement bias in one or both of the senses or
a real difference in the world properties measured with each. In that case,
if it is determined as very likely that the visual and haptic estimates derive
from different worldly size properties, then the estimates would not be fur-
ther integrated or combined.
(C) Suppose once more that subjects are determining size or width by
both looking at an object and gripping it (cf. Ernst & Di Luca, 2011, figure
12.3). Ernst and Banks (2002) found that haptic and visual size information
is combined in reaching an upshot estimate of size in a way that maxi-
mizes precision. But it seems quite possible that even though perceivers
combine the visual and haptic sensory information in this way, they still
retain access to the visual and haptic signals or estimates. If so, there would
not be complete perceptual fusion in the sense that the initial visual and
haptic estimates can still be accessed and used. This is just what is found for
touch and sight via the oddity task used in Hillis et al. (2002) and described
in van Dam, Parise, and Ernst (this volume). By contrast, Hillis et al. (2002)
found that for purely visual assessment of slant as gauged by stereo and
texture information, subjects could not draw upon the individual (visual)
stereo and texture signals or estimates, or at least not to the same extent as
when each was presented in isolation. Thus, at least for this sort of purely
visual slant perception, there was perceptual fusion, more or less automati-
cally leading to optimal integration or combination of the estimates.
In modeling when and why fusion occurs, Ernst and collaborators make
use of the coupling prior distributions noted in (B) (see van Dam, Parise,
and Ernst, this volume, and Ernst & De Luca 2011). Such a coupling prior
reflects assumptions about how likely it is that the different sorts of sen-
sory signals occur together. These might be haptic and stereo-vision signals
or the signals might both be visual, like texture and stereo. In effect, the
strength of the association between, say, haptic and the stereo-vision sen-
sory signals is encoded in the width or spread of this coupling prior distribu-
tion. Therefore, a tight or sharp-ridged coupling prior indicates that haptic
size signals and stereo (vision) size signals tend to vary together. With this
sort of prior information on hand, it is sensible for perceptual systems to
treat any discrepancy between haptic and stereo signals or estimates as due

to measurement noise and to be guided instead by the (strongly peaked)
coupling prior—discarding the individual haptic and stereo estimates.
4 Brief Pointers to the Philosophy of Perception
Though the Bayesian approach to modeling cue combination is the domi-
nant approach in perception science, the computational modeling details
of this work are just starting to find their way into discussions of perception
in philosophy (see Rescorla, forthcoming; see also Colombo & Seriès, 2012).
In this section we briefly point to places where connections are there to be
drawn and explored, with the science potentially stimulating, enriching,
and constraining the work in philosophy.
As we have seen, the chief aim in this Bayesian cue combination model-
ing tradition is to explain how different sources of information are com-
bined in reaching upshot estimates of single worldly properties, such as size
or slant. This general kind of perceptual combination is what de Vignemont
(this volume) discusses and explores as “integrative binding.”
First there are apparent connections to Molyneux’s question, which is
explored in the Sinha et al. and Van Cleve contributions to this volume. As
Van Cleve discusses, formulations of Molyneux’s question can differ subtly
and importantly. But Molyneux’s question is essentially whether a person
born blind would, upon regaining sight, recognize by vision the shape of
an object previously known by touch.
On the working outlook of cue combination modeling, the same spatial
properties are accessed and estimated through different sources. This is in
line with the idea, dating to Aristotle, that there are perceptual “common
sensibles”—here spatial common sensibles. Such common sensible views
have traditionally been associated with a proposed or hypothesized “yes”
to Molyneux’s question. This observed, the connection between the cue
combination computational modeling and commitments on Molyneux is
not straightforward. A full charting would include exploring just how sub-
jects gain the ability to access the same spatial properties across informa-
tion gained from touch and sight—perhaps through some form of learning.
There is still much to be learned about how the relevant kind(s) of percep-
tual learning is/are achieved (on the modeling of learning, see van Dam,
Parise, & Ernst, this volume, and Ernst, 2012).
Here is another kind of connection between the modeling we have
described and topics in the philosophy of perception.
A number of contributors to this volume discuss how and when differ-
ent properties—for example, slant and color—are bound to the same object
or object surface. See especially the chapters by O’Callaghan, Deroy, and

Bayne. De Vignemont (this volume) refers to this sort of perceptual-binding
achievement as “additive binding.”
Explaining this kind of binding of multiple properties to an object/event
is not the primary aim of the kind of Bayesian cue combination modeling
described in this chapter. By way of (oversimplified) illustration: in the case
of Bayesian cue combination, the modeling task is to understand how to
combine information that is at least partly redundant—haptic and visual
information about slant, for example. By contrast, in the multiple property
cases that O’Callaghan, Deroy, and Bayne focus on, the sensory informa-
tion specifying one property (such as slant) and the sensory information
specifying the other property (such as color) are not redundant; two differ-
ent properties are indicated (to be bound to the same object).
Nonetheless, there is an important connection between the Bayesian cue
combination case (redundant information) and the multiple property case
(nonredundant information). A core challenge faced by perceptual systems
that must be resolved if different properties are to be accurately bound to
the same object or surface is determining whether the information guiding
the detection of each property derives from the same object. Once again,
see the contributions to this volume by Deroy, O’Callaghan, Bayne, and
de Vignemont. Meeting this challenge requires solving a correspondence
problem (or a problem of causal inference) of the sort described in part (B)
of section 3 and in the van Dam, Parise, and Ernst (this volume) section on
the correspondence problem.
Acknowledgments
Thanks to Cesare Parise for helpful discussion, as well as for figure 1.1.
Notes
1. Figure due to Cesare Parise.
2. A perennial challenge is to account for the origin of “prior knowledge” or
“assumptions” engaged in perceptual response (see Ernst & De Luca, 2012, for
extended discussion). One possibility is that this assumed distribution has been pro-
grammed into the organism over evolutionary time in encounters over eons with
typical earthly natural environments. Another is that the prior distributions and/or
the likelihoods (reflecting world models) are learned. The question of whether the
prior information is innate or learned—and if learned, under what conditions—can
often be approached through experiment (van Dam, Parise, & Ernst, this volume;
Ernst, 2012). The exact mechanisms of learning the relevant distributions are not
well understood (as observed in both van Dam, Parise, & Ernst, this volume, and
Ernst, 2012).

3. Strictly, for particular (real-valued) slants, the probability will be zero. So, if we
are careful, we’d talk instead of areas under the “probability density function” cor-
responding to the probability that a slant will fall within a certain range.
4. Prior information or assumptions about the distribution of slants can be incorpo-
rated into the upshot estimate by adding a further estimate, pr (for “prior”), with its
own weight:
Slant = w1*vs + w2*hs + w3*pr. (4)
Here again it is assumed that the weights add up to 1.
5. The upshot estimate is only guaranteed to be unbiased if the individual estimates
are unbiased. Assuming Gaussian noise, you can think of the assumption of lack of
bias as the assumption that the distribution of measurements from a source—of
slant, say, via texture—will center on the actual, worldly slant value present. Details
are debated, but this is likely often an unrealistic assumption. Too much perceiving
is biased or inaccurate. For sustained discussion of this “challenge of bias” within
the Bayesian modeling tradition, see Ernst and Di Luca (2011) and part (B) of section
3. The challenge was independently emphasized outside the Bayesian tradition by
Domini and Caudek (2011).
6. The basic idea is that the likelihood distributions—one for each source of infor-
mation—are multiplied together and with the prior. Therefore, with two sources of
information,
P(s|vs,hs) ∝ P(vs|s)*P(hs|s)*P(s). (5)
Then a MAP estimate of the worldly property such as slant is arrived at by taking
the maximum of the posterior P(s|vs,hs).
Equation (3) above is a special case of (5) in the following way. First, there is no
prior in (3), so it corresponds to the case where (5) has a flat prior. The posterior,
P(s|vs,hs), in (5) will in that case simply correspond to multiplying the two likeli-
hoods and thus to MLE estimation. Again, the maximum of the posterior distribution
is the most likely slant given the two inputs, vs and hs. This should lead to the same
perceived slant as equation (3) in the case where the weights in equation (3) are
optimally chosen to minimize variability.
7. Ernst (2012) contains a detailed discussion of the computational effects of posit-
ing such thick-tailed distributions and of the motivations and justifications for
doing so.
References
Colombo, M., & Seriès, P. (2012). Bayes on the brain—on Bayesian modeling in neu-
roscience. British Journal for the Philosophy of Science, 63, 697–723.

Domini, F., & Caudek, C. (2011). Combining image signals before three-
dimensional reconstruction: The intrinsic constraint model of cue integration. In
J. Trommershäuser, K. Kording, & M. S. Landy (Eds.), Sensory cue integration (pp. 120–
143). Oxford: Oxford University Press.
Ernst, M. O. (2012). Optimal multisensory integration: Assumptions and limits. In
B. E. Stein (Ed.), The new handbook of multisensory processes (pp. 1084–1124). Cam-
bridge, MA: MIT Press.
Ernst, M. O., & Banks, M. S. (2002). Humans integrate visual and haptic information
in a statistically optimal fashion. Nature, 415, 429–433.
Ernst, M. O., & Di Luca, M. (2011). Multisensory perception: From integration to
remapping. In J. Trommershäuser, K. Kording, & M. S. Landy (Eds.), Sensory cue inte-
gration (pp. 224–250). Oxford: Oxford University Press.
Hillis, J. M., Ernst, M. O., Banks, M. S., & Landy, M. S. (2002). Combining sensory
information: Mandatory fusion within, but not between, senses. Science, 298,
1627–1630.
Knill, D. C. (2003). Mixture models and the probabilistic structure of depth cues.
Vision Research, 43, 831–854.
Knill, D. C. (2007). Robust cue integration: A Bayesian model and evidence from
cue-conflict studies with stereoscopic and figure cues to slant. Journal of Vision, 7(5),
1–24.
Landy, M. S., Banks, M. S., & Knill, D. C. (2011). Ideal observer models of cue inte-
gration. In J. Trommershäuser, K. Kording, & M. S. Landy (Eds.), Sensory cue integra-
tion (pp. 5–29). Oxford: Oxford University Press.
Rescorla, M. (Forthcoming). Bayesian perceptual psychology. In M. Matthen (Ed.),
The Oxford handbook of the philosophy of perception. Oxford: Oxford University Press.
Seydell, A., Knill, D. C., & Trommershäuser, J. (2010). Adapting Bayesian priors for
the integration of visual depth cues. Journal of Vision, 10, 1–27.
Seydell, A., Knill, D. C., & Trommershäuser, J. (2011). Priors and learning in cue
integration. In J. Trommershäuser, K. Kording, & M. S. Landy (Eds.), Sensory cue inte-
gration (pp. 155–172). Oxford: Oxford University Press.
Shams, L., & Beierholm, U. R. (2010). Causal inference in perception. Trends in Cog-
nitive Sciences, 14, 425–432.

2 The Multisensory Nature of Perceptual Consciousness
Tim Bayne
1 Introduction
Philosophical reflection on perceptual consciousness has typically adopted
a modality-specific perspective as its point of departure. According to this
approach, an account of perceptual consciousness as a whole will simply
fall out of an account of each of the various perceptual modalities. In this
chapter, I argue against one manifestation of this atomistic approach to
perceptual experience: the decomposition thesis. According to the decom-
position thesis, a person’s overall perceptual experience can be identified
with the sum of his or her modality-specific experiences. If the decom-
position thesis is correct, then perceptual experience can be exhaustively
factored into modality-specific states. Modality-specific ways of referring
to perceptual experience would not merely pick out distinctive clusters of
phenomenal features and objects, they would also carve the structure of
consciousness at its joints, so to speak.
Michael Tye has suggested that the decomposition thesis—or at least
something very much like it—is the “received view” among philosophers
of mind (Tye, 2003, 17). I suspect that he is right. Whether or not that is
the case, there is no doubt that the decomposition thesis captures an influ-
ential and intuitively plausible conception of consciousness. It is, however,
a conception that faces significant objections—or so I will argue. I begin by
considering whether the decomposition thesis can accommodate the unity
of consciousness (sec. 2) and the perception of common sensibles (sec. 3),
before turning to the question of whether it can account for multisensory
integration (secs. 4–6).
Before addressing those tasks, let me first consider a more fundamental
issue raised by the decomposition thesis. One might think that it is not
possible to evaluate the decomposition thesis without having an account of
how many perceptual modalities there are, where the boundaries between

16 T. Bayne
them lie, and what makes it the case that an experience is assigned to a
particular sensory modality. Unfortunately, there are no answers to these
questions—at least, none that command widespread assent (Macpherson,
2011a, 2011b). Some theorists recognize only the five traditional Aristote-
lian senses of vision, audition, touch, taste, and smell, whereas others argue
that human perception involves a number of additional senses. Those who
argue for more than the traditional five senses, however, do not themselves
agree about the identity of the additional senses. Some theorists regard pas-
sive touch and active touch as separate modalities; others hold that the
detection of temperature, pressure, and vibration involve separate modali-
ties (or, at the very least, submodalities). There are also debates about the
individuation of the chemical senses, with open questions concerning
the relations between the perception of taste, smell, and flavor (Auvray &
Spence, 2008; Stevenson, 2009). In addition, there are unresolved questions
about whether perceptual experiences are to be assigned to specific modali-
ties on the basis of their intentional objects, their phenomenal character, or
their underlying neurofunctional structure.
Although these questions raise important issues that any thoroughgo-
ing treatment of the decomposition thesis must address, I will (largely) set
them to one side here. As we will see, significant progress can be made in
evaluating the decomposition thesis by relying only on our intuitive sense
of how to individuate experiences. In what follows, I will focus my atten-
tion on vision and audition, although many of the issues raised in the fol-
lowing discussion apply also to other modalities.
2 Phenomenal Unity
One source of pressure on the decomposition thesis derives from reflec-
tion on the unity of consciousness. Consider what it’s like to experience a
jazz trio, where one both sees the musicians and hears the music that they
produce. In such cases, one doesn’t merely enjoy visual and auditory expe-
riences at one and the same time, but instead these experiences are unified
with each other within the context of one’s overall perceptual awareness
of the world. One way to highlight this unity is by contrasting normal
perceptual experience with the kind of perceptual experience that might
be enjoyed by a creature with two streams of consciousness. In such a crea-
ture, the visual experience of the trio might be restricted to one stream of
consciousness, and the auditory experience might be restricted to the other
stream of consciousness. Such an individual will have visual and auditory
experiences at one and the same time, but these experiences will not enjoy

The Multisensory Nature of Perceptual Consciousness 17
a conjoint phenomenal character in the way that those of a normal percep-
tual subject will.
These considerations suggest that we need to recognize a phenomenal
unity relation in order to characterize normal perceptual experience (Bayne
& Chalmers, 2003; Bayne, 2010; Dainton, 2000/2006; Tye, 2003). Might
recognizing such a relation put pressure on the decomposition thesis? It
would if phenomenal unity has its own distinctive phenomenology—a
phenomenology over and above that which is possessed by the experiences
that it relates. The idea that phenomenal unity does possess its own phe-
nomenal character is not difficult to motivate. If phenomenal unity is a
relation that two experiences bear when they share a conjoint phenomenal
character, then it seems to follow that relations of phenomenal unity must
possess a proprietary phenomenal character. When phenomenal unity
holds between experiences drawn from different modalities—as it often
does—then that phenomenal character could not be identified with any
one experiential modality. For example, the character of the unity relation
that holds between an auditory experience (a) and a visual experience (v)
could be neither purely auditory nor purely visual; instead, it must be either
audio-visual or perhaps generically perceptual. If that is so, then the decom-
position thesis would be false: “subtracting” modality-specific experiences
from one’s overall perceptual experience would leave in place relations of
phenomenal unity.
Although it is not wholly unattractive, it seems to me that this line of
argument ought to be resisted. Despite its name, phenomenal unity does
not itself possess any phenomenal character. Although there is something
it is like to experience a and v together, this phenomenal togetherness
should not be assigned to the phenomenal unity relation itself. Phenom-
enal unity cannot be singled out in an act of introspective awareness, nor
can the “what it’s likeness” of one phenomenal unity relation (say, that
which holds between a visual and an auditory experience) be compared
and contrasted with the “what it’s likeness” of another phenomenal unity
relation (say, that which holds between an emotional experience and a tac-
tile experience).
A further reason to deny that phenomenal unity possesses its own phe-
nomenal character is that ascribing phenomenal character to phenomenal
unity would threaten to generate a vicious regress (Hurley, 1998). If rela-
tions of phenomenal unity did possess their own phenomenal character,
then they would need to be unified with the rest of the subject’s experi-
ences, and in order to account for that unity, we would need to posit an
additional set of phenomenal unity relations. In positing these relations,

18 T. Bayne
we would seem to have embarked on a regress that shows every sign of
being vicious.
The lesson to draw from these considerations is that although the pres-
ence of phenomenal unity does make a phenomenal difference, the relation
does not itself possess any phenomenology. The phenomenal difference
that phenomenal unity makes should not be assigned to the phenomenal
unity relation itself but to the complex experiential state that is generated
by its presence. Whenever two experiences are phenomenally unified with
each other, there is a complex experiential state that subsumes them both
(Bayne & Chalmers, 2003; Bayne, 2010). For example, if a visual experience
(v) and an auditory experience (a) are phenomenally unified, then there
will be a complex audio-visual experience that subsumes both v and a. And
here we do have an objection to the decomposition thesis, for the complex
state a-v is not itself a modality-specific state, nor can it be identified with
the mere conjunction of modality-specific states. Consider again an experi-
ence of a jazz trio in which one’s auditory and visual experiences are unified
in the form of a complex multimodal perceptual state. This state cannot
be identified with the conjunction of the visual and auditory states since a
subject with two streams of consciousness could enjoy both the visual state
and the auditory state without enjoying the complex state a-v. Insofar as the
phenomenology of a-v “outstrips” that of its modality-specific constituent
experiences, it would appear to be at odds with the decomposition thesis.
3 Common Sensibles
A second source of trouble for the decomposition thesis involves the exis-
tence of common sensibles—that is, features of the world that can be
experienced via more than one perceptual modality. Paradigm examples
of common sensibles include spatial, temporal, and causal relations. How
might the advocate of the decomposition thesis accommodate common
sensibles?
One commonly adopted strategy for meeting this challenge involves an
appeal to modality-specific modes of presentation (Kulvicki, 2007; Lopes,
2000; O’Dea, 2006). The idea here is that common sensibles are “com-
mon” (or amodal) at the level of reference but not at the level of sense. In
the same way that ‘the morning star’ and ‘the evening star’ pick out the
same object in different ways, so too might visual experience and audi-
tory experience (for example) represent spatial relations in different ways.
Visual experiences of space have a visual phenomenology whereas auditory
experiences of space have an auditory phenomenology.1
This conception

of perceptual content is sometimes referred to as a Fregean account and is
contrasted with a Russellian account of the common sensibles (and of per-
ceptual content more generally) according to which the phenomenal char-
acter of a perceptual experience is fixed by the properties that it represents
(Chalmers, 2004; Thompson, 2009). According to the Russellian, there are
no modality-specific modes of presentation for the simple reason that there
are no modes of presentation at all.
What reason might be given for the Fregean account of perceptual con-
tent? It is not uncommon for Fregeans to appeal to introspection in support
of the claim that there are modality-specific perceptual modes. However, it
is far from clear—to me, at any rate—that introspection can do the work
asked of it here. I don’t deny that there is typically an experiential (and
introspectively discernible) difference between experiences of space that
involve one modality and those that involve another modality, but it seems
to me that one can account for this fact without appealing to modality-
specific modes of presentation.
The first thing to note is that different modalities will typically represent
spatial properties and relations with different degrees of grain and precision.
For example, normal human vision exhibits greater spatial resolution than
either audition or touch do. Second, perceptual experience typically has a
self-representational aspect to it, for one is implicitly aware of which sen-
sory organs one is using to perceive the world. For example, one is usually
aware of whether one’s awareness of a surface is mediated by visual explora-
tion, tactile exploration, or both. Third, if the representation of space did
involve modality-specific modes of presentation, then such modes ought
to be apparent to one in those contexts in which one is aware of the same
property via multiple sensory modalities. For example, suppose that one
were both visually and tactually aware of the length of a table. In such a
case, the visual experience of the table’s length ought to be introspectively
discernible from the tactile experience of its length. But for what it’s worth,
I cannot identify any such multiplicity in my own experience.
A further problem with the modal-specificity view is that we are clearly
aware of spatial relations between stimuli that are presented in different
modalities. For example, one can be aware of the sound of a siren as being
to the left of a visually presented dog. What, according to the Fregean,
is the phenomenal character of one’s awareness of this spatial relation?
Clearly it could be neither purely visual nor purely auditory, but if that is
the case then positing modality-specific modes of representation wouldn’t
provide the advocate of the decomposition thesis with a fully satisfactory
account of the common sensibles.

20 T. Bayne
The picture becomes even bleaker for the advocate of the decomposition
thesis if we consider other common sensibles such as time. The suggestion
that there might be modality-specific modes of presentation for the repre-
sentation of time seems to lack even the prima facie plausibility that accom-
panies corresponding claims about the representation of space. This point
applies to temporal relations between events irrespective of whether they
are mediated by the same sensory modality or not. In short, a Russellian
account of the representation of temporal relations seems to be so much
more attractive than a Fregean account.2
What might the advocate of the decompositional thesis say at this
point? One line of thought involves appealing not to the phenomenal
character of an experience but to the causal history and/or neurofunc-
tional basis of the relevant representations. For example, one might argue
that a particular representation of space is visual not in virtue of its phe-
nomenal character, but in virtue of the fact that its causal history can be
traced back to processing in the retina and/or in virtue of the fact that
the representation itself—its vehicle, if you prefer—is located in the visual
cortex. This position motivates a version of the decomposition thesis
according to which at least some aspects of perceptual experience can be
assigned to particular modalities in virtue of their history and/or neuro-
functional basis.
Might this line of thought save the decomposition thesis? I think not.
But in order to see why, we must turn to the topic of multisensory integra-
tion. That is the business of the next section.
4 Multisensory Integration and the Unity Assumption
What explains the popularity of the decomposition thesis? Arguably much
of its appeal derives from assuming that perceptual experience involves
modality-specific systems. According to this view, not only is perceptual
processing in one perceptual modality doxastically impenetrable, it is also
encapsulated from processing that occurs in those neural systems associ-
ated with the subject’s other perceptual modalities.
This picture of perceptual processing has been extremely influential,
but its influence is not commensurate with its plausibility, for contempo-
rary cognitive neuroscience has replaced this view of perceptual processing
with one that is robustly multimodal. Research has revealed the existence
of significant numbers of bimodal and multimodal cells—that is, cells that
respond to stimulation in more than one sensory modality—and multisen-
sory association areas (Driver & Noesselt, 2008). For example, the superior

colliculus has been implicated in the integration of visual, auditory, and
tactile cues; the parieto-insular vestibular cortex has been implicated in the
integration of vestibular and visual cues; and the temporoparietal junction
has been implicated in the integration of tactile, proprioceptive, kines-
thetic, and visual cues. Moreover, many of the areas that have traditionally
been regarded as “unisensory” may be so only in terms of their afferent pro-
jections, for back-projections from multisensory convergence areas often
result in input from a secondary modality. Indeed, the ubiquity of multi-
sensory interaction in the brain has led some researchers to describe the
cortex as “essentially multisensory” (Foxe & Schroeder, 2005; Ghazanfar &
Schroeder, 2006).
But what really matters here is not the anatomical basis of multisen-
sory integration (MSI) but its effects on information processing. Research
extending back for over a century has revealed that perceptual process-
ing in one modality is often modulated by perceptual processing in other
modalities (Bertelson, 1988; Calvert, Stein, & Spence, 2004; de Gelder &
Bertelson, 2003). Let me begin with some examples of MSI before turning
to the implications of this research for the questions at hand.
One of the most well-known examples of MSI is the ventriloquism effect,
in which the apparent source of a spatially discrepant sound source is mislo-
calized so that it more closely matches the seen source, be it the articulated
lips of the ventriloquist’s dummy or the sight of a loudspeaker cone (Bertel-
son, 1999; Bertelson & de Gelder, 2004). There are also temporal forms of
“ventriloquism,” in which the presentation of a stream of auditory stimuli
modulates the perceived temporal location of visual stimuli (Bertelson &
Aschersleben, 2003; Morein-Zamir et al., 2003; see also Hanson et al., 2008).
Other examples of MSI involve the modulation of “categorical” perception.
For example, in the McGurk effect, dubbing the phoneme /ba/ onto the lip
movements for /ga/ produces a percept of the phoneme /da/ (McGurk &
MacDonald, 1976). Indeed, MSI can even affect the number of objects that
are presented in perceptual experience. In the sound-induced flash illusion,
participants are presented with multiple tones concurrently with a single
flash of light (Shams, Kamitani, & Shimojo, 2000, 2002; Shams, Ma, & Bei-
erholm, 2005). As a result of the interaction between vision and audition,
participants typically see (or “seem to see”) two flashes of light. A similar
interaction has been found between touch and audition: participants who
are presented with multiple tones will experience a single tap as two taps
(Hötting & Röder, 2004).
What bearing does multisensory integration have on the decomposition
thesis? Let us return to the suggestion that perceptual representations of

22 T. Bayne
common sensibles can be assigned to a particular modality in virtue of their
history and/or neurofunctional basis. We can now see that MSI problema-
tizes this proposal. In the ventriloquism effect, the representation of the
sound’s spatial location is not solely a function of the information received
via the ears or of processing in early auditory cortex but is the joint result of
both auditory and visual processing. Looking to the causal history of these
representations will not save the decomposition thesis.
What about appeals to the neurofunctional basis of perceptual repre-
sentations? Contrast two conceptions of MSI: a (purely) causal conception
and a constitutive conception. According to the (purely) causal conception,
the “multisensory” nature of MSI is restricted to processing the relevant
perceptual input: by contrast, the representations that result from multisen-
sory integration are wholly unisensory. For example, a causal treatment of
ventriloquism would take it to involve a purely visual representation of a
visual stimulus and a purely auditory representation of an auditory stimu-
lus. According to the constitutive conception, by contrast, the multisensory
nature of MSI is not restricted to perceptual processing but extends to the
neurofunctional nature of the perceptual representations that are generated
by that processing. In the context of ventriloquism, the representations in
question are not confined to either auditory cortex or to visual cortex but
instead are located in areas that are inherently audiovisual.
It might be useful to consider how these two accounts apply to the
sound-induced flash illusion. A causal treatment of the sound-induced flash
illusion takes it to result in a purely visual experience as of multiple flashes
(rather than the single flash that actually occurred) and a purely auditory
experience of multiple tones. A constitutive treatment of the sound-induced
flash illusion, by contrast, takes the multi-sensory interaction to result in a
representation that is audio-visual. Rather than generating a representation
of a visual object and a representation of an auditory object, the interaction
generates a representation of an audio-visual object—an entity that is both
flashing and tone-emitting.
If the causal conception of MSI were correct, then the line of thought
outlined earlier might suffice to save (a version of) the decomposition thesis.
However, it seems to me that a strong case can be made for treating many
multisensory interactions in constitutive terms.3
At the heart of this case is
an appeal to what Welch and Warren dub “the unity assumption” (Welch &
Warren, 1980). Here is Welch’s characterization of the assumption:
An intersensory conflict can be registered as such only if the two sensory modalities
are providing information about a sensory situation that the observer has strong rea-

sons to believe (not necessarily consciously) signifies a single (unitary) distal object
or event. (Welch, 1999, 373)
The idea is that in order to explain why the perceptual system integrates
information across modalities in the way that it does, we must take it to
identify the objects (and properties) of one modality with those of another.
(Thus, “the identity assumption” would be a more accurate label than “the
unity assumption.”) For example, we can understand the ventriloquism
effect only by assuming that the perceptual system identifies the heard
event with the seen event. If the perceptual system failed to identify one
event with the other, then it would not register the conflict between vision
and audition as a conflict, and we would have no information-processing
explanation of the effect. A similar story can be told with respect to the
McGurk effect. The perceptual system has reason to integrate its representa-
tions of the visual and auditory cues only if it takes these cues to concern
a single speech act. As Ernst (2006, 128) has remarked, “The integration of
signals is only reasonable if they are derived from the same object or event;
unrelated signals should be kept separate.”
Before examining how the unity assumption puts pressure on the
decomposition thesis, let me make a number of clarifying remarks about
the assumption itself. First, although Welch presents the assumption in
terms of what the subject believes (see above), it is more plausibly regarded
as subdoxastic in nature. MSI can be modulated by “top-down” effects (see
e.g., Vatakis & Spence, 2007), but it is not—in general, at least—doxastically
penetrable (Radeau, 1994). Knowing that one is currently subject to the
ventriloquism effect will not typically have any impact on the nature of
one’s perceptual experience.
Second, we can appeal to the unity assumption not only to explain the
resolution of intersensory conflict but also the enhancement of intersen-
sory coherence. Consider again the sound-induced flash illusion. It is unclear
how the unity assumption might account for this effect if it were concerned
only with intersensory conflict, for there is no inconsistency in represent-
ing a tone and representing the absence of any corresponding visual event
at (or near) the relevant time. However, there is an intuitive sense in which
hallucinating an additional flash of light increases the coherence of the
subject’s overall perceptual experience.4
Finally, the unity assumption does not itself constitute a complete
account of MSI. For one thing, it does not have anything to say about the
conditions under which an object that is perceived via one modality is
identified with an object that is perceived via another modality. Moreover,

24 T. Bayne
the unity assumption says nothing about how intersensory conflict ought
to be resolved in any particular case. Answers to these questions will be
provided only by broader explanatory frameworks such as that provided by
the Bayesian cue combination account (Ernst & Bülthoff, 2004; Trommer-
shäuser, Landy, & Körding, 2011). Such accounts do not render the unity
assumption redundant but provide a framework that explains when the
assumption is “triggered” and why the perceptual system responds to inter-
modal inconsistency in the way that it does.5
5 Multisensory Object Files
So much for multisensory integration and the unity assumption—what rel-
evance does the foregoing have for the decomposition thesis? The connec-
tion can be couched in terms of the notion of an object file, where “object
file” talk is simply a convenient way of referring to the fact that various
perceptual features have been tagged by the perceptual system as belong-
ing to the same intentional object. If it is implicit in the operation of the
perceptual system that an object perceived via one modality is identical to
an object perceived via another modality, then it seems to follow that the
object files created as the result of this processing must be multimodal: they
are not restricted to the features drawn from a single modality but extend
across modalities.
In fact, these object files are multisensory in two respects. First, they con-
tain “amodal” information about certain kinds of properties—the “com-
mon sensibles.” Consider the spatial content of an object file concerning a
ventriloquized event. Is this feature a visual feature or an auditory feature?
A causal conception of MSI would imply that the subject has two object
files: a visual file that represents its location in a visual manner and an
auditory file that represents its location in an auditory manner. However,
the fact that one’s awareness of the location of the ventriloquized event is a
function of both the visual and the auditory input indicates that the spatial
information in question is not modality-specific.
A second respect in which these files are multisensory is that they are
not restricted to features associated with one modality but include features
associated with multiple modalities. For example, in the sound-induced
flash illusion, subjects do not have one object file associated with the tone
and another associated with the flash but instead have a single object file
containing information associated with both. In effect, this object file rep-
resents that there is a flashing, beeping object at such-and-such location
in the subject’s perceptual field. Similarly, in the McGurk effect, one has

a single object file for the perceived speech act that contains information
derived from both audition and vision.
That, in a nutshell, is the unity assumption argument against the decom-
position thesis. The argument turns on three claims: (i) MSI requires the
unity assumption; (ii) the unity assumption requires a constitutive inter-
pretation, an idea that I have developed in terms of multisensory object
files; and (iii) if perception involves multisensory object files, then the
decomposition thesis is false. Let me flesh out this argument by consider-
ing some objections to it.
6 Objections and Replies
6.1 From Object Files to Perceptual Content
One objection to the argument is that it confuses levels of description:
multisensory integration and the notion of a multisensory object file are
subpersonal matters, whereas the decomposition thesis is a claim about per-
ceptual experience, a personal-level phenomenon. As such, the objection
runs, claims about multisensory integration and the structure of object files
couldn’t have any bearing on the prospects of the decomposition thesis, for
claims about personal-level phenomena are not hostage to the fortunes of
subpersonal, information-processing models.
This challenge is certainly worth taking seriously, for there are a number
of reasons why one should be reluctant to identify the contents of object
files with the objects of perceptual experience. First, experiments by Mitroff
and colleagues using a motion-induced blindness paradigm indicate that
object files can be updated independently of awareness (Mitroff & Scholl,
2005; Mitroff, Scholl, & Wynn, 2005; see also Bonneh, Cooperman, & Sagi,
2001). Second, although patients with object-centered neglect are often
consciously aware of only the right side of an object, evidence from prim-
ing indicates that they possess information about the left side of neglected
objects (Marshall & Halligan, 1988; see also Dorichi & Galati, 2000; Wojciu-
lik & Kanwisher, 1998). Presumably information about both the neglected
and consciously perceived sides of an object is bound together within a
single object file.
Third, and perhaps of most direct relevance to the current question,
there is evidence that at least certain kinds of multisensory integration can
occur “outside” of consciousness, in the sense that the subject need not
be conscious of the sensory cues responsible for the modulation of per-
ceptual content. Visually presented lip movements that are presented in
the ipsilesional hemifield of a patient suffering from hemineglect can give

26 T. Bayne
rise to a McGurk effect when combined with auditory speech stimuli
presented in the patient’s (neglected) contralesional hemifield (Soroker,
Calamaro, & Myslobodsky, 1995a, 1995b; see also Leo et al., 2008). Similar
results have been found for ventriloquism (Bertelson et al., 2000) and the
integration of emotion as detected via facial expression and tone of voice
(de Gelder & Vroomen, 2000). Assuming that neglect patients are unaware
of stimuli that they do not report, these results provide further evidence of
a gap between the contents of an object file and the structure of perceptual
content.
However, we can recognize that there is an important distinction
between the contents of object files and the nature of perceptual objects
without giving up on the claim that the nature of object files might con-
strain the nature of perceptual objects. One way to connect the two levels of
description is to consider the implications that MSI has for the contents of
the perceptual experiences that result from it. What “constraints” does the
perceptual experience generated by the sound-induced flash illusion place
on one’s environment? The answer recommended by the causal account
of MSI is that the veridicality of one’s experience requires only that one’s
environment contain sounds of a certain kind and flashes of a certain kind.
By contrast, the constitutive account of MSI would indicate that unless
one’s experience of lights and tones has a common causal ground, then
it is not wholly veridical. If one’s environment contained a flash of light
and a sound that failed to have a common ground—as indeed is the case in
the flash-induced sound illusion—then one’s experience would be to some
degree illusory.
Following O’Callaghan (2008b), it seems to me that the perceptual
experience that one has in the context of the sound-induced flash illusion
doesn’t leave it as an open question whether the flash and the beep are
manifestations of a single event, but instead imposes this requirement on
one’s environment. In other words, the perceptual content of one’s experi-
ence is in line with that suggested by the constitutive conception of MSI.
Consider a parallel between the issues just discussed and the interpreta-
tion of certain unisensory phenomena such as the phi phenomenon (Kol-
ers & Von Grunau, 1976) and the “cutaneous rabbit” (Geldard & Sherrick,
1972), a tactile illusion in which a rapid sequence of taps delivered first near
the wrist and then near the elbow creates the experience of an object “hop-
ping” from one’s wrist to one’s elbow. These illusions of apparent motion
are best accounted for by supposing that the perceptual system makes
assumptions about the numerical identity of the objects of perception. It is
only because the visual system assumes that the first object is numerically

identical to the second object that it constructs a (nonveridical) percept of
a single object moving from one location to another, and it is only because
the tactile system assumes that there is a single object “hopping” up one’s
arm that it revises the order in which the taps to the arm are experienced. In
these cases, it is part of the perceptual content of one’s experience that one
is presented with a single object in motion. Similarly, we should recognize
that claims about the numerical identity of perceptual objects are built into
the content of one’s perceptual experience in the sound-induced flash illu-
sion and many other examples of MSI.
It might be objected that this view is at odds with an influential con-
ception of perceptual content according to which perceptual content is
purely general rather than singular (McGinn, 1997; Davies, 1992; Horgan
& Tienson, 2002). Suppose one is looking at a book. On the general view
of perceptual content, the particular book at which one is looking does not
enter into the content of one’s experience, but one’s perceptual experience
would instead have been satisfied by any number of qualitatively identical
books. However, although the argument presented above might appear to
presuppose a singular account of perceptual content, any such appearance
would be misleading. All the argument requires is that if perceptual content
is purely general, then the features relating to more than one modality
must fall within the scope of the existential quantifier. In other words, the
content must have the form “There is an x such that x is both flashing and
beeping” rather than “There is an x such that x is flashing and there is a y
such that y is beeping.”
6.2 Multisensory Objects
A second challenge to the argument from multisensory integration con-
cerns the very notion of a multisensory object. The objects of sight are typi-
cally physical entities, whereas the objects of hearing are typically sounds.
We see the dog but hear its bark; we see the coffee pot but hear its gurgling;
we see the ambulance but hear its siren. And, our imagined critic might
continue, given that physical objects and sounds are different types of
things, it is unclear what an audiovisual object could possibly be. The idea
that visual and auditory features could be bound together in the form of
a single intentional object seems to be a deep mistake—indeed, a category
mistake, no less!—for there is nothing that could have both visual and audi-
tory features. A dog can be brown but not loud, whereas its bark can be loud
but not brown—nothing, it seems, could be both brown and loud.6
The challenge is a serious one, but there is a respectable reply to it.7
The response in question is best approached by considering a distinction

28 T. Bayne
that is often made in discussions of audition between primary (or “direct”)
perceptual objects and secondary (or “indirect”) perceptual objects (e.g.,
O’Callaghan, 2008a; Matthen, 2010). According to this proposal, although
the primary objects of auditory experiences are sounds, audition has sec-
ondary objects in the form of sound-emitting objects. We are aware of the
secondary objects of auditory experience by hearing the primary objects
that they produce. We hear the dog by hearing its bark; we hear the coffee
pot by hearing its gurgle; and we hear the ambulance by hearing its siren.
According to the account of multisensory objects that I am sketching
here, although the primary objects of perceptual experience are modality
specific, these objects will themselves be bound together in the form of a
composite “object.” Visual features will be bound to a visual object, and
auditory features will be bound to an auditory object, but auditory and
visual features will also be bound together insofar as the visual and auditory
objects are bound together. On this view, we need object files of two kinds:
one kind of file individuates the primary objects of perception and another
individuates its secondary objects. The role of the secondary object file is to
keep track of when two primary perceptual objects are associated with the
same secondary perceptual object.
But what exactly are secondary perceptual objects? They are, I suggest,
events. (It is perhaps no accident that Bertelson entitled his influential over-
view of multisensory integration “The perception of multimodal events.”)8
Consider the McGurk effect. Here, both visual and auditory features are
bound together as a representation of the utterance of the phoneme <da>.
In simple cases of ventriloquism, we might think of the intentional object
as something akin to an “explosion”—an event involving the production
of a sound and a spark of light. In other examples of ventriloquism—such
as when seeing a drum being struck “ventriloquizes” an accessory sound—
we might think of the intentional object as the event of the drum’s being
struck. And in the sound-induced flash illusion, the intentional objects
of one’s experience are events of flashing/beeping. In short, a plausible
account can be given of the intentional objects of multisensory integra-
tion: they are events.
6.3 Partial Integration
A rather different challenge to the argument from MSI involves the fact
that multisensory integration is often merely partial. For example, studies
of ventriloquism often find that although the presence of a visual cue biases
the perceived location of the auditory cue, and the presence of the auditory
cue may also bias the perceived location of the visual cue, the auditory and

visual cues may not be experienced as co-located; rather, they will simply be
experienced as closer to each than they would have been had each cue been
presented independently (see e.g., Bertelson & Radeau, 1981; Bertelson et
al., 2000).
Partial integration is prima facie puzzling. On the one hand, the unity
assumption suggests that the perceptual system must have bound the visual
and auditory cues together, for if it didn’t, then we would lack an informa-
tion-processing account of why their apparent locations were modulated
in the way that they were. But if the perceptual system did bind these cues
together, why did it not also co-locate them? After all, it seems unlikely
that a single event would produce a flash in one location and a sound in
another location.
We can put the partial integration challenge in terms of the following
dilemma: On the one hand we could deny that partial integration results in
multisensory object files. But if we do that, then we must either deny that
the unity assumption is required in order to explain partial integration,
or we must deny that the unity assumption is indicative of the presence
of multisensory object files. On the other hand, if we allow that partial
integration involves multisensory object files, then we must allow that an
object file can represent an object as having incompossible attributes—that
it can, for example, fully locate an event at two distinct locations. And that
too seems prima facie implausible. What should we do?
Embracing the first horn of this dilemma seems to me to be particularly
unpalatable. The link between multisensory integration (even when merely
partial) and the unity assumption seems to me to be highly robust, as does
the link between the unity assumption and the existence of multisensory
objects. So, that leaves the second horn of the dilemma—is there any way
of making this response to the objection palatable? Perhaps so.
One line of response would be to hold that object files can attribute
incompossible properties to an object. Consider the waterfall illusion, in
which a stationary object appears to be both moving and, at the same time,
stationary. Some theorists describe the content of the visual experience pro-
duced by the waterfall illusion as “logically impossible” (Frisby, 1979, 101;
see also Smeets & Brenner, 2008). This description—which has some phe-
nomenological plausibility—seems to require that the relevant object file
represents its object as both stationary and moving (relative to the same
frame of reference). And if we are willing to allow that unisensory object
files can attribute incompossible properties to their objects, perhaps we
should also allow that multisensory object files can attribute incompossible
properties to their objects.

30 T. Bayne
Another—and to my mind rather more attractive—response to the chal-
lenge posed by partial integration leans heavily on the idea that the inten-
tional objects of multisensory perception are not objects (strictly speaking)
but events. Events are not (usually) instantaneous, point-like occurrences,
but have both temporal and spatial dimensions. Because events have both
temporal and spatial duration, it is possible for an event-involving object
file to bind together visual and auditory objects even when those objects
are themselves assigned to distinct spatiotemporal locations. Roughly
speaking, we can think of perceptual content in cases of partial integration
on the model of the following: “There is a visual object at such-and-such a
spatio-temporal location, and there is an auditory object at such-and-such a
spatio-temporal location, and these two objects are implicated in the same
event.” Much more remains to be said on this topic, but I hope that this is
at least the start of a plausible account of partial integration.
7 Conclusion
In this chapter I have examined the decomposition thesis—the claim that per-
ceptual experience can be parceled out into modality-specific chunks. I have
argued that there are three respects in which the decomposition thesis is open
to challenge. First, it struggles to accommodate the unity of consciousness,
for—I have argued—we must recognize distinctively multimodal experiences
in order to do justice to the unity between experiences associated with differ-
ent modalities. Second, the decomposition thesis struggles to accommodate
the perceptual representation of common sensibles, such as spatial, temporal,
and causal relations. Third, the decomposition thesis struggles to accommo-
date the phenomenon of multisensory integration, which—I have argued—
requires conceiving of perceptual content in terms of multisensory perceptual
objects. Over and above the pressure that it puts on the decomposition thesis,
I hope that this chapter draws attention to some of the many questions of
philosophical interest that are raised by the multimodal nature of perception.
Multisensory integration “is the rule and not the exception in perception”
(Shimojo & Shams, 2001, 505), and there is a pressing need to explore its
implications for philosophical accounts of perception.9
Notes
1. Note that the Fregean conception of phenomenal content need not be unpacked
in modality-specific terms, for one could hold that phenomenal content is Fregean
rather than Russellian but deny that the modes of presentation associated with phe-

nomenal content track differences between sensory modalities. However, as a matter
of fact, many Fregeans do seem to think that there are modality-specific modes of
presentation.
2. As Nudds (2001) has argued, there are also cross-modal perceptual relations
involving causation. For example, in perceiving a drum being struck, one might be
aware of the sound as being generated by the contact of the drumstick with the
drum—an event that one is visually aware of.
3. The following discussion is heavily indebted to Casey O’Callaghan’s (2008b,
2012) important work on the multisensory nature of perception.
4. Another example of the perceptual system “creating” an additional perceptual
object in order to foster the coherence of the subject’s experience is provided by the
Aristotle illusion. When an object is placed between one’s fingers while they are
crossed, one will feel as though one is touching two objects.
5. The Bayesian account of multisensory integration can explain why one flash is
perceived as two when accompanied by multiple tones, whereas two flashes are not
perceived as one when they are accompanied by a single tone. Briefly, the account
appeals to the fact that audition has a finer temporal grain than vision does, and
hence the perceptual system puts more weight on those time-dependent numerosity
judgments that involve audition than those that involve vision.
6. The information-processing variant of this objection can be put in terms of the
“indispensable attributes” account of visual and auditory objects associated with the
work of Kubovy and colleagues (Kubovy & van Valkenburg, 2001; Kubovy & Schutz,
2010). The idea, briefly, is that visual objects are individuated in terms of their posi-
tion in a spatial frame of reference, whereas auditory objects are individuated in
terms of their position in a pitch-based frame of reference. In this framework, the
question posed by multisensory objects can be put as follows: with respect to what
frame of reference are audiovisual objects individuated? By appeal to both pitch and
space? In terms of some independent frame of reference? Or does the answer to this
question depend on the kind of audio-visual integration in question? I hope to
explore these questions in future work.
7. Note that this challenge applies with particular force in the context of interac-
tions between vision and audition. It is less clear how much force it has in the con-
text of other forms of multisensory interaction such as interactions between vision
and touch (for example).
8. Note too that in the literature on multisensory binding, theorists typically refer
to the intentional objects of such binding as “events” rather than “objects.” See
Zmigrod, Spapé, and Hommel (2009) and Zmigrod and Hommel (2011, 2013).
9. Earlier versions of this chapter were presented at the Thumos group (Geneva), the
Logos group (Barcelona), and at a conference on the unity of consciousness and

Discovering Diverse Content Through
Random Scribd Documents

in the modest home of a bourgeois family. There was a good deal of
criticism of this innovation, but Their Majesties would not interfere. They
were so unaffected in their own private life that they could not but encourage
the same tastes in their children.
Yet the war had already brought some very remarkable change in our life
at the palace. It had always been austere, and now became even more so. The
Czar was away a good deal. The Czarina and her two elder daughters almost
always wore the costume of a nurse, and divided their time between visits to
the hospitals and the innumerable duties arising out of their work for the
relief of the wounded. The Czarina was very tired even when the war began.
She had spent herself without counting the consequences, devoting herself
with the enthusiasm and ardour she brought to everything to which she set
her hand. Although her health was severely shaken, she displayed
remarkable physical elasticity. She seemed to derive comfort and strength
from the accomplishment of the splendid task which she had undertaken. It
was as if she found that it satisfied her craving for self-devotion and enabled
her to forget the poignant anxiety and apprehension that the Czarevitch’s
illness caused, even in its inactive periods.
Another result of the war, as agreeable as unexpected, was that Rasputin
had retired into the background. At the end of September he had returned
from Siberia completely recovered from the terrible wound which had all but
ended his days. But everything pointed to the fact that since his return he
was being more or less neglected. In any case, his visits were more and more
infrequent. It was true that as Alexis Nicolaïevitch had been so much better
during the winter there had been no need to resort to his intervention, so that
he had found himself deprived of what had been his great stand-by.
But when all is said, his power remained quite formidable. I had proof of
the fact a short time after, when Madame Wyroubova was all but killed in a
terrible railway accident. She was nearly dead when she was dragged from
under the fragments of a shattered carriage, and had been brought to
Tsarskoïe-Selo in a condition which seemed desperate. In her terror the
Czarina had rushed to the bedside of the woman who was almost her only
friend. Rasputin, who had been hastily sent for, was there also. In this
accident the Czarina saw a new proof of the evil fate which seemed to
pursue so relentlessly all those whom she loved. As she asked Rasputin in a
tone of anguish whether Madame Wyroubova would live, he replied:

“God will give her back to you if she is needed by you and the country. If
her influence is harmful, on the other hand, He will take her away. I cannot
claim to know His impenetrable designs.”
It must be admitted that this was a very clever way of evading an
awkward question. If Madame Wyroubova recovered he would have earned
her eternal gratitude, as, thanks to him, her recovery would seem to
consecrate, as it were, her mission with the Czarina. If she died, on the other
hand, Her Majesty would see in her death a manifestation of the inscrutable
ways of Providence, and thus be the more easily consoled in her loss.[36]
Rasputin’s intervention had helped him to recover his influence, but his
triumph was short-lived. In spite of everything, we felt that something had
changed, and that he was not so important as he had been. I was delighted to
note the fact, particularly as shortly before I had had a long talk on the
subject of the staretz with the Swiss Minister in Petrograd.[37] The
information he gave me in the course of our conversation left me in no doubt
as to the real character of Rasputin. As I had always suspected, he was a
misguided mystic who possessed a kind of psychic power, an unbalanced
creature who worked alternately through his carnal desires and his mystic
visions, a being quite capable of having weeks of religious ecstasy after
nights of infamy. But before this interview I had never realised the
importance that was attached to Rasputin’s influence on politics, not merely
in Russian circles, but even in the embassies and legations of Petrograd. That
influence was greatly exaggerated, but the mere fact that it could exist was a
kind of challenge to public opinion. The presence of this man at Court was
also a subject of mystery and abhorrence to all who knew the debauchery of
his private life. I fully realised that all this involved the greatest danger to the
prestige of Their Majesties and furnished a weapon which their enemies
would sooner or later try to use against them.
The mischief could only have been remedied by sending Rasputin away;
but where was the power strong enough to bring about his disgrace? I knew
the deep, underlying causes of his hold over the Czarina too well not to fear
the restoration of his influence if circumstances took a turn favourable to
him.
The first six months of the war had not brought the results hoped for, and
everything pointed to a long and bitter struggle. Unexpected complications
might arise, for the prolongation of the war might well bring very serious
economic difficulties which could foster general discontent and provoke

actual disorder. The Czar and Czarina were much concerned at this aspect of
the matter. It made them very anxious.
As ever in moments of trouble and uncertainty, it was from religion and
the affection of their children that they drew the comfort they needed. With
their usual natural simplicity and good humour the Grand-Duchesses had
accepted the increasing austerity of life at Court. It is true that their own
lives, so utterly destitute of the elements which young girls find most
agreeable, had prepared them for the change. When war broke out in 1914,
Olga Nicolaïevna was nineteen and Tatiana Nicolaïevna had just had her
seventeenth birthday. They had never been to a ball. The only parties at
which they had appeared were one or two given by their aunt, the Grand-
Duchess Olga Alexandrovna. After hostilities one thought, and one thought
alone, inspired them—to relieve the cares and anxieties of their parents by
surrounding them with a love which revealed itself in the most touching and
delicate attentions.
If only the world had known what an example the Imperial family were
setting with their tender and intimate association! But how few ever
suspected it! For it was too indifferent to public opinion and avoided the
public gaze.

I
CHAPTER XI
THE RETREAT OF THE RUSSIAN ARMY—THE CZAR PLACES
HIMSELF AT THE HEAD OF HIS ARMY—THE GROWING
INFLUENCE OF THE CZARINA
(FEBRUARY-SEPTEMBER, 1915)
N spite of the successes gained by the Russians in Galicia in the autumn,
the situation was very uncertain in the spring of 1915. On both sides
preparations were being made for a fierce renewal of the struggle to which
the fighting of January and February was only the prelude. On the Russian
side it looked as if everything possible had been done to strengthen the
army’s fighting power and assure the normal flow of supplies. The Czar, at
any rate, believed that it was so, on the faith of the reports he had received.
He had placed all his hopes on the success of this spring campaign.
The Austrians were the first to take the offensive, but the Russians
counter-attacked vigorously, and their superiority was soon made manifest
all along the front. In the first fortnight of March their successes were
continued. On the 19th they captured the fortress of Przemysl. The whole
garrison and considerable booty in war material fell into their hands. There
was tremendous excitement in the country. The Czar returned from G.H.Q.
on March 24th. He was in high spirits. Were the fortunes of war at length
going to turn in favour of Russia?
In the middle of April Russian divisions stood on the crest of the
Carpathians and menaced the rich plains of Hungary. The Austrian army was
at the end of its tether. But these successes had been bought at the price of
enormous losses, and the mountain fighting continued under conditions
which were extremely trying even for the victor. The prolongation of the war
was also beginning to show effects on the population at home. It had begun
to feel the high cost of food and the poverty of communications was
paralysing all economic life. There must be no delay in finding a solution.

But Germany could not remain indifferent to the dissolution of the
Austrian army, and as soon as she clearly appreciated the danger she made
up her mind to avert it by taking every step in her power. Several German
army corps had been massed east of Cracow and placed under the command
of General Mackensen, who was to take the offensive against the flank of the
Russian army and try to cut the communications of the troops operating in
the Carpathians. The onslaught began in the first days of May, and under the
pressure of the Germans the Russian army of Western Galicia was obliged to
retreat rapidly to the east. It had to accept the loss of the Carpathians, the
capture of which had cost so much blood and effort, and descend into the
plains. The troops fought with remarkable courage and endurance, but they
were cruelly short of arms and ammunition.
The retreat continued. On June 5th Przemysl was lost, and on June 22nd
Lemberg. By the end of the month all Galicia—that Slav land the conquest
of which had filled all Russian hearts with joy—had been evacuated.
THE CZAR.

THE CZAREVITCH.
Facing page 134.
Meanwhile the Germans had begun a vigorous offensive in Poland and
made rapid progress in spite of the fierce resistance of the Russians. It was a
grave moment. The whole Russian front had been shaken and given way
under the pressure of the Austro-German armies. Men wished to know who
was responsible for these disasters. They called for the guilty and demanded
their punishment.
The development of events had been a terrible blow to the Czar. It had
been a shock, especially as he had certainly not expected anything of the
kind. But he set his teeth against adversity. On June 25th he dismissed the
Minister of War, General Sukhomlinoff, whose criminal negligence seemed
to have been responsible for the fact that it was impossible to secure the
army’s supplies. He replaced him by General Polivanoff. On the 27th he
summoned a conference at G.H.Q., at which all the Ministers were present.
It was a question of rousing all the energies of the country, of mobilising all
its forces and resources for the life-and-death struggle with the hated foe.
It was decided to summon the Duma. The first sitting took place on
August 1st, the anniversary of the declaration of war by Germany on Russia.
The firm and courageous attitude of the Assembly did a good deal to calm
the public agitation. But while calling on the whole nation to co-operate in
the defence of the Fatherland, the Duma demanded that the guilty should be
discovered and punished. A few days later the Czar appointed a

“Commission of Enquiry” with a view to fixing responsibility for the
nation’s misfortunes.
Meanwhile the German offensive in Poland had made further progress.
On August 5th Warsaw was abandoned by the Russians, who withdrew to
the right bank of the Vistula. On the 17th Kovno was lost. One after the
other all the Russian fortresses fell before the onslaught of the enemy, whose
advance no obstacle seemed capable of staying. By the end of August the
whole of the Government of Poland was in the hands of the Germans.
The reverses assumed the proportions of a catastrophe which endangered
the very existence of the country. Should we be able to stop the invading
hordes or should we have to follow the precedent of 1812 and withdraw into
the interior, thus abandoning Russian soil to the enemy? Had all our willing
sacrifices brought us nothing?
The country was suffering from the incessant withdrawals of men and
from requisitions. Agriculture was short of labour and horses. In the towns
the cost of living was rising with the disorganisation of the railways and the
influx of refugees. The most pessimistic news passed from mouth to mouth.
There was talk of sabotage, treason, etc. Russian public opinion, so
changeable and prone to exaggeration whether in joy or sorrow, indulged in
the most gloomy forebodings.
It was just when Russia was passing through this acute crisis that
Nicholas II. decided to take the command of his armies in person.
For several months the Czarina had been urging the Czar to take this step,
but he had stood out against her suggestion as he did not like the idea of
relieving the Grand-Duke Nicholas of the post he had given him. When the
war broke out his first impulse had been to put himself at the head of his
army, but, yielding to the representations of his Ministers, he had abandoned
an idea which was very close to his heart. He had always regretted it, and
now that the Germans had conquered all Poland and were advancing on
Russian soil, he considered it nothing less than criminal to remain away
from the front and not take a more active part in the defence of his country.
The Czar had returned from G.H.Q. on July 11th, and spent two months
at Tsarskoïe-Selo before making up his mind to this new step. I will relate a
conversation I had with him on July 16th, as it shows quite clearly what
were the ideas that inspired him at that time. On that day he had joined

Alexis Nicolaïevitch and myself in the park, and had just been telling his son
something about his recent visit to the army. Turning to me, he added:
“You have no idea how depressing it is to be away from the front. It
seems as if everything here saps energy and enfeebles resolution. The most
pessimistic rumours and the most ridiculous stories are accepted and get
about everywhere. Folk here care nothing except for intrigues and cabals,
and regard low personal interests only. Out at the front men fight and die for
their country. At the front there is only one thought—the determination to
conquer. All else is forgotten, and, in spite of our losses and our reverses,
everyone remains confident. Any man fit to bear arms should be in the army.
Speaking for myself, I can never be in too much of a hurry to be with my
troops.”[38]
The Czarina was able to take advantage of this great ambition. She set
herself to overcome the scruples which considerations of another character
inspired. She desired the removal of the Grand-Duke Nicholas, whom she
accused of secretly working for the ruin of the Czar’s reputation and prestige
and for a palace revolution which would further his own ends. On the
strength of certain information she had received from Madame Wyroubova,
she was also persuaded that G.H.Q. was the centre of a plot, the object of
which was to seize her daring the absence of her husband and confine her in
a convent.
The Czar, on the other hand, had full confidence in the loyalty of the
Grand-Duke Nicholas. He considered him incapable of any criminal action,
but he was compelled to admit his complicity in the intrigue against the
Czarina. Yet he did not give way until the imperious instinct urging him to
put himself at the head of his army had become an obligation of conscience.
By intervening personally in the struggle he hoped to show the world that
the war would be fought out to the bitter end and prove his own unshakable
faith in ultimate victory. In this tragic hour he thought it was his duty to
stake his own person, and as head of the state to assume the full burden of
responsibility. By his presence among the troops he wished to restore their
confidence, for their morale had been shaken by the long series of reverses,
and they were tired of fighting against an enemy whose strength consisted
principally in the superiority of his armament.
In spite of the recent retreats, the prestige of the Grand-Duke Nicholas
was still considerable in Russia. During this first twelve months of the war
he had given proof of resolution and an iron will. The fact that he was

deprived of his command in times of defeat indicated that he was held
responsible, and was bound to be interpreted as a punishment, as unjust on
the merits as insulting to his honour. The Czar fully realised all this, and only
decided as he did much against his will. His first idea had been to keep the
Grand-Duke with him at G.H.Q., but that would have made the position of
the ex-Generalissimo somewhat delicate. The Czar decided to appoint him
Lieutenant-General of the Caucasus and Commander-in-Chief of the army
operating against the Turks.
The Czar communicated his decision to take over the Supreme Command
to his Ministers at a council which took place at Tsarskoïe-Selo a few days
before his departure for G.H.Q. The news threw most of those present into
utter consternation, and they did their best to dissuade him from his project.
They pointed to the grave difficulties in the way of public business if the
head of the state was to spend practically all his time at G.H.Q., more than
five hundred miles from the seat of government. They referred to his
innumerable duties and asked him not to take new and crushing
responsibilities upon himself. In the last resort they begged him not to place
himself at the head of his troops at a moment so critical. In case of failure he
was running a risk of exposing himself to attacks which would undermine
his prestige and authority.
Yet the Czar was not to be moved. Several members of his immediate
entourage made several further attempts to convince him, but these failed
also, and on the evening of September 4th he left for Mohileff, where
G.H.Q. was established at that time. The next day he signed the Prikase, in
which he announced to the troops that he was taking command in person,
and at the foot he added in his own hand:
“With unshakable faith in the goodness of God and firm confidence in
final victory we shall accomplish our sacred duty in defending our
Fatherland to the end, and we shall never let the soil of Russia be outraged.”
He was repeating the oath he had taken at the outset of the war and
casting his crown into the arena.
In France and England this announcement came as a surprise which was
not without a certain element of apprehension, but this action was regarded
as a pledge which irrevocably associated the Russian Empire, in the person
of its Czar, with the fortunes of the Entente, and this at a moment when a
series of defeats would have been grounds for fearing separatist tendencies.
All the great newspapers of the Allied countries emphasised the importance

of this decision. It was hoped that it would have a considerable effect on the
morale of the Russian army and contribute to further the cause of final
victory. In Russia the whole Press raised a shout of triumph, but in sober
reality opinion about the wisdom of changing the command was sharply
divided at first. In the army itself we shall see that the presence of the Czar
helped to raise the spirits and courage of the men and gave the campaign a
new impetus.
History will some day reveal the political and military consequences of
this step, which was certainly an act of courage and faith on the part of the
Czar himself.
As I had feared, the apparent indifference with which Rasputin had been
treated during the winter had only been temporary, and at the time of the
disasters in May there was a revival of his influence, which grew steadily
stronger. The change is easily explained. At the beginning of the war the
Czar and Czarina were utterly obsessed by the greatness of their task, and
had passed through hours of exaltation in the knowledge of the love they
bore their people, a love they felt was reciprocated. That fervent communion
had filled them with hope. They believed that they were really the centre of
that great national movement which swept over the whole of Russia. The
military events of the following months had not shaken their courage. They
had maintained their ardent faith

THE FOUR GRAND-DUCHESSES.
[Facing page 140.
in that spring offensive which was to bring about the final success of the
Russian armies.
When the great catastrophe followed they passed through a time of
unspeakable anguish. In her sorrow the Czarina was bound to feel impelled
to seek moral support from him whom she already regarded not only as the
saviour of her son, but as the representative of the people, sent by God to
save Russia and her husband also.
It is not true that personal ambition or a thirst for power induced the
Czarina to intervene in political affairs. Her motive was purely sentimental.
She worshipped her husband as she worshipped her children, and there was
no limit to her devotion for those she loved. Her only desire was to be useful
to the Czar in his heavy task and to help him with her counsel.
Convinced that autocracy was the only form of government suited to the
needs of Russia, the Czarina believed that any great concessions to liberal
demands were premature. In her view the uneducated masses of the Russian
people could be galvanised into action only by a Czar in whose person all
power was centralised. She was certain that to the moujik the Czar was the
symbol of the unity, greatness, and glory of Russia, the head of the state and
the Lord’s Anointed. To encroach on his prerogatives was to undermine the
faith of the Russian peasant and to risk precipitating the worst disasters for

the country. The Czar must not merely rule: he must govern the state with a
firm and mighty hand.
To the new task the Czarina brought the same devotion, courage, and,
alas! blindness she had shown in her fight for the life of her son. She was at
any rate logical in her errors. Persuaded, as she was, that the only support for
the dynasty was the nation, and that Rasputin was God’s elect (had she not
witnessed the efficacy of his prayers during her son’s illness?), she was
absolutely convinced that this lowly peasant could use his supernatural
powers to help him who held in his hands the fate of the empire of the Czars.
Cunning and astute as he was, Rasputin never advised in political matters
except with the most extreme caution. He always took the greatest care to be
very well informed as to what was going on at Court and as to the private
feelings of the Czar and his wife. As a rule, therefore, his prophecies only
confirmed the secret wishes of the Czarina. In fact, it was almost impossible
to doubt that it was she who inspired the “inspired,” but as her desires were
interpreted by Rasputin, they seemed in her eyes to have the sanction and
authority of a revelation.
Before the war the influence of the Czarina in political affairs had been
but intermittent. It was usually confined to procuring the dismissal of anyone
who declared his hostility to the staretz. In the first months of the war there
had been no change in that respect, but after the great reverses in the spring
of 1915, and more particularly after the Czar had assumed command of the
army, the Czarina played an ever-increasing part in affairs of state because
she wished to help her husband, who was overwhelmed with the burden of
his growing responsibilities. She was worn out, and desired nothing more
than peace and rest, but she willingly sacrificed her personal comfort to what
she believed was a sacred duty.
Very reserved and yet very impulsive, the Czarina, first and foremost the
wife and mother, was never happy except in the bosom of her family. She
was artistic and well-educated, and liked reading and the arts. She was fond
of meditation, and often became wholly absorbed in her own inward
thoughts and feelings, an absorption from which she would only emerge
when danger threatened. She would throw herself at the obstacle with all the
ardour of a passionate nature. She was endowed with the finest moral
qualities, and was always inspired by the highest ideals. But her sorrows had
broken her. She was but the shadow of her former self, and she often had

periods of mystic ecstasy in which she lost all sense of reality. Her faith in
Rasputin proves it beyond a doubt.
It was thus that in her desire to save her husband and son, whom she
loved more than life itself, she forged with her own hands the instrument of
their undoing.

T
CHAPTER XII
NICHOLAS II. AS COMMANDER-IN-CHIEF—THE ARRIVAL OF THE
CZAREVITCH AT G.H.Q.—VISITS TO THE FRONT
(SEPTEMBER—DECEMBER, 1915)
HE Grand-Duke Nicholas left G.H.Q. on September 7th, two days after
the arrival of the Czar. He left for the Caucasus, taking with him General
Yanushkevitch, who had been replaced as First Quartermaster-General of
the Russian armies by General Alexeieff a short time before. This
appointment had been very well received by military circles, who had high
hopes of Alexeieff. He it was who had drawn up the plan of campaign in
Galicia in the autumn of 1914, and as Commander of the North-Western
Front he had just given further proof of his military talent. The burden which
was now laid upon his shoulders was a crushing one, for as a result of the
irresistible advance of the Germans the Russian army was in a very critical
position, and the decisions which he had to take were exceptionally grave.
From the outset the Czar gave him an entirely free hand with regard to the
operations, confining himself to covering him with his authority and taking
responsibility for everything he did.
A few days after Nicholas II. took over the Supreme Command the
situation suddenly took a turn for the worse. The Germans, who had massed
large forces north-west of Vilnam, had succeeded in breaking the Russian
front, and their cavalry was operating in the rear of the army and threatening
its communications. On September 18th we seemed on the verge of a great
disaster.
Thanks to the skill of the dispositions which were taken and the
endurance and heroism of the troops, the peril was averted. This was the last
effort of the enemy, who himself had shot his bolt. In the early days of
October the Russians in turn gained a success over the Austrians, and
gradually the immense front became fixed and both sides went to ground.
This marked the end of the long retreat which had begun in May. In spite
of all their efforts the Germans had not obtained a decision. The Russian
armies had abandoned a large stretch of territory, but they had everywhere
escaped the clutches of their foes.

The Czar returned to Tsarskoïe-Selo on October 6th for a few days, and it
was decided that Alexis Nicolaïevitch should go back with him to G.H.Q.,
for he was most anxious to show the Heir to the troops. The Czarina bowed
to this necessity. She realised how greatly the Czar suffered from loneliness,
for at one of the most tragic hours of his life he was deprived of the presence
of his family, his greatest consolation. She knew what a comfort it would be
to have his son with him. Yet her heart bled at the thought of Alexis leaving
her. It was the first time she had been separated from him, and one can
imagine what a sacrifice it meant to the mother, who never left her child,
even for a few minutes, without wondering anxiously whether she would
ever see him alive again.
We left for Mohileff on October 14th, and the Czarina and the Grand-
Duchesses came to the station to see us off. As I
THE CZAR AND CZAREVITCH ON THE BANKS OF THE DNIEPER. SUMMER OF 1916.

THE CZAR AND CZAREVITCH NEAR MOHILEFF. SUMMER OF 1916.
[Facing page 148.
was saying good-bye to her, Her Majesty asked me to write every day to
give her news of her son. I promised to carry out her wishes faithfully the
whole time we were away.
The next day we stopped at Riegitza, where the Czar wished to review
some troops which had been withdrawn from the front and were billeted in
the neighbourhood. All these regiments had taken part in the exhausting
campaigns in Galicia and the Carpathians, and their establishment had been
almost entirely renewed two or three times over. But in spite of the terrible
losses they had suffered, they marched past the Czar with a proud and
defiant bearing. Of course, they had been resting behind the line for several
weeks, and had had time to recover from their weariness and privations. It
was the first time that the Czar had passed any of his troops in review since
he had taken over the Command. They now looked upon him both as their
Emperor and Generalissimo. After the ceremony he mixed with the men and
conversed personally with several of them, asking questions about the severe
engagements in which they had taken part. Alexis Nicolaïevitch was at his
father’s heels, listening intently to the stories of these men, who had so often
stared death in the face. His features, which were always expressive, became
quite strained in the effort not to lose a single word of what the men were
saying. His presence at the Czar’s side greatly interested the soldiers, and

when he had gone they were heard exchanging in a whisper their ideas about
his age, size, looks, etc. But the point that made the greatest impression upon
them was the fact that the Czarevitch was wearing the uniform of a private
soldier, which had nothing to distinguish it from that of a boy in the service.
On October 16th we arrived at Mohileff, a little White Russian town of a
highly provincial appearance to which the Grand-Duke Nicholas had
transferred G.H.Q. during the great German offensive two months before.
The Czar occupied the house of the Governor, which was situated on the
summit of the steep left bank of the Dnieper. He was on the first floor in two
fairly large rooms, one of which was his study and the other his bedroom. He
had decided that his son should share his room. Alexis Nicolaïevitch’s camp-
bed was accordingly placed next to his father’s. I myself and some members
of the Czar’s military suite were lodged in the local court-house, which had
been converted for use by G.H.Q.
Our time was spent much as follows. Every morning at half-past nine the
Czar called on the General Staff. He usually stayed there until one o’clock,
and I took advantage of his absence to work with Alexis Nicolaïevitch in his
study, which we had been obliged to make our workroom owing to lack of
space. We then took lunch in the main room of the Governor’s house. Every
day there were some thirty guests, which included General Alexeieff, his
principal assistants, the heads of all the military missions of the Allies, the
suite, and a few officers who were passing through Mohileff. After lunch the
Czar dealt with urgent business and then about three we went for a drive in a
car.
When we had proceeded a certain distance from the town we stopped and
went for a walk in the neighbourhood for an hour. One of our favourite
haunts was the pretty pine-wood in the heart of which is the little village of
Saltanovka, where the army of Marshal Davout met the troops of General
Raievsky on July 29th, 1912.[39] On our return the Czar resumed work while
Alexis Nicolaïevitch prepared the lessons for the next day in his father’s
study. One day when I was there as usual the Czar turned towards me, pen in
hand, and interrupted me in my reading to remark abruptly:
“If anyone had told me that I should one day sign a declaration of war on
Bulgaria I should have called him a lunatic. Yet that day has come. But I am
signing against my will, as I am certain that the Bulgarian people have been
deceived by their king and the partisans of Austria, and that the majority

remain friendly to Russia. Race feeling will soon revive and they will realise
their mistake, but it will be too late then.”
The incident shows what a simple life we led at G.H.Q., and the intimacy
which was the result of the extraordinary circumstances under which I was
working.
As the Czar was anxious to visit the troops with the Czarevitch, we left
for the front on October 24th. The next day we arrived at Berditcheff, where
General Ivanoff, commanding the South-Western Front, joined our train. A
few hours later we were at Rovno. It was in this town that General Brussiloff
had established his headquarters, and we were to accompany him to the
place where the troops had been assembled. We went by car, as we had more
than twelve miles to cover. As we left the town a squadron of aeroplanes
joined us and escorted us until we saw the long grey lines of the units
massed behind a forest. A minute later we were among them. The Czar
walked down the front of the troops with his son, and then each unit defiled
in turn before him. He then had the officers and men on whom decorations
were to be bestowed called out of the ranks and gave them the St. George’s
Cross.
It was dark before the ceremony was over. On our return the Czar, having
heard from General Ivanoff that there was a casualty station quite near,
decided to visit it at once. We entered a dark forest and soon perceived a
small building feebly lit by the red flames of torches. The Czar and Alexis
Nicolaïevitch entered the house, and the Czar went up to all the wounded
and questioned them in a kindly way. His unexpected arrival at so late an
hour at a spot so close to the front was the cause of the general astonishment
which could be read on every face. One private soldier, who had just been
bandaged and put back in bed, gazed fixedly at the Czar, and when the latter
bent over him he raised his only sound hand to touch his sovereign’s clothes
and satisfy himself that it was really the Czar who stood before him and not
a ghost. Close behind his father stood Alexis Nicolaïevitch, who was deeply
moved by the groaning he heard and the suffering he felt all around him.
We rejoined our train and immediately left for the south. When we woke
next morning we were in Galicia. During the night we had crossed the
former Austrian frontier. The Czar was anxious to congratulate the troops,
whose prodigies of valour had enabled them to remain on hostile soil
notwithstanding the dearth of arms and ammunition. We left the railway at

Bogdanovka and gradually mounted the plateau on which units from all the
regiments of General Tcherbatcheff’s army had been assembled. When the
review was over the Czar disregarded the objections of his suite and visited
the Perchersky Regiment, three miles from the front lines, at a place which
enemy artillery fire could have reached. We then returned to our cars, which
we had left in a forest, and went to General Lechitzsky’s army, which was
some thirty miles away. We were overtaken by darkness on our way back. A
thick mist covered the countryside; we lost our way and twice had to go
back. But after many wanderings we at length struck the railway again,
though we were sixteen miles from the place where we had left our train!
Two hours later we left for G.H.Q.
The Czar brought away a most encouraging impression from his tour of
inspection. It was the first time that he had been in really close contact with
the troops, and he was glad that he had been able to see with his own eyes,
practically in the firing-line, the fine condition of the regiments and the
splendid spirit with which they were inspired.
We returned to Mohileff in the evening of October 27th, and the next
morning Her Majesty and the Grand-Duchesses also arrived at G.H.Q.
During their journey the Czarina and her daughters had stopped at several
towns in the Governments of Tver, Pskoff, and Mohileff, in order to visit the
military hospitals. They stayed three days with us at Mohileff and then the
whole family left for Tsarskoïe-Selo, where the Czar was to spend several
days.
I have somewhat lingered over the first journey which the Czar made
with his son, and to avoid mere repetition I shall confine myself to a short
summary of the visits we paid to the armies in the month of November.
We left Tsarskoïe-Selo on the 9th. On the 10th we were at Reval, where
the Czar visited a flotilla of submarines which had just come in. The boats
were covered with a thick coating of ice, a sparkling shell for them. There
were also two English submarines which had surmounted enormous
difficulties in penetrating into the Baltic, and had already succeeded in
sinking a certain number of German ships. The Czar bestowed the St.
George’s Cross on their commanding officers.
During our next day at Riga, which formed a kind of advanced bastion in
the German lines, we spent several hours with the splendid regiments of
Siberian Rifles, which were regarded as some of the finest troops in the
Russian army. Their bearing was magnificent, as they marched past before

the Czar, answering his salute with the traditional phrase: “Happy to serve
Your Imperial Majesty,” followed by a tremendous round of cheers.
A few days later we were at Tiraspol, a little town sixty miles north of
Odessa, where the Czar reviewed units from the army of General
Tcherbatcheff. After the ceremony the Czar, desiring to know for himself
what losses the troops had suffered, asked their commanding officers to
order all men who had been in the ranks since the beginning of the campaign
to raise their hands. The order was given, and but a very few hands were
lifted above those thousands of heads. There were whole companies in
which not a man moved. The incident made a very great impression on
Alexis Nicolaïevitch. It was the first time that reality had brought home to
him the horrors of war in so direct a fashion.
The next day, November 22nd, we went to Reni, a small town on the
Danube on the Rumanian frontier. An immense quantity of supplies had been
collected there, for it was a base for the river steamers which were engaged
in taking food, arms and ammunition to the unfortunate Serbians whom the
treachery of Bulgaria had just exposed to an Austro-German invasion.
The following day, near Balta in Podolia, the Czar inspected the famous
division of Caucasian cavalry whose regiments had won new laurels in the
recent campaign. Among other units were the Kuban and Terek Cossacks,
perched high in the saddle
THE CZAR AND THE CZAREVITCH AT A RELIGIOUS SERVICE AT G.H.Q., MOHILEFF.
[Facing page 154.

and wearing the huge fur caps which make them look so fierce. As we
started to return, the whole mass of cavalry suddenly moved forward, took
station on both sides of the road, broke into a gallop, tearing up the hills,
sweeping down the banks of ravines, clearing all obstacles, and thus escorted
us to the station in a terrific charge in which men and animals crashed
together on the ground while above the mêlée rose the raucous yells of the
Caucasian mountaineers. It was a spectacle at once magnificent and terrible
which revealed all the savage instincts of this primitive race.
We did not return to G.H.Q. until November 26th, after having visited
practically the whole of the immense front from the Baltic to the Black Sea.
On December 10th we heard that the Czar was intending to visit the
regiments of the Guard which were then on the frontier of Galicia. On the
morning of our departure, Thursday, December 16th, Alexis Nicolaïevitch,
who had caught cold the previous day and was suffering from a heavy
catarrh in the head, began to bleed at the nose as a result of sneezing
violently. I summoned Professor Fiodrof,[40] but he could not entirely stop
the bleeding. In spite of this accident we started off, as all preparations had
been made for the arrival of the Czar. During the night the boy got worse.
His temperature had gone up and he was getting weaker. At three o’clock in
the morning Professor Fiodrof, alarmed at his responsibilities, decided to
have the Czar roused and ask him to return to Mohileff, where he could
attend to the Czarevitch under more favourable conditions.
The next morning we were on our way back to G.H.Q., but the boy’s state
was so alarming that it was decided to take him back to Tsarskoïe-Selo. The
Czar called on the General Staff and spent two hours with General Alexeieff.
Then he joined us and we started off at once. Our journey was particularly
harrowing, as the patient’s strength was failing rapidly. We had to have the
train stopped several times to be able to change the plugs. Alexis
Nicolaïevitch was supported in bed by his sailor Nagorny (he could not be
allowed to lie full length), and twice in the night he swooned away and I
thought the end had come.
Towards morning there was a slight improvement, however, and the
hæmorrhage lessened. At last we reached Tsarskoïe-Selo. It was eleven
o’clock. The Czarina, who had been torn with anguish and anxiety, was on
the platform with the Grand-Duchesses. With infinite care the invalid was
taken to the palace. The doctors ultimately succeeded in cauterizing the scar

which had formed at the spot where a little blood-vessel had burst. Once
more the Czarina attributed the improvement in her son’s condition that
morning to the prayers of Rasputin, and she remained convinced that the boy
had been saved thanks to his intervention.
The Czar stayed several days with us, but he was anxious to get away as
he was wishful to take advantage of the comparative stagnation at the front
to visit the troops and get into the closest possible touch with them.
His journeys to the front had been a great success. His presence had
everywhere aroused immense enthusiasm, not only among the men but also
among the peasants, who swarmed in from the country round whenever his
train stopped, in the hope of catching a glimpse of their sovereign. The Czar
was certain that his efforts would tend to revive feelings of patriotism and
personal loyalty in the nation and the army. His recent experiences
persuaded him that he had succeeded, and those who went with him thought
the same. Was it an illusion? He who denies its truth can know little of the
Russian people, and cannot have the slightest idea how deep-rooted was
monarchical sentiment in the moujik.

T
CHAPTER XIII
THE CZAR AT THE DUMA—THE CAMPAIGN IN GALICIA—OUR
LIFE AT G.H.Q.—GROWING DISAFFECTION IN THE REAR
(1916)
HE Czar had returned to G.H.Q. alone on December 25th, and three days
later he reviewed on the Galician frontier the divisions of the Guard
which had been concentrated in view of an imminent offensive. The
absence of Alexis Nicolaïevitch was a real sorrow to him, as he had been
looking forward eagerly to presenting him to his Guard. He had then
returned to Mohileff.
Towards the end of the year 1915 the military situation of the Russians
had greatly improved. The army had taken advantage of the quiet months
which followed the conclusion of the great German offensive at the end of
September, 1915, and, thanks to the enormous reserves in man-power at the
disposal of the country, it had easily made good the very heavy losses it had
suffered in the retreat. Once more the Germans found themselves baulked of
the great prize they had promised themselves—a prize which their brilliant
successes at the opening of the campaign seemed to have assured. They had
growing doubts about their ability to overcome the stubborn Russian
resistance by arms, and by clever propaganda and cunning intrigues they
were now endeavouring to stir up such disaffection in the interior of the
country as would hasten, they hoped, the consummation so devoutly to be
desired. But in the person of the Czar they found an insurmountable obstacle
to the realisation of their designs. That obstacle must be removed.
By assuming the command of his troops and thus staking his crown on
the struggle, the Czar had definitely deprived his enemies of all hopes of a
reconciliation. At Berlin the authorities now knew that Nicholas II. would
stand by his allies to the bitter end, and that all attempts at a rapprochement
would be broken against his unswerving determination to continue the war at
any cost. They also knew that the Czar was the sole bond between the

different parties in the Empire, and that once it was removed no organised
power would be capable of averting dismemberment and anarchy.
The German General Staff therefore devoted itself unceasingly to ruin the
prestige of the monarchy and bring about the downfall of the Czar. To attain
that object the essential step was to compromise the Czar in the eyes of his
people and his allies. Germany had in Russia many sources of intelligence
and powerful means of action, and she devoted herself to spreading the idea
that the Czar was thinking of liquidating the war and making a separate
peace.
The Czar decided to nip these intrigues in the bud and to define his
intentions beyond doubt. On January 2nd, at Zamirie, where he was
inspecting the regiments of General Kuropatkin’s army, he ended his address
to the troops with the following formal declaration:
“You need have no fear. As I announced at the beginning of the war, I
will not make peace until we have driven the last enemy soldier beyond our
frontiers, nor will I conclude peace except by agreement with our allies, to
whom we are bound not only by treaties but by sincere friendship and the
blood spilt in a common cause.”
Nicholas II. thus confirmed in the presence of his army that solemn
compact which had been entered upon on August 2nd, 1914, and renewed
when he had become Commander-in-Chief of the Russian armies. The
Government was anxious to give the widest possible publicity to the Czar’s
speech, and had it printed and distributed among the armies and in the
country districts.
In January and February the Czar continued his visits to the front and
G.H.Q. (it was at Mohileff that he spent the Russian New Year), and returned
to Tsarskoïe-Selo on February 21st, the day before the opening of the Duma.
Five days before, the news of the capture of the fortress of Erzerum, which
had so long been the backbone of the Turkish resistance, had caused great
joy throughout Russia. It was certainly a fine success, and the offensive of
the army of the Caucasus continued to make rapid headway.
The morning after his arrival the Czar carried out his intention of going
with his brother, the Grand-Duke Michael, to the Tauride Palace, where the
Duma was to resume its labours that day. It was the first time that the
representatives of the nation had received a visit from their sovereign, and in
political circles great importance was attached to this historical event. It bore

witness to the Czar’s ardent desire for closer co-operation with the people’s
representatives, and the step was particularly warmly welcomed, as
confidence in the Government had been shaken as the result of the reverses
suffered by the army and the crushing charges made against the former
Minister of War, General Sukhomlinoff.
The Czar was received on his arrival at the Tauride Palace by M.
Rodzianko, President of the Duma, who conducted him into the Catherine
Hall, where he was present at a Te Deum to celebrate the capture of Erzerum.
Then turning to the deputies, the Czar expressed his great pleasure at being
among them, and voiced his absolute conviction that in the tragic days
through which Russia was passing they would all unite their efforts and
work together in perfect harmony for the welfare of the country. His words
were received with vociferous cheers.
The Czar withdrew after a visit to the chambers and offices of the Tauride
Palace. Half an hour later the President, in opening the session, ended his
speech with these words:
“The direct association of the Czar with his people, that benefit which is
inestimable and indispensable to the prosperity of the Russian Empire, is
now strengthened by a tie which is still more potent. This good news will fill
all hearts with, joy even in the remotest corners of our land, and give fresh
courage to our glorious soldiers, the defenders of their country.”
On that memorable day it seemed that the sovereign, the Ministers, and
the representatives of the nation had one thought, and one thought alone—to
conquer at whatever cost.
The same evening the Czar went to the Council of State, which was also
resuming its labours that day. Then he returned to Tsarskoïe-Selo, which he
left next morning for G.H.Q. This was the time of the great onslaught on
Verdun, and it was essential that Russia should intervene without delay in
order to draw a larger portion of the German forces upon herself. It was
decided to take the offensive.
The attack was launched about March 15th in the Dvinsk and Vilna
sectors, and at first it was crowned with success, but the progress of the
Russians was slow, for the Germans offered a very stubborn resistance.
There had been a thaw, the roads were almost impracticable, and the men
had to wade through mud and marsh. The attack died down about the
beginning of April and soon came to a standstill. Yet the diversion had borne

fruit, for the Germans had found themselves compelled to send considerable
reinforcements to the threatened sectors.
Alexis Nicolaïevitch had remained very weak as the result of the
excessive hæmorrhage which had so endangered his life in December. It was
February before he was quite strong again, but the Czarina had learned from
experience, and intended to keep him at Tsarskoïe-Selo until the return of the
fine weather.[41]
I was far from complaining of the Czarina’s decision, for the Czarevitch’s
education was suffering as the result of our long visits to the front.
We did not return to G.H.Q. until May 17th. The Czar was to remain
there for a considerable time. A fortnight after our arrival—on June 4th—the
great offensive of General Brussiloff opened in Galicia. It was a complete
triumph, and our successes were greatly extended in the following days.
Under the pressure of the Russian army the Austrian front gave way and was
withdrawn towards Lemberg. The number of prisoners was very large, and
the situation of the Austrians in the Lutzk sector became highly critical. The
news of this fine victory was received with immense enthusiasm at G.H.Q. It
was to be the last cause of rejoicing for the Czar.
Since our return to Headquarters our life had followed the same course as
during our previous visits, though I no longer gave the Czarevitch his
lessons in his father’s study, but in a little verandah which we had converted
into a schoolroom or in a large tent in the garden, which was also our dining-
room. It was here that the Czar took his meals after the hot weather began.
We took advantage of the fine summer days to go sailing on the Dnieper. We
had the use of a small yacht which had been placed at our disposal by the
Ministry of Ways and Communications.
From time to time the Czarina and the Grand-Duchesses paid short visits
to G.H.Q. They lived in their train, but joined the Czar at lunch and came
with us on our excursions. The Czar in return dined with the Czarina and
spent part of the evening with his family whenever he could. The Grand-
Duchesses greatly enjoyed these visits to Mohileff—all too short to their
taste—which meant a little change in their monotonous and austere lives.
They had far more freedom here than at Tsarskoïe-Selo. As is so often the
case in Russia, the station at Mohileff was a very long way from the town
and almost in the open country. The Grand-Duchesses spent their spare time

visiting the peasants of the neighbourhood or the families of railway
employees. Their simple ways and natural kindness soon won all hearts, and
as they adored children you could see them always accompanied by a mob
of ragamuffins collected on their walks and duly stuffed with sweets.
Unfortunately, life at Mohileff grievously interrupted Alexis
THREE OF THE GRAND-DUCHESSES (OLGA, ANASTASIE, AND TATIANA) VISITING THE
WIFE AND CHILDREN OF A RAILWAY EMPLOYEE AT MOHILEFF.

THE CZARINAAND THE GRAND-DUCHESS TATIANA TALKING TO REFUGEES.
MOHILEFF, MAY, 1916.
[Facing page 166.
Nicolaïevitch’s studies and was also bad for his health. The impressions
he gained there were too numerous and exciting for so delicate a nature as
his. He became nervous, fretful, and incapable of useful work. I told the
Czar what I thought. He admitted that my objections were well founded, but
suggested that these drawbacks were compensated for by the fact that his son
was losing his timidity and natural wildness, and that the sight of all the
misery he had witnessed would give him a salutary horror of war for the rest
of his life.
But the longer we stayed at the front the stronger was my conviction that
it was doing the Czarevitch a lot of harm. My position was becoming
difficult, and on two or three occasions I had to take strong steps with the
boy. I had an idea that the Czar did not entirely approve, and did not back me
up as much as he might have done. As I was extremely tired by my work in
the last three years—I had had no holiday since September, 1913—I decided
to ask for a few weeks’ leave. My colleague, M. Petroff, came to take my
place, and I left General Headquarters on July 14th.
As soon as I arrived at Tsarskoïe-Selo the Czarina summoned me, and I
had a long talk with her, in the course of which I tried to show the grave
disadvantages for Alexis Nicolaïevitch of his long visits to the front. She
replied that the Czar and herself quite realised them, but thought that it was

better to sacrifice their son’s education temporarily, even at the risk of
injuring his health, than to deprive him of the other benefits he was deriving
from his stay at Mohileff. With a candour which utterly amazed me she said
that all his life the Czar had suffered terribly from his natural timidity and
from the fact that as he had been kept too much in the background he had
found himself badly prepared for the duties of a ruler on the sudden death of
Alexander III. The Czar had vowed to avoid the same mistakes in the
education of his son.
I realised that I had come up against a considered decision, and was not
likely to secure any modification. All the same, it was agreed that Alexis
Nicolaïevitch’s lessons should be resumed on a more regular plan at the end
of September, and that I should receive some assistance in my work.
When our conversation was over the Czarina made me stay behind to
dinner. I was the only guest that evening. After the meal we went out on the
terrace. It was a beautiful summer evening, warm and still. Her Majesty was
stretched on a sofa, and she and two of her daughters were knitting woollen
clothing for the soldiers. The two other Grand-Duchesses were sewing.
Alexis Nicolaïevitch was naturally the principal topic of conversation. They
never tired of asking me what he did and said. I spent an hour thus in this
homely and quiet circle, suddenly introduced into the intimacy of that family
life which etiquette had forbidden me from entering, save in this casual and
rare fashion.
In the days following I spent my time in a round of visits and renewing
relationships which my journeys to the front had compelled me to neglect. I
thus saw people in different strata of society in the capital, and was not slow
to realise that far-reaching changes had taken place in public opinion in
recent months. People did not confine themselves to violent attacks on the
Government, but went on to attack the person of the Czar.
Since that memorable February 22nd on which Nicholas II. had presented
himself to the Duma in his sincere desire for reconciliation, the differences
between the sovereign and the representatives of the nation had only
increased. The Czar had long been hesitating to grant the liberal concessions
which had been demanded. He considered it was the wrong time, and that it
was dangerous to attempt reforms while the war was raging. It was not that
he clung to his autocratic personal prerogatives, for he was simplicity and
modesty itself, but he feared the effect such radical changes might have at so

critical a moment. When the Czar declared on February 22nd that he was
happy to be among the representatives of his people, the Czar had spoken his
real thoughts. In inviting them to unite all their efforts for the welfare of the
country in the tragic days through which it was passing, he was urging them
to forget all their political differences and have only one goal—victory and
belief in their Czar until the end of the war.
Why did he not make a solemn promise that day to give the nation the
liberties they asked as soon as circumstances permitted? Why did he not try
to recover by his acts that confidence of the Duma which he felt he was
losing? The answer is that those around him had made it impossible for him
to find out for himself what was really going on in the country.
The Czar’s visit to the Tauride Palace had given rise to great hopes. They
had not been fulfilled, and men were not slow to see that nothing had been
changed. The conflict with the Government was immediately resumed. The
demands became more pressing and recrimination more violent. Frightened
by the false reports of those who abused his confidence, the Czar began to
regard the opposition of the Duma as the result of revolutionary agitation,
and thought he could re-establish his authority by measures which only
swelled the general discontent.
But it was the Czarina who was the special object of attack. The worst
insinuations about her conduct had gained currency and were believed even
by circles which hitherto had rejected them with scorn. As I have said, the
presence of Rasputin at Court was a growing blot on the prestige of the
sovereigns, and gave rise to the most malicious comments. It was not as if
the critics confined themselves to attacks upon the private life of the
Czarina. She was openly accused of Germanophile sympathies, and it was
suggested that her feelings for Germany could become a danger to the
country. The word “treason” was not yet heard, but guarded hints showed
that the suspicion had been planted in a good many heads. I knew that all
this was the result of German propaganda and intrigues.[42]
I have explained above that in the autumn of 1915 the Berlin Government
had realised that they could never overthrow Russia as long as she stood
united round her Czar, and that from that moment her one idea had been to
provoke a revolution which would involve the fall of Nicholas II. In view of
the difficulties of attacking the Czar directly, the Germans had concentrated
their efforts against the Czarina and begun a subterranean campaign of
defamation against her. It was skilfully planned and began to show results

Welcome to Our Bookstore - The Ultimate Destination for Book Lovers
Are you passionate about books and eager to explore new worlds of
knowledge? At our website, we offer a vast collection of books that
cater to every interest and age group. From classic literature to
specialized publications, self-help books, and children’s stories, we
have it all! Each book is a gateway to new adventures, helping you
expand your knowledge and nourish your soul
Experience Convenient and Enjoyable Book Shopping Our website is more
than just an online bookstore—it’s a bridge connecting readers to the
timeless values of culture and wisdom. With a sleek and user-friendly
interface and a smart search system, you can find your favorite books
quickly and easily. Enjoy special promotions, fast home delivery, and
a seamless shopping experience that saves you time and enhances your
love for reading.
Let us accompany you on the journey of exploring knowledge and
personal growth!
ebookgate.com

Sensory Integration and the Unity of Consciousness 1st Edition David Bennett

More Related Content

Similar to Sensory Integration and the Unity of Consciousness 1st Edition David Bennett (20)

Recently uploaded (20)

Sensory Integration and the Unity of Consciousness 1st Edition David Bennett