The Somatosensory System - Deciphering The Brain's Own Body Image
The Somatosensory System - Deciphering The Brain's Own Body Image
The Somatosensory System - Deciphering The Brain's Own Body Image
SOMATOSENSORY SYSTEM
Deciphering the Brains Own Body Image
Published Titles
Apoptosis in Neurobiology
Yusuf A. Hannun, M.D., Professor/Biomedical Research and Department Chairman/
Biochemistry and Molecular Biology, Medical University of South Carolina
Rose-Mary Boustany, M.D., tenured Associate Professor/Pediatrics and Neurobiology,
Duke University Medical Center
Methods for Neural Ensemble Recordings
Miguel A.L. Nicolelis, M.D., Ph.D., Professor/Department of Neurobiology,
Duke University Medical Center
Methods of Behavioral Analysis in Neuroscience
Jerry J. Buccafusco, Ph.D., Professor/Pharmacology and Toxicology,
Professor/Psychiatry and Health Behavior, Medical College of Georgia
Neural Prostheses for Restoration of Sensory and Motor Function
John K. Chapin, Ph.D., MCP and Hahnemann School of Medicine
Karen A. Moxon, Ph.D., Department of Electrical and Computer Engineering,
Drexel University
Computational Neuroscience: Realistic Modeling for Experimentalists
Eric DeSchutter, M.D., Ph.D., Department of Medicine, University of Antwerp
Methods in Pain Research
Lawrence Kruger, Ph.D., Professor Emeritus/Neurobiology, UCLA School of Medicine
Motor Neurobiology of the Spinal Cord
Timothy C. Cope, Ph.D., Department of Physiology, Emory University School of Medicine
Nicotinic Receptors in the Nervous System
Edward Levin, Ph.D., Associate Professor/Department of Pharmacology and Molecular
Cancer Biology and Department of Psychiatry and Behavioral Sciences,
Duke University School of Medicine
Methods in Genomic Neuroscience
Helmin R. Chin, Ph.D., NIMH, NIH Genetics Research
Steven O. Moldin, Ph.D, NIMH, NIH Genetics Research
Methods in Chemosensory Research
Sidney A. Simon, Ph.D., Professor/Department of Neurobiology,
Duke University Medical Center
Miguel A.L. Nicolelis, M.D., Ph.D., Professor/Department of Neurobiology,
Duke University Medical Center
CRC PR E S S
Boca Raton London New York Washington, D.C.
2002 by CRC Press LLC
Library of Congress Cataloging-in-Publication Data
This book contains information obtained from authentic and highly regarded sources. Reprinted material
is quoted with permission, and sources are indicated. A wide variety of references are listed. Reasonable
efforts have been made to publish reliable data and information, but the author and the publisher cannot
assume responsibility for the validity of all materials or for the consequences of their use.
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic
or mechanical, including photocopying, microfilming, and recording, or by any information storage or
retrieval system, without prior permission in writing from the publisher.
All rights reserved. Authorization to photocopy items for internal or personal use, or the personal or
internal use of specific clients, may be granted by CRC Press LLC, provided that $1.50 per page
photocopied is paid directly to Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923 USA.
The fee code for users of the Transactional Reporting Service is ISBN 0-8493-2336-3/02/$0.00+$1.50. The
fee is subject to change without notice. For organizations that have been granted a photocopy license by
the CCC, a separate system of payment has been arranged.
The consent of CRC Press LLC does not extend to copying for general distribution, for promotion, for
creating new works, or for resale. Specific permission must be obtained in writing from CRC Press LLC
for such copying.
Direct all inquiries to CRC Press LLC, 2000 N.W. Corporate Blvd., Boca Raton, Florida 33431.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation, without intent to infringe.
Maria Chu
Department of Physiology
and Neuroscience
New York University School
of Medicine
New York, New York
Daniel J. Debowy
Department of Physiology
and Neuroscience
New York University School
of Medicine
New York, New York
Erika Fanselow
Department of Neurobiology
Duke University Medical Center
Durham, North Carolina
Edward H. Hu
Department of Physiology
and Neuroscience
New York University School
of Medicine
New York, New York
Shari Reitzen
Department of Physiology
and Neuroscience
New York University School
of Medicine
New York, New York
Jill Sakai
Department of Physiology
and Neuroscience
New York University School
of Medicine
New York, New York
Sergiy Yakovenko
Centre for Neuroscience
University of Alberta
Edmonton, Alberta
Chapter 2
Cerebral Cortical Regions Devoted to the Somatosensory System:
Results from Brain Imaging Studies in Humans
Harold Burton
Chapter 3
Neural Mechanisms of Tactile Form and Texture Perception
Kenneth O. Johnson and Takashi Yoshioka
Chapter 4
Feeling Surfaces and Objects Remotely
Susan J. Lederman and Roberta L. Klatzky
Chapter 5
Somatosensory and Proprioceptive Contributions to Body Orientation,
Sensory Localization, and Self-Calibration
James R. Lackner and Paul DiZio
Chapter 6
Locomotor Control: From Spring-like Reactions of Muscles
to Neural Prediction
Arthur Prochazka and Sergiy Yakovenko
Chapter 7
Sensory Learning and the Brains Body Map
Mathew E. Diamond, Justin A. Harris,
and Rasmus S. Petersen
Chapter 8
Attention in the Somatosensory System
Steven S. Hsiao and Francisco Vega-Bermudez
Chapter 10
Deciphering the Code Dynamic Modulation of Neural Activity
During Tactile Behavior
Randall J. Nelson, Mikhail A. Lebedev, and Yu Liu
Chapter 11
A Computational Perspective on Proprioception and
Movement Guidance in Parietal Cortex
Paul E. Cisek
Chapter 12
A Critique of the Pure Feedforward Model of Touch
Miguel A. L. Nicolelis, Erika Fanselow, and Craig Henriquez
Chapter 13
The Changeful Mind: Plasticity in the Somatosensory System
Sherre L. Florence
Chapter 14
Functional Implications of Plasticity and Reorganizations in the
Somatosensory and Motor Systems of Developing and Adult Primates
Jon H. Kaas
CONTENTS
1.1 Introduction
1.2 The Basic Somatosensory System
1.3 The Somatosensory System in Prosimian Primates
1.4 The Simian Elaboration
1.5 Somatosensory Cortex in Chimpanzees and Humans
1.6 Conclusions
References
1.1 INTRODUCTION
An understanding of how the somatosensory system processes and uses sensory
information depends in large part on knowing how the system is organized. One
might think that our understanding of the system would have been complete some-
time ago, but actually our knowledge is far from complete, and considerable research
is yet needed. There are several reasons for our incomplete understanding. Most
importantly, evolution has created considerable variability in the organization of the
mammalian somatosensory system. Part of the variability stems from adaptations
that lead to the use of different parts of the body to gather important sensory
information. Many mammals use the whiskers on the face, nose, and mouth for
exploring the environment, but this is a bit risky for primates, which have emphasized
binocular vision with forward directed eyes and a short snout. Explorations with the
face risk damage to the eyes, and reaching and feeling with the hand became a
suitable substitute. Some New World monkeys also use their highly sensitive pre-
hensile tail. But primates have adapted in other ways as well, and have evolved into
one of the most varied orders in terms of brain size and organization. Prosimian
primates occupy a range of environmental niches, but have only moderate brain
expansion. New World monkeys vary from small marmosets which have brains that
are yet large relative to body size, to spider and capuchin monkeys with large brains
that at least superficially resemble those of Old World macaque monkeys. Old World
caudal area of the flying fox has been referred to as area 1/2 (Krubitzer and Calford,
1992) as it is in the same relative position as area 1 of primates and appears to be
responsive to both cutaneous and deep receptors as area 2 of primates.
In all investigated mammals, S1 systematically represents the mechanoreceptors
of the skin of the opposite side of the body (see Kaas, 1983). The body parts are
usually represented from tail to tongue in a mediolateral sequence. The representa-
tions of different body parts often have a morphological counterpart in the cortex
that can be visualized using appropriate histochemical techniques. These isomorphs
of the body are best known for S1 of rats and mice, where an orderly arrangement
of oval-like aggregates of neurons, one for each whisker on the side of the face,
have long been described as the cortical barrels (Woolsey and Van Der Loos, 1970).
Metabolic markers, cytochrome oxidase (CO), and succinic dehydrogenase have
been used to reveal more of the isomorph including discrete cellular clusters for
other whiskers and for the digits and pads of the forepaws and hindpaws (Dawson
and Killackey, 1987; Li et al., 1990; Pearson et al., 1996). An isomorphic array of
cortical bands, one for each of the 11 rays of the contralateral half of the nose, is
especially prominent for the star-nosed mole (Figure 1.2; Catania and Kaas, 1995,
1996). Such isomorphs help identify S1, demonstrate existence of a single systematic
representation in S1, and distinguish S1 from other representations. The star-nosed
mole is unusual in that two other cortical representations, S2 and possibly PV, also
have isomorphs of the rays of the nose. In addition, S2 also has a very prominent
isomorph of the forepaw, while a comparable isomorph of the forepaw does not
exist in S1 (Catania, 2000). This suggests a more pronounced or specialized role
for S2 for forepaw afferents than for S1 in moles.
In a wide range of mammals, S1 projects directly to S2, PV, SR, and SC (see
Beck et al., 1996). Thus, all of these areas are involved in further processing of
information from S1, as well as processing inputs from the thalamus. S2 and PV
receive VP inputs, while SR and SC receive most of their thalamic inputs from
1996). Somatosensory cortex includes a primary area, S1, rostral and caudal bor-
dering areas, R and C, and S2 and PV (Figure 1.3; Sur et al., 1980; Wu et al., 1995,
1996, 1997). In galagos, S1 occupies a strip of koniocortex that is similar to area
3b of monkeys, but not as well differentiated. The representation of the body surface
in S1 resembles that of area 3b of monkeys, and thalamic inputs are from SA and
RA neurons of the ventroposterior nucleus. All these observations indicate that this
single S1 representation is area 3b, and not a configuration of fields or area l as
proposed by some investigators. The rostral area, R, is almost certainly homologous
with area 3a of other primates. The rostral area of galagos appears to be activated
by muscle spindles, and it does have a motor component as does area 3a of monkeys.
In addition, the rostral area receives thalamic input from the ventroposterior superior
nucleus (VPS) just dorsal to VP. The caudal area is in the same relative position as
area l, and it has some of the architectonic features of area l. In addition, it receives
projections from area 3b, as does area 1. Nevertheless, the caudal area is not as
highly responsive to cutaneous stimuli as area 1 in most monkeys. In anesthetized
galagos, neurons are either unresponsive to cutaneous stimuli or require intense
stimulation, much as the caudal area does in some non-primates (e.g., Slutsky et al.,
FIGURE 1.4 (next pg.) The somatosensory system of monkeys. The diagram shows how
afferents enter the spinal cord to rise in the dorsal column nuclei in the lower brain stem (not
to scale), or activate second order neurons in the spinal cord that contribute to the ascending
spinothalamic system. Slowly adapting and rapidly adapting mechanoreceptor inputs to grac-
ile, cuneate, and trigeminal nuclei (N) are relayed to the ventroposterior nucleus (VP) of the
thalamus. VP has two large subnuclei; a medial subnucleus (VPM) represents the face while
a lateral subnucleus (VPL) represents the rest of the body. VPL contains hand and foot
subnuclei, as well as a lateralmost tail subnucleus (not shown). Muscle-spindle receptor inputs
to the dorsal column nuclei are relayed to the ventroposterior superior nucleus (VPS), while
spinothalamic terminals relay to the ventroposterior inferior nucleus (VPI). VPI includes septa
that extend dorsally into VP. A more caudal region of the somatosensory thalamus contains
a nucleus specific for spinothalamic inputs responsive to pain and temperature (not shown).
This nucleus has been called the ventral medial posterior nucleus (VMpo). VP projects densely
to area 3b, less strongly to area 1, and sparsely to area 2 of anterior parietal cortex. VPS
projects to area 3a and area 2. VPI projects widely to many somatosensory areas, but most
densely to areas S2 and PV. Connections and recordings suggest the existence of two additional
areas, the parietal rostral area, PR, and ventral somatosensory area, VS. Posterior parietal
cortex, traditionally subdivided into areas 5 and 7, is complexly subdivided into areas with
somatosensory, motor, and visual functions. In primates with a central sulcus, some or most
of area 3b is in the central sulcus (CS).
packed with densely stained neurons. These neurons are highly active metabolically
and thus VP is also apparent as the nucleus with the most expression of cytochrome
oxidase (CO) in the somatosensory thalamus. The medial subnucleus, ventroposte-
rior medial (VPM), representing the face, is separated from the lateral portions,
ventroposterior lateral (VPL), representing the rest of the body, by a cell-sparse fiber
septum, the arcuate lamina and many early investigators considered them as separate
nuclei. A second, less obvious band separates the VPL into a more medial subnu-
cleus, representing the hand, from a more lateral subnucleus representing the foot,
The anterior parietal cortex of monkeys contains four strip-like areas that extend
mediolaterally (Figure 1.5). All four areas were originally thought to be within a
single cortical representation, S1 (Marshall et al., 1937). But, it is now clear that
each of these architectonic fields contains a separate representation of the body and
is a distinct area of the somatosensory cortex (for early evidence, see Merzenich
et al., 1978; Kaas et al., 1979). In addition, only the representation in area 3b is the
homologue of the single representation that is commonly described as S1 in other
D5
D4
D3 Palm
D2
D1
Chin+ Upper
Lower Face
Upper Lip
Lip
1.6 CONCLUSIONS
REFERENCES
Akbarian, S., Grusser, O. J., and Guldin, W. O. Thalamic connections of the vestibular cortical
fields in the squirrel monkey (Saimiri sciureus), J. Comp. Neurol., 326, 423, 1992.
Andersen, R. A., Asanuma, C., Essick, G., and Siegel, R. M. Corticocortical connections of
anatomically and physiologically defined subdivisions within the inferior parietal
lobule, J. Comp. Neurol., 296, 65, 1990.
Andersen, R. A., Snyder, L. H., Bradley, D. C., and Xing, J. Multimodal representation of
space in the posterior parietal cortex and its use in planning movements, Annu. Rev.
Neurosci., 20, 303, 1997.
CONTENTS
2.1 Introduction
2.2 Primary Somatosensory Cortex (SI)
2.2.1 Defining the Boundaries of SI
2.2.1.1 Cytoarchitecture
2.2.1.2 Evidence for Medial to Lateral Somatotopy
2.2.1.3 SI Finger Representation
2.2.1.4 Separate Anterior and Posterior Maps
2.3 Secondary Somatosensory Cortex (SII)
2.3.1 Identification of SII
2.3.1.1 Gross Anatomy of the Parietal Operculum
2.3.1.2 Cytoarchitecture
2.3.1.3 Brain Imaging Evidence on SII Organization
2.3.1.3.1 Temporal Order of Processing
through SII
2.3.1.3.2 One or Two Somatotopic Maps in SII
2.3.2 A Role for SII: Sensorimotor Effects
in Somatosensory Areas
2.4 Other Somatosensory Areas in Frontal and Parietal Cortex
2.4.1 Somatosensory Activation of Insula and Frontal Operculum
2.4.2 Posterior Parietal Cortex
2.4.2.1 Homologous Regions in Monkeys and Humans
2.4.2.2 Identification of Active Foci
2.5 Brain Regions Active during Tactile Discrimination Tasks
2.6 Tactile Attention
2.1 INTRODUCTION
Studies in monkeys indicate that somatosensory processing of innocuous tactile
stimuli occurs within an interconnected cortical network, which mostly resides in
the parietal cortex. The major regions include multiple subdivisions of the primary
somatosensory area (SI) in the postcentral gyrus, the secondary somatosensory area
(SII) in the parietal operculum, additional lateral cortical areas buried within the
Sylvian fissure, portions of the supramarginal gyrus, and granular prefrontal cortex.
This review focuses on homologous somatosensory representations in the cortex
of humans that are revealed with non-invasive recording and neuroimaging methods.
Critical to the validation of these observations is confirming characteristic features
known from direct neural recordings, especially in monkeys. Brain imaging studies
like positron emission tomography (PET) and functional magnetic resonance imag-
ing (fMRI) are especially appropriate in exposing the many cortical foci that
respond during a single somatosensory stimulation paradigm. Non-invasive tech-
niques have the great advantage of allowing replicate studies in conscious normal
or patient populations performing a variety of tasks affecting tactile discrimination,
sensorimotor integration, attention, and object recognition.
A discussion of different brain imaging techniques is beyond the scope of this
review. However, some comments about underlying characteristics of these images
Initial vetting of many brain imaging techniques derives from detecting somatosen-
sory-driven activations attributable to anterior parietal cortex, and especially when
results confirm medial to lateral sequences of activated foci from stimulating a small
number of contralateral body regions (Table 2.1). For example, PET (reviewed in
Reference 28) and fMRI (e.g., Reference 210) studies report a medial to lateral
somatotopy for widely separate body representations (e.g., foot, hand, and face).
Several studies describe activation of only a single peri-Rolandic site contralateral
to stimulating the hand with high amplitude vibratory stimuli.26,40,71,213 These obser-
vations thus corroborate earlier reports of a single distorted contralateral represen-
tation spanning the postcentral gyrus and having an approximate medial to lateral
topography with respect to successive segmental sacral to cervical to trigeminal
levels of the body.181,233 Several studies also note ipsilateral activations even in the
hand region of SI.28,90,130,138
Current imaging results must be tempered by recent findings of only 55%
concordance between the centers of activated regions using fMRI and direct record-
ings in the same brains of anesthetized monkeys.55 Thus, discrepancies from expected
SI map features may only be a consequence of inherent technical limitations.199
Further limiting possible findings is the crude spatial resolution of most imaging
studies and the limited number of body representations studied.
Some PET28 and fMRI32,72,130,138,144,164 studies reveal at least two finger foci
occupying anterior and posterior portions of the postcentral gyrus. In a PET study,
tactile stimulation of two fingertips activates contralateral foci extending into the
posterior bank of the central sulcus, without reaching the fundus, and a second
FIGURE 2.1 (previous pg.)(See Color Figure 2.1 in color insert.) Z-score maps (P<.05) of MR
signals in single subjects cued to attend to vibrotactile stimuli applied to fingertip 2 on the
right hand with an MR compatible, computer-controlled vibrator. Results obtained using a
general linear model, an assumed hemodynamic response, and a Boynton model.15 Data shown
in A are from trials when the subject selectively attended to changes in vibration frequency
to detect whether the first or second sequentially presented stimulus was a higher frequency.
Data shown in B are from another subject attending during different trials either selectively
to frequency or to the duration of paired vibratory stimuli, or who divided attention between
both frequency and duration features (see Reference 215). MR signals were collected and
analyzed per trial (TR 3.68 seconds, 3.75mm voxel size, ~16 seconds per trial). A. Functional
MR data superimposed on a three-dimensional, partially inflated view of left hemisphere for
one subject. B. Slices at three orientations illustrate the location of MR signals in anterior
and posterior portions of SI in another subject. Pink lines on each slice show registration of
slices at orthogonal orientations. Coordinate values are referenced to the Talairach atlas.222
Anatomical abbreviations (shown in red): CS, central sulcus; LS, lateral sulcus; PoCS, post-
central sulcus; SFS, superior frontal sulcus; functional abbreviations (shown in yellow): aSI,
anterior primary somatosensory area; pSI, posterior primary somatosensory area; SII, second-
ary somatosensory area; preMC, premotor cortex; iSG, inferior supramarginal region; sSG,
superior supramarginal region.
postcentral gyrus, and 68% in the vicinity of the postcentral sulcus. The anterior SI
region also shows almost no overlap for foci from non-adjacent digits (see color
coded insets in Figure 2.2A, Reference 131), while more posterior regions still
exhibit overlaps of more than 25%.131 Multiple digit representations occur within a
single horizontal plane, but with clear medial to lateral separations in presumptive
area 3b (Figure 2.2A). In addition, representations for different digits are separable
in the superior to inferior axis (Figures 2.2 and 2.3). Another fMRI study reports
considerable overlap between foci for fingers 2-3 vs. 4-5,90 but a larger representation
for fingers 23. This study, however, does not distinguish different anteroposterior
sectors of the postcentral gyrus, and vein inflow effects contaminate the images. Yet
another study describes nearly complete overlap in anterior foci from non-adjacent
digits 2 and 5 and almost no overlap in more posterior postcentral regions.72 Each
fingertip was stimulated with individually mounted piezoelectric vibrators. Even
group averaged images, obtained with higher field strength magnets and piezoelectric
vibrators, show (Figure 2.3) distinct foci separated by >3 mm in the transverse axis
for each vibrated fingertip. 144 However, three-dimensional reconstructions
(Figure 2.2B, Reference 131) show some overlap between adjacent digits in anterior
SI. The activated foci for even non-adjacent digits overlap in posterior SI
(Figure 2.2B). Despite differences in imaging techniques, the fMRI studies agree
with earlier MEG data in noting a 1218 mm tangential length to the finger
region.131,144 These studies also confirm earlier observations of larger regions devoted
to the thumb and index finger. Collectively, current brain imaging data provides a
partial confirmation of maps obtained in monkeys. However, as techniques improve,
this discrepancy will likely disappear.
A precise anatomical localization of SII within the inferior parietal lobule must
consider the sulcal arrangement surrounding the parietal operculum. On average,
the anterior and posterior borders of the parietal operculum are, respectively, the
central sulcus and posterior ascending ramus of the Sylvian fissure.218 An imaginary
lateral extension of the postcentral sulcus points to the middle of the parietal oper-
culum, which is relatively co-extensive with the horizontal ramus of the Sylvian
fissure (Figure 2.5). This horizontal segment, from inspection of 67 brains, is found
in 99% and 70%, respectively, of left and right hemispheres.232 It is twice as long
on the left. Men with a consistent right hand preference show bilateral greater length
horizontal rami; however, an asymmetry always exists, which is longer on the left.232
An inferior portion of the supramarginal gyrus lies immediately posterior and super-
ficial to the parietal operculum in most brains.218 This gross anatomical location of
the parietal operculum coincides with the position of SII in summary illustrations
by Penfield179,181 and occupies that portion of the parietal operculum included within
area 40, by Brodmann17 or area PF, of subsequent analyses (e.g., Reference 59).
Most studies of the parietal operculum with MEG, PET, or fMRI show bilateral
distributions of increased regional activity (Table 2.2). All report larger responses
contralateral to the stimulated site; however, the coordinates for responses from both
sides are identical. FMRI findings of reduced or absent ipsilateral SII responses in
patients with partial transections of the corpus callosum confirm anatomical studies
in monkeys that ipsilateral responses in SII (and posterior portions of SI) are con-
ducted by fibers crossing through the posterior third of the corpus callosum.63 Studies
in monkeys show these connections arising from SI and SII in the opposite hemi-
sphere.43,112,127,147,148 The average and standard deviation of stereotactic atlas
coordinates222 from eight studies (Table 2.2) are/ 49.3, 4.8, 21 7.3, 17.4
1.9 for the center of mass or peak activation. The small standard deviations for these
averages are notable because the data include PET and fMRI images reconstructed
to atlas space with different techniques. They demonstrate considerable consistency
in the discovered centers of contralateral and ipsilateral responses during several
types of somatosensory stimulation to different body regions. However, this spatial
uniformity presents a problem to hypothesizing a somatotopic map with distinct
zones for different body representations (see below).
2.3.1.3.1 Temporal Order of Processing through SII
The first brain imaging demonstration of bilateral somatosensory responses on the
upper bank of the Sylvian fissure in humans comes from MEG recordings.91,92 These,
and several subsequent recordings, note long latency (~100ms), mostly positive
somatosensory-evoked fields (SEF)69,70,93-95,101,139,150 or SEPs.2 Shorter latency SEPs
are reported with intracortical electrodes implanted in the upper bank of (60ms
negative cited in Reference 76) or surface cortex next to the Sylvian fissure (20 ms
cited in Reference 141) in patients. Some MEG studies121,128 emphasize short latency
(20 to 26 msec), negative responses followed by additional, larger amplitude, mostly
positive responses at the longer times noted by others.
Explanations for differences in reported response latencies require some com-
ments on modeling dipole source localizations from the upper bank of the Sylvian
monkeys illustrate large receptive fields that include face and hand, multi-finger,
hand and arm, or all of the foot, leg, and sacral/lumbar trunk.237 Potential comparable
receptive fields in humans would likely generate extensive spread of activity in fMRI
or PET images when stimulating any finger, part of the hand, or lower leg, which
might then eclipse separate representations from more selected body regions. An
Activity in SII increases more than in SI during motor tasks with a tightly coupled
sensory component that is topographically close to the body location of the contracting
muscles.101,139 For example, responses from the parietal operculum increase during trials
with combined electrical stimulation of one median nerve and synchronized voluntary
finger flexion movements only in the same hand.101 Comparable topographical tuning
effects occur in the absence of overt movements when isometric voluntary contractions
of thenar or deltoid muscles in the same limb accompany electrical stimulation of the
median nerve.69,139 SII responses decrease when median nerve stimulation is combined
with voluntary contractions of masseter or anterior tibial muscles, which are remote
from the receptive territory of the stimulated nerve.139 Additional examples of related
findings come from PET data showing increases in regional cerebral blood flows (rCBF)
in SI and SII during passive movements, but even greater rCBF changes, especially in
contralateral SII, during active vs. passive movement of finger 3.161 Similarly, bilateral
MR signals in SII, but not SI, are greater when subjects must execute multiple manip-
ulations in identifying, non-nameable objects vs. fewer movements when recognizing
simpler shapes, such as a smooth sphere.11,12
These motor related modulations of SII activity possibly indicate mechanisms
for enhancing sensory information from a limb as a guide to behavior involving that
limb. A particular example of this behavior would be active manipulation of an
object for purposes of identification. We previously hypothesized that SII provides
FIGURE 2.6 (previous pg.)(See Color Figure 2.6 in color insert.) Distribution of activated foci
in the parietal operculum from tactile stimulation of a variety of body regions in individual
subjects. Results obtained using a cross-correlation analysis. A1-3. Axial slices from three
subjects depict foci from stimulating the foot, hand, and face. Outline drawings to the right
of the MR images summarize the distribution of activity for each stimulated body region.
These cases illustrate a medial to lateral organization for, respectively, the foot to face. In
addition, the foot region occupies a more posterior position in two subjects. B1, C1. The
location of activated foci from stimulating the upper arm-shoulder with respect to the hand in
two subjects. In one subject (AO), the representation for the more proximal body region possibly
surrounds that for the foci activated from the more distal hand. B2, C2. The location of activated
foci from stimulating the upper leg-thigh with respect to the foot in two subjects. In both subjects,
the representation for the more proximal body region appears to surround that for the foci activated
from the more distal foot. [A is modified from Figure 7, and B, C are modified from Figure 8 in
Reference 54. 2000 Wiley-Liss, Inc. Reproduced with permission of John Wiley & Sons, Inc.]
Attention tasks alter activity in SI and SII. However, results from different studies
show varying magnitudes of changes in these two regions (see below). The disparities
largely rest on the question of attention effects in SI. Most neurophysiology studies
in monkeys report fewer cells influenced by attention in SI.31,100,102 A corresponding
distinction exists for the visual system for which only some recent studies confirm
attention effects in V1.78,104,190,203,223 Underlying studies of visual attention is the
hypothesis that processing through the attention network is hierarchical with the
least attention-related activity in primary sensory cortex.78,104,190,203,223 A similar con-
ception of a hierarchical somatosensory attention network may be applicable.29-32
Numerous brain imaging studies show attention related modulation of SI. For exam-
ple, tasks requiring discriminating differences between touched objects demand
attention and nearly always evoke increases relative to no stimulation/rest conditions
in regional cerebral blood flows in peri-Rolandic cortex that is contralateral to the
stimulated hand.204 Several studies report blood flow increases more confined to just
the contralateral postcentral gyrus during trials when subjects attend to vibratory
stimuli compared to performing alternative visual tasks in the presence of the same
tactile stimulation.158,176 Relative increases in rCBF also occur bilaterally in SI when
subjects selectively attend to different tactile stimulus attributes as compared to trials
with the same stimuli, but a distracting counting task.30 Blood flow changes also
arise in the ipsilateral SI in the vicinity of the postcentral sulcus, but more promi-
nently when ipsilateral is the right hemisphere. In fMRI studies, MR signal intensity
increases on trials cued to detecting touches of the big toe compared to trials with
attention to a visual display,111 or the spatial extent of activations increases when
contrasting responses during trials with a tactile vs. a distracting counting task.88
Thus, activity in SI increases for attended tactile information. Unfortunately, there
are currently no distinctions regarding attention effects in anterior vs. posterior parts
of SI.
Brain imaging studies also document a role for SII in tactile attention. Examples
from MEG recordings include greater magnitude SEF responses at 55 to 130 ms
from SII when using an oddball paradigm that requires vigilance to detect variations
in different stimulus attributes: spatial location,94,151 stimulus intensity,160 or both
location and intensity.99 Neither short (~20 ms) nor long (>35 ms) latency responses
from SI show enhancements in the same tasks.99 In contrast, several PET studies
report rCBF increases in SI, but greater shifts in SII. This includes finding rCBF
increases when subjects selectively attend to the roughness or length attributes of
gratings passively rubbed against the skin30 or when attending to roughness or the
length of manipulated objects.133 The distinctions regarding relative modulation in
SI compared to SII also entails significantly larger spatial extent of MR signal
Two sites within frontal cortex routinely show activity changes during a variety
of tasks including those involving attention. One is in area 6 of premotor cortex
and mostly occupies the depths of medial precentral sulcus (Figure 2.1, preMC).
A second is in the anterior cingulate cortex. Both regions show increased rCBF
when subjects attend to selected attributes of tactile30 or visual44-46 stimuli. Motor
tasks engage corresponding parts of premotor cortex, which as noted below, has
prompted suggestions about common mechanisms between attention and intention
to move.
Models of attention often incorporate the notion of amplified neural responses for
selected, relevant stimuli, which improves probable preferential detection of desired
target events.52,187,188 In single neuron recording experiments, the most common
expression of gain is enhanced firing rates to task relevant stimuli (reviewed in
Reference 32). A potential expression of this process in brain imaging findings is
the many examples of magnified responses in somatosensory cortical areas, and
especially in SII, whenever attention focuses on some body part. Such amplifications
might reflect a priming mechanism for any input applied to the receptive field of a
neuron29,99 or, more generally, to the collective responses of neurons representing an
attended body region.80
Somewhat unique to brain imaging studies is the revelation of potential global shifts
in relative activity depending on the focus of attention. For example, attending to
tasks that involve non-tactile modalities downwardly modulates blood flows in
SI.65,97,124,142 A similar decrease in blood flow arises in the somatosensory cortical
foci for one body representation whenever attention is directed to a distant, different
part of the body.56 Evidence from fMRI studies in the visual system also shows that
Several brain imaging studies support the notion of a reorganized SI in adult subjects
following peripheral or central injuries, sustained somatosensory stimulation through
training, habituation to repetitive, simultaneous stimulation, or local anesthesia pro-
tocols. Much of this literature relies on observations from EEG or MEG recordings
that are projected to be from the postcentral gyrus. However supportive these
results are for findings of plasticity in animals, the data is sometimes contradictory
and possibly open to alternative explanations. Brain imaging data by itself also
cannot identify the locus responsible for observed changes. Instances of immediate,
behaviorally mediated changes are likely due to unmasking existing subthreshold
cortical connections through mechanisms, such as disinhibition.117 However,
patients with long-term amputations may show reorganizations based on more
extensive anatomic changes, such as the axonal sprouting that has been observed
in animals.67,108,118 Changes in the connections between neighboring somatotopic
representations in brainstem structures may be especially important (reviewed in
Reference 165).
2002 CRC Press LLC
2.7.2.1 Effects Following Injuries
2.8 SUMMARY
Brain imaging techniques offer a unique global overview of brain activity. Many
experiments involving the somatosensory system have focused principally on the
organization of the body representation in SI. Currently, the SI maps examined from
brain images are relatively crude replicas of the more detailed representations studied
previously with more direct recordings in animals. The explanations for differences
between these two sets of observations are not yet fully known. Obvious are technical
considerations related to spatial and temporal resolution. But functional distinctions
not previously evident in the more focused animal studies may alter ideas about the
likely rigidity of map organization. Recent studies in humans have uncovered
remarkable dynamics in SI organization by manipulating tactile attention or through
demonstrating alterations after training or injuries. Both attention and training effects
may be based on manipulating synchronicity of inputs. These findings point toward
a conception of SI maps as a neural resource within which the total pool of engaged
neurons changes to meet behaviorally meaningful demands. In actuality, there may
be no single SI map as originally conceived and refined by neurophysiology studies
through much of the 20th Century. Missing still from most brain imaging data on
SI is consistent evidence reflecting its four cytoarchitectonic subdivisions known
for nearly 100 years. It will be difficult to extend theoretical predictions about the
role of SI in tactile perceptions without such information.
Brain imaging has also provided substantial evidence that multiple cortical areas
contribute to somatosensory perceptions. These have included the parietal operculum,
posterior parietal cortex, frontal operculum, insula, and possibly portions of precentral
frontal cortex. There are still other activated regions not discussed that include cingulate
cortex and medial premotor regions. Intriguing has been converging data about over-
lapping functional domains for attention and intention for movements within some of
these same regions. Although enticing, much of this literature is still preliminary and
merely identifies the existence of several regions. It has yet to clarify specific roles for
any region. Brain imaging studies, when coupled with well-designed behavioral proto-
cols, will be particularly apt in helping define the spatial and temporal sequences of
activation. This data will be needed in forming new hypotheses. One example already
accessible is the relative interdependence between SI and SII or between anterior and
posterior parts of the postcentral gyrus in tactile discrimination tasks.
ACKNOWLEDGMENTS
I thank the authors of several papers who generously sent me copies of their figures.
My colleague, Dr. Robert Sinclair, provided especially valuable edits and comments.
Work on this review was supported by funds from the National Institutes of Health,
NS31005.
CONTENTS
3.1 Overview
3.2 Peripheral Neural Mechanisms
3.2.1 Mechanoreception
3.2.2 Form and Texture Perception
3.2.2.1 Form Perception
3.2.2.1.1 Spatial Acuity
3.2.2.1.2 Pattern Recognition
3.2.2.1.3 Studies with the Optacon
3.2.2.1.4 Curvature Perception
3.2.2.2 Texture Perception
3.2.2.2.1 Roughness
3.2.2.2.2 Softness
3.2.2.3 Texture Perception with a Probe
3.3 Cortical Mechanisms
3.3.1 Function of Area 3b
3.3.2 Receptive Fields in Area 3b
3.3.3 Delayed Inhibition
3.3.4 Functional Implications
3.4 Conclusion
References
3.1 OVERVIEW
Information about the external world is analyzed and subdivided into separate pro-
cessing streams in each of the sensory systems. That division begins at the very first
stage of sensory processing within the somatosensory system: nociceptors, thermore-
ceptors, proprioceptors, and cutaneous mechanoreceptors transduce different stimulus
properties and channel their information into separate, parallel streams. The same
principle applies to the four cutaneous mechanoreceptors that are responsible for tactile
is asked whether the dot or ridge spacing of one surface is greater than another the
response is a judgment about the surface and it provides a clue to the subjects
capacity for form perception (e.g., Reference 33 and 34); if the same subject is asked
whether the second surface feels rougher (or softer) than the first, the response is a
description of his or her experience and it provides a clue to the subjects perception
of texture.
The ability to discriminate object or surface features and the capacity for pattern
recognition at the fingertip are the same whether the object is contacted by active
touch or is applied to the passive hand.35 Form perception is affected only marginally
by whether the object is stationary or moving relative to the skin; it is unaffected
by scanning speed up to 40 mm/s; it is unaffected by contact force, at least over the
range from 0.21 N; and it is affected only marginally by the heights (relief) of
spatial features over a wide range of heights.36-39
Three psychophysical studies of the limit of tactile spatial acuity are illustrated
in Figure 3.1. In all three studies, the element width that resulted in performance
midway between chance and perfect discrimination was between 0.9 and 1.0 mm,
which is close to the theoretical limit set by the density of SA1 and RA primary
afferents at the fingertip.40 Acuity declines progressively from the index to the fifth
finger41 and it declines progressively with age.42-44 Whether these differences in
acuity are due to differences in innervation density is not known. Spatial acuity at
the fingertip is the same in man and monkey.45 Spatial acuity at the lip and tongue
is significantly better than at the fingertip.46-48
3.2.2.1.1 Spatial Acuity
Tactile spatial resolution of about 1 mm requires an innervation density of (at least)
about one afferent per square mm and it requires that individual afferents resolve
the spatial details at least as well as human subjects. Neither the PC nor the SA2
system comes close on either score.13,49 Note that the human performance illustrated
in Figure 3.1 begins to rise above chance at element sizes around 0.5 mm, which
means that either the SA1 or RA system must begin to resolve spatial detail at
0.5 mm or less. Evidence that only the SA1 afferents account for the spatial reso-
lution illustrated in Figure 3.1 comes from neurophysiological experiments in which
SA1 and RA afferents were studied with the same periodic gratings used in the
psychophysical experiments. SA1 responses to a periodic grating convey information
about spatial structure when the groove and ridge widths are 0.5 mm wide (e.g., see
Figure 3.2). When the grooves and ridges are 1 mm wide, SA1s provide a robust
neural image of the stimulus. In contrast, RAs require grooves that are at least 3 mm
wide before their responses begin to distinguish a grating from a flat surface; most
RAs fail even to register grooves 3 mm wide.7 The RA response illustrated in
Figure 3.2 was the most sensitive to spatial detail of all the RAs studied. Kops and
Gardner50 obtained nearly identical results with an Optacon, a dense array of vibro-
tactile probes designed as a reading aid for the blind.51 PC afferents were unable to
resolve grooves that were 5 mm wide (Figure 3.2).
well, that it fails to activate the SA1 system, and that the PC system cannot account
for spatial pattern recognition performance with the Optacon;52,53 therefore, the many
psychophysical studies employing the Optacon54 are studies of the sensory capacity
provided by the RA system. The human ability to resolve spatial patterns with the
Our knowledge of texture perception and its neural mechanisms has changed dra-
matically in the last decade. A major step is the demonstration that texture perception
involves two strong dimensions, roughness and softness, and a weaker third dimen-
sion described as something like stickiness. Multidimensional scaling studies have
shown that texture perception includes softhard and smoothrough as independent
perceptual dimensions, surface hardness and roughness can occur in almost any
combination, and that they account for most or all of texture perception.64,65 A third,
FIGURE 3.2 (previous pg.)Responses of SA1, RA, and PC afferents to a grating pressed into
the skin. The grating is shown in cross section beneath each response profile. The bars are
0.5 mm wide; the grooves are deeper than illustrated (2.0 mm deep) and are 0.5, 0.5, 0.75,
1.0, 1.5, 2.0, 3.0, and 5.0 mm wide. The responses displayed in each profile were obtained
by indenting the skin to a depth of 1 mm, holding the indentation for 1 second, raising the
grating, and then moving it laterally by 0.2 mm before the next indentation. The horizontal
dimension of the response profile represents the location of the center of the receptive field
relative to the grating; for example, the left peak in the SA1 response profile (approximately
95 imp/s) occurred when the center of the SA1 receptive field was directly beneath the left
edge of the grating. The RA illustrated here was the most sensitive to the spatial structure of
the grating of all RAs studied. Some RAs barely registered the presence of the 5 mm gap
even though they responded vigorously at all grating positions. Adapted from Phillips, J.R.
and Johnson, K.O., J. Neurophysiol. 46, 11921203, 1981, with permission.
FIGURE 3.5 SA1 population response to indentation with spheres of varying curvature. The
left plot shows the mean responses of SA1 afferents as a function of proximaldistal distance
from the center of indentation. Data is shown for seven curved surfaces with radii ranging
from 1.44 mm (curvature = 694 m-1) to a flat surface (curvature = 0 m-1). The right plot shows
population response profiles in proximaldistal slices at varying distances from the center of
indentation. From Goodwin, A.W., Browning, A.S., and Wheat, H.E., J. Neurosci. 15,
798810, 1995, with permission.
FIGURE 3.6 (next pg.) Perceived roughness and spatial variation in SA1 firing rates in four
studies with different textured surfaces. The left ordinate, filled circles, and solid lines in each
graph represents the mean reported roughness for each surface (first normalized to a grand
mean of 1.0 for each subject). The right ordinate, open circles, and dashed lines represent a
measure of spatial variation in SA1 firing rates. The same measure of spatial variation was
used for all four studies. The surface pattern used in each study is illustrated below the data
to which it applies. The top row (A) shows results from Connor et al.,71 who used 18 raised
dot patterns with different dot spacings and diameters. The middle row (B) shows results
from Blake et al.,73 who used 18 raised dot patterns with different dot heights and diameters.
The two left graphs in the bottom row (C) show results from Connor and Johnson,72 who
varied pattern geometry to distinguish temporal and spatial neural coding mechanisms. The
right graph shows data from Yoshioka et al.,75 who used fine gratings with spatial periods
ranging from 0.1 to 2.0 mm. The lines connect stimulus patterns with constant spatial periods.
From Yoshioka, T., Dorsch, A.K., Hsiao, S.S., and Johnson, K.O. J. Neuroscience, in press,
2001, with permission.
the studies described above. The mean firing rate of a population of such neurons
with excitatory and inhibitory receptive fields separated by 23 mm would corre-
spond closely to a subjects roughness judgments.75 Such a mechanism has several
advantages. Like roughness perception, it is unidimensional and it is affected only
secondarily by factors such as scanning velocity and contact force.78 Whether or not
neurons in area 3b are responsible for the computations that underlie roughness
When we use a tool or probe, we perceive distant events almost as if our fingers
were present at the working surfaces of the tool or probe. An early demonstration
of this was by Katz,83 who showed that we can discriminate the texture of a surface
as well with a probe as with a finger applied directly to the surface. He showed
further that this capacity is lost when vibrations in the probe are damped. A recent
study21 has shown that when subjects grasp a probe, some subjects can detect
transmitted vibrations with amplitudes less than 10 nm (the mean is 30 nm,
Figure 3.7). Only the PC system can account for this capacity.
The hypothesis that the PC system is responsible for the perception of vibrations
transmitted through an object held in the hand supposes that Pacinian receptors
detect the transmitted vibration and that the PC population transmits a neural rep-
resentation of the vibratory signal sufficient to account for this perceptual capacity.
Work on the human ability to detect and discriminate complex vibratory stimuli
supports this idea by showing that we are sensitive to the temporal structure of high
frequency stimuli that only activate PC afferents.84-86 For example, humans can
discriminate the frequency with which a 250-Hz carrier stimulus is modulated for
modulation frequencies as high as 60 Hz.86 In contrast, Bensmaa and Hollins87 have
shown that the discrimination of complex waveforms composed of high frequencies
is poor and they have suggested that RA afferents play a significant role in temporal
coding.
The receptive fields of 247 area 3b neurons mapped with scanned, random-dot
stimuli are illustrated in Figure 3.9, which shows that nearly all receptive fields are
characterized by a single central region of excitation with inhibition on one, two, or
three sides. Surround inhibition occurred rarely. The inhibitory area was, on average,
about 30% larger than the excitatory area (means were 18 and 14 mm2) and, like
the excitatory area, varied greatly (from 1 to 47 mm2). The inhibitory mass (absolute
value of inhibition integrated over the entire inhibitory field) like the excitatory mass
(comparable definition), varied by 50 to 1 between neurons (125 to 6830 mass units;
mean 1620 mass units). There was no evidence of clustering into distinct receptive
FIGURE 3.9 (opposite) Receptive fields in area 3b of the alert monkey. Each panel illustrates
an example of a receptive field type, the total number of receptive fields of that type, and
their percentage of the total receptive field sample (n = 247, all on distal fingerpads). The
gray scale represents the grid of excitation and inhibition (25 25 bins = 10 10 mm) that
best described the neurons response to a random stimulus pattern. Dark regions represent
excitatory regions; lighter regions represent inhibitory regions. The uniform background gray
level represents the region where stimuli had no effect. The types are shown in decreasing
order of frequency: A. A single inhibitory region located on the trailing (distal) side of the
excitatory region (left). B. A region of inhibition located on one of the three nontrailing sides
of the excitatory region. C. Two regions of inhibition on opposite sides of the excitatory
region. D. Inhibition on three sides of the excitatory region. E. Inhibition on two contiguous
sides of the excitatory region. F. A complete inhibitory surround. G. An excitatory region
only. H. Receptive field dominated by inhibition. I. Receptive fields not easily assigned to
one of the preceding categories. This categorization is meant only to show the range and
relative frequency of these receptive field types. In fact, the receptive fields formed a contin-
uum spanning these shapes. From DiCarlo, J.J., Johnson, K.O., and Hsiao, S.S., J. Neurosci.
18, 26262645, 1998, with permission.
3.4 CONCLUSION
In this chapter, we have reviewed the peripheral and cortical neural mechanisms of
tactile form and texture perception. Three decades of combined psychophysical and
neurophysiological experiments suggest a sharp division of function among the four
cutaneous afferent systems that innervate the human hand. The SA1 system provides
a high-quality neural image of the spatial structure of objects and surfaces that
contact the skin and this is the basis of form and texture perception. The RA system
provides a neural image of motion signals from the whole hand. From this, the brain
extracts information that is critical for grip control and also information about the
motion of objects contacting the skin. The PC system provides a neural image of
vibrations transmitted to the hand from objects contacting the hand or, more fre-
quently, objects grasped in the hand. The SA2 system provides a neural image of
skin stretch over the whole hand. The evidence is less secure but the most likely
hypothesis is that the brain extracts information about hand conformation and the
direction of motion of objects moving across the skin. S1 cortex transforms this
information to an altered form of representation that is, as yet, not well understood.
The available evidence suggests that area 3b is specialized for processing spatial
information. Evidence of that is its magnification of body parts with high spatial
resolution compared with area 1 and its neurons response properties, which appear
to be specialized for processing spatial information.
REFERENCES
1. Johnson, K.O., Yoshioka, T., and Vega-Bermudez, F., Tactile functions of mechan-
oreceptive afferents innervating the hand. J. Clin. Neurophysiol. 17, 539, 2000.
2. Johnson, K.O. and Hsiao, S.S., Neural mechanisms of tactual form and texture
perception. Annu. Rev. Neurosci. 15, 227, 1992.
3. Iggo, A. and Andres, K.H., Morphology of cutaneous receptors. Annu. Rev. Neurosci.
5, 1, 1982.
CONTENTS
4.1 Introduction
4.2 Research Background and Methodologies:
Perceiving with the Bare Finger
4.2.1 Roughness Perception
4.2.1.1 Psychophysics
4.2.1.2 Neurophysiology
4.2.2 Common-Object Identification
4.3 Perceiving Surface Textures via Remote Touch
4.3.1 Effects of Surface Geometry: Interelement Spacing
4.3.2 Effects of Geometric Characteristics of the Probe:
Probe-Tip Diameter
4.3.3 Effects of the Manual Exploration Process
4.3.3.1 Relative Speed of Motion
4.3.3.2 Mode of Touch
4.3.3.3 Force
4.4 Identifying Common Objects Haptically via Remote Touch
4.5 Remote Touch in Medicine
Acknowledgment
References
4.1 INTRODUCTION
Experimental psychology offers a perspective on the workings of the somatosensory
system that is complementary to various wet-lab approaches considered by most
chapters in this book. The latter address the neural mechanisms and neural repre-
sentations that underlie how the somatosensory system processes information about
the external world and about our bodies. Such approaches examine the relationship
between the environment (external, internal) and the underlying neural responses.
Experimental psychologists study how the intact organism processes those same
environmental events in terms of the behavioral responses of the whole organism.
Ultimately, our joint goal is to understand how organisms sense, perceive, think,
represent, and act on the environment.
4.2.1.1 Psychophysics
4.2.1.2 Neurophysiology
Johnson and Connor7,16,17 have argued for a spatial-intensive neural code for
representing the neural events that underlie the perception of roughness magnitude.
Their work suggests a spatial code at the peripheral level in terms of the relative
activity across spatially distributed Slowly Adapting Type I (SAI) mechanoreceptors.
This spatial code is maintained in SI cortex (areas 3b and 1), using the neural
differences in activity of adjacent SAI units separated by about 1 mm. The neural
difference operations must be performed by cortical neurons whose receptive fields
possess neighboring excitatory and inhibitory areas. The researchers further suggest
that the spatially distributed neural differences are passed on to neurons in area SII,
which integrate the information received from areas 3b and 1 into a single intensive
code.
Most recently, Dorsch, Yoshioka, Hsiao, and Johnson18 have provided additional
evidence that suggests that roughness perception below the limit of cutaneous spatial
resolution (i.e., for gratings with interelement spacings as low as .1 mm) is also
mediated by spatial variation in the SAI population response. These data converge
nicely with the psychophysical findings reported by Hollins and Risner.15
y = ax2 + bx + c (4.1)
where a is the coefficient for the quadratic term, b is the coefficient for the linear
component, and c is a constant that raises or lowers the function. Alternatively, it is
possible to express the same function as follows:
where a now represents the curvature of the quadratic component (i.e., the sharpness
with which the function peaks); x0 is the location of the peak along the x-axis, and
h is the height of the peak on the y-axis. In terms of the coefficients from Equation 1,
x0 is equal to b/2a and h is equal to c (b2/4a).
Using Equation 2, it is now possible to evaluate the effects of candidate physical
variables on perceived roughness more systematically in terms of the curvature of
the quadratic component (a), peak position on the x axis (x0), and peak height on
the y-axis (h), which are all meaningful features. We plan to present a comprehensive
analysis of the three quadratic parameters with respect to the results of our probe
experiments in a future paper. Here, we will report only the most consistent findings.
To date, we have noted a consistent effect of probe-tip diameter on perceived
roughness: the peak of the quadratic function shifted toward the wider end of the
interelement-spacing continuum as probe diameter increased. Presumably, this
occurred as the probe switched from riding along the tops of the raised elements to
penetrating the interelement spaces. A wider probe tip would penetrate only the
wider spacings, leading to an increasingly smooth percept as the probe rode along
the base substrate. Another result of the displacement of the peak along the x-axis
was a concomitant reversal in the roughness estimates that occurred toward the
higher end of the spacing continuum. More specifically, surfaces with interelement
spacings below the peak value were perceived to be rougher with the small probe
than with the large probe; the order was reversed for interelement spacings beyond
the peak value such surfaces felt rougher with the large probe than with the small
probe. The reversal in roughness may have been due to the greater mechanical impact
that occurred when the narrower probe penetrated the interelement spaces, subse-
quently catching on the leading edge of the raised elements.
A subsequent study by Lederman, Klatzky, Hamilton, and Grindley [Reference
25, Expt. 2] has recently confirmed our earlier results, this time using identical
spherical shapes for the probe tip and diameters of 1, 2, and 3 mm.
Lederman, Klatzky, Hamilton, and Ramsay3 examined the consequences for rough-
ness perception of changing the relative speed of motion between the textured
surfaces and the stick probe used to explore them. Both active and passive modes
of touch were employed. Active touch required that the subject move the probe over
a stationary surface, whereas passive touch required that an external agent move the
surface under a stationary probe.
In one experiment, speed was altered by a factor of 10: 20.5 (slow), 73.2
(medium), and 207.3 (fast) mm/s. Like the probe studies just mentioned, the
psychophysical function for each speed was best fit by a quadratic (as opposed to
a linear) equation: perceived roughness initially increased with increases in inter-
element spacing, then progressively decreased with further increases in interelement
spacing. Using Equation 2 to convert to the alternate form of the second-order
polynomial, we determined that the peak, x0, of the quadratic function shifted
progressively further toward the wide end of the interelement-spacing continuum
with increasing speed. An associated reversal in perceived roughness magnitude was
also observed: roughness magnitude diminished with increasing speed for the nar-
rower interelement spacings (i.e., over the monotonically increasing portion of the
function); beyond the peak value, the effect of increasing speed was reversed (i.e.,
perceived roughness increased with decreasing speed for the progressively wider
interelement spacings). Increasing speed influenced perceived roughness in ways
that were similar to those due to our increasing probe size. The results are shown
in Figure 4.4. A second experiment that used only a four-fold change in speed
produced similar results in terms of the maximum speed effects obtained.*
We attribute not only the quadratic shape of the function, but also the peak shifts
and the reversals in roughness magnitude due to speed and probe size to mechanical
changes in the proximal vibratory signal: for the relatively narrow spacings, each
probe will ride along the tops of the raised elements. At some spacing, the probe
will begin to penetrate the gap between the raised elements, ride along the smooth
base, and catch on the leading edges of the raised elements encountered in its
* Reconsideration of the mathematical validity of the doubling ratio, a statistic originally devised to
compare the relative magnitude of various factors to that of interelement spacing, led us to conclude
subsequently that the range of speeds in fact had no effect on perceived roughness estimates.
movement path. The value of the interelement spacing at which this change occurs
will likely be determined by both speed of motion and probe size as well by
element height, which we are currently investigating.
In the experiment above, subjects used either active or passive modes of touch to
explore the textured surfaces. The maximum effect of speed on perceived roughness
was statistically significant for both modes of touch (see Figure 4.4). These results
indicate that subjects failed to show complete roughness constancy; that is, they did
not alter their roughness estimates to fully compensate for the effects of changing
speed on the proximal cutaneous and kinesthetic inputs. However, the maximum
effect of speed was greater for passive than for active touch. This result suggests
greater compensation in the active condition, as expected because of the additional
kinesthetic cues to velocity.
4.3.3.3 Force
Lederman et al. [Reference 25, Expt. 1] also examined the effects of varying the
force applied to the surface while exploring it with a probe. A passive-touch mode
was selected, as this provided better control over the constant speed. Once again,
subjects provided numeric estimates of the roughness magnitude of a set of raised-
dot surfaces. All functions were well fit by quadratic equations. As shown earlier
with the bare finger, increasing the applied force increased subjects estimates of
Blindfolded subjects explored objects with either the small or large probe,
choosing to hold it much as they would a pencil. Each subject was fitted with a
fabric knit glove. A total of 2 cm was cut away from each fingertip, thereby restricting
contact information to the skin areas in contact with the probe. Objects were placed
before the observer in a prototypical orientation. The mean accuracy and mean
response times from initial contact to naming are shown in Table 4.1 for the small
and large probes.
The observed performance values are considerably poorer than corresponding
levels in the Klatzky et al. study (cf. bare index finger: mean accuracy = 84%, mean
response time = 22.8 s; gloved index finger: mean accuracy = 74% and mean
response time = 45.2 s). T-tests indicated that the differences between the two probes
did not even approach significance. Relatively poor performance scores with a rigid
sheath, relative to unconstrained exploration with multiple bare fingers, were also
obtained in this study of object identification via intermediary links. A more com-
prehensive paper that presents details of the complete set of experiments is now
under preparation.
REFERENCES
1. Lederman, S.J. and Klatzky, R.L., The hand as a perceptual system. In: The Psycho-
biology of the Hand, Connolly, K., Ed., London: MacKeith Press, Chapter 2, 16, 1998.
2. Klatzky, R.L. and Lederman, S.J., The haptic glance: A route to rapid object identi-
fication and manipulation. In: Attention and Performance XVII: Cognitive Regulation
of Performance: Interaction of Theory and Application, Gopher, E. and Koriat, A.,
Eds., Mahwah, NJ: Erlbaum, Chapter 6, 165, 1999.
3. Lederman, S.J., Klatzky, R.L., Hamilton, C.L., and Ramsay, G.I., Perceiving rough-
ness via a rigid probe: Psychophysical effects of exploration speed and mode of touch,
Haptics-e (Electronic Journal of Haptics Research), 1(1), 1, 1999. Available at
http://www.haptics-e.org.
4. Klatzky, R.L. and Lederman, S.J., Tactile roughness perception with a rigid link
interposed between skin and surface, Perception and Psychophysics, 61(4), 591, 1999.
5. Lederman, S.J., Tactile roughness of grooved surfaces: The touching process and
effects of macro- and micro-surface structure, Perception and Psychophysics, 16(2),
385, 1974.
6. Stevens, S.S. and Harris, J.R., The scaling of subjective roughness and smoothness,
Journal of Experimental Psychology: General, 64(5), 489, 1962.
7. Connor, C.E., Hsiao, S.S., Phillips, J.R., and Johnson, K.O., Tactile roughness: Neural
codes that account for psychophysical magnitude estimates. Journal of Neuroscience,
10, 3823, 1990.
8. Lederman, S.J. Tactual roughness perception: Spatial and temporal determinants,
Canadian Journal of Psychology, 37, 498, 1983.
9. Lederman, S.J. and Taylor, M.M., Fingertip force, surface geometry, and the percep-
tion of roughness by active touch, Perception and Psychophysics, 12(5), 401, 1972.
10. Taylor, M.M. and Lederman, S.J., Tactile roughness of grooved surfaces: A model
and the effect of friction, Perception and Psychophysics, 17(1), 23, 1975.
11. Katz, D., The World of Touch, L. Krueger, Trans., Hillsdale, NJ: Erlbaum, 1989
(original work published 1925).
12. Johnson, K.O. and Lamb, G.D., Neural mechanisms of spatial tactile discrimination:
Neural patterns evoked by braille-like dot patterns in the monkey, Journal of Physi-
ology, 310, 117, 1981.
13. Lederman, S.J., Loomis, J.M., and Williams, D., The role of vibration in the tactual
perception of roughness, Perception and Psychophysics, 32(2), 109, 1982.
14. Hollins, M., Bensmaia, S.J., and Risner, S.R., The duplex theory of tactile texture
perception, Proceedings of the Fourteenth Annual Meeting of the International Society
for Psychophysics, 115, 1998.
15. Hollins, M. and Risner, S.R., Evidence for the duplex theory of tactile texture per-
ception, Perception and Psychophysics, 62(4), 695, 2000.
16. Connor, C.E. and Johnson, K.O., Neural coding of tactile texture: Comparison of
spatial and temporal mechanisms for roughness perception, Journal of Neuroscience,
12, 3414, 1992.
CONTENTS
We take contact of our body with the ground for granted. Except for brief moments
during running or jumping, or falling from a tree or ladder, those of us who are not
skydivers are rarely without physical contact between our body and the support
surface of the earth. The acceleration of earth gravity pulls us downward and the
reaction force provided by the surface of the earth or the floors of structures arrests
our downward motion and provides a physical base on which stance and locomotion
can be carried out. What a scale measures as weight is actually the contact force it
provides in supporting the body against the pull of gravity. In the absence of an
effective gravitational pull on the body, walking and running would not be possible
independently articulated head and neck, the vestibular system could still be used
as a primary source of information about head orientation and, coupled with head-
on-trunk information, be used to determine body orientation to gravity. Touch and
pressure cues on the body surface, however, provide additional information directly
representing the direction of down and the location of the body center of mass
relative to the parts of the body supporting it against the downward acceleration of
gravity. These body contact cues are very important in the control of balance and
locomotion. Patients with dorsal column lesions or degeneration (e.g., tabes dorsalis,
Friedrichs ataxia) are unstable and adopt a broad-based shuffling gait. Diabetic
patients with peripheral neuropathies show similar impairments of function.
* Occasionally, a subject will feel as if he/she is undergoing orbital motion in a horizontal plane while
always face-up or face-down.
subjects feel upright while describing their orbital path (see Figure 5.3C), and pres-
sure on the top of the head makes them feel upside down. Thus, the overall pattern
of touch and pressure cues impinging on the body dramatically influences experi-
enced orientation. The transition from horizontal to vertical orbital motion is not
experienced as a physical rotation from horizontal to vertical, but rather as a gradual
feeling of being less and less in one orientation and then progressively more com-
pelling in the new orientation, until it is completely vivid. The absence of a sense
of spatial displacement from horizontal to vertical is likely related to the absence
of the semicircular canal and otolith activity that would accompany such a change.
A similar fading-out, fading-in of experienced orientation occurs in the orientation
illusions experienced in parabolic flight described above.
The ZARR studies emphasize the importance of touch and pressure cues for
orientation and indicate that dynamic patterns of somatosensory stimulation can
supplement and even supplant vestibular contributions to perceived orientation. This
is especially obvious in parabolic flight experiments involving the ZARR where
background force level varies between 0 and 1.8 g. When blindfolded subjects ride
as per-rotation. Lifting and lowering the feet will cause the aftereffect to alternately
change direction and to dissipate faster than the usual 2030 seconds. With stops
from higher velocities of rotation, the duration of the aftereffect is increased. It is
important to note that when the subjects feet remain on the footrest of the rotating
chair, he experiences his feet as stationary in relation to the footrest and himself and
the entire chair as turning. On lowering his feet to touch the floor, he experiences
it to be stationary and his body and the chair to be turning in relation to it. This
pattern means that contact of the feet with different reference frames chair vs.
floor greatly influences the apparent motion of the body. Such an influence points
to top-down cognitive effects on orientation rather than a bottom-up integration
of sensory signals.
Similar top-down influences can be demonstrated for hand and arm movement
control. When a subject, in darkness, is accelerated in a rotating chair nystagmus is
elicited. The nystagmus has a slow phase that is opposite the direction of rotation.
If a target light is attached to a boom on the chair and fixed relative to the observer,
during angular acceleration, the target will be seen to change its position relative to
the subject. The subject will see it displace in the direction of acceleration. After
reaching a peak subject-relative displacement, the target will still seem to be moving
relative to the observer but no longer increasing its subject-relative displacement.
reflex is resisted, apparent motion of that limb will be experienced in the direction
that would be associated with lengthening of the vibrated muscle. Thus, biceps
brachii activation leads to apparent forearm extension and triceps brachii activation
to apparent flexion. The arm movement illusion has two components: a spatial
displacement and a velocity one. The limb will seem to move physically to a new
spatial location and then to still be moving but no longer displacing. The induced
motion of a visual target attached to the hand of the vibrated arm has the same
features and shows a peaking of apparent spatial displacement while still being
perceived to be in continuous motion.
If a target light is attached to the index finger of each hand and the fingers are
separated by a few centimeters, as illustrated in Figure 5.5, then when illusory
extension of both arms is induced by vibration of both bicep muscles, the targets
will be seen to increase in separation. Bilateral triceps vibration induces apparent
flexion of the two arms and the targets will be seen to move closer together. This
visual direction remapping occurs for both monocular and binocular viewing and
eye position recordings indicate that steady fixation is maintained.18 The illusory
visual effect is eliminated if the fingers are not in direct contact with the targets.
These results demonstrate that proprioceptive-tactile information about visual target
location can influence the apparent direction of gaze, visual direction, and the
apparent physical separation of two targets (the latter, without affecting apparent
distance, which means that there must be a central remapping of retinal local signs).
These findings indicate that somatosensory and proprioceptive information provides,
through hand contact, a mechanism for calibrating the direction of gaze and visual
spatial localization.
FIGURE 5.8 Perceptual remapping of body dimensions in the vicinity of contact with an
appendage in which a vibratory myesthetic illusion is induced. In the test situation, the subject
holds his arms akimbo in darkness (center) during bilateral biceps (left) or triceps (right)
brachii muscle vibration. Biceps vibration elicits illusory forearm extension and body expan-
sion at the point of hand contact. Triceps vibration induces illusory forearm flexion and
pinching of the waist.
measured the threshold for detecting the motion of a surface moving under the
fingertips. He found that the stationary, restrained fingertips resting on a surface
could detect motion of 1 mm/s or even less. He also had subjects move their hands
back and forth laterally with their fingertips contacting a yardstick that also could
be in motion. He found that the direction and approximate speed of yardstick motion
could be detected although he did not provide details about the frequencies, veloc-
ities, and amplitudes of arm and yardstick movements studied. The apparatus illus-
trated in Figure 5.9 permits related questions to be asked. It couples motion of the
contact surface with motion of the hand in the same or opposite direction by a
fraction thereof. Consequently, the tactile feedback associated with voluntary move-
ments of the hand over a surface can be systematically remapped. This allows
measurement of the detection thresholds for identifying motion of the surface during
conjoint hand movements.
Subjects are better able to detect surface motion when it is in the same direction
as arm motion. The displacement detection threshold is about 7% for with dis-
placement of the surface but 40% for against displacements. Thus, subjects are
much more able to detect displacements of the contact surface during voluntary
motion of the hand when the surface moves so as to reduce the magnitude of slip
displacement across the fingertips. Surface displacement against the direction of
hand motion is considerably more difficult to detect reliably. An interesting addi-
tional feature is that subjects also tend to overestimate the magnitude of their arm
movements in experimental conditions involving counter-displacements of the con-
tact surface and to underestimate them when the surface moves in the same direction
as their arm movements. This pattern suggests that the slip cues at the fingertips as
the hand sweeps over a surface contribute to the appreciation of the magnitude of
the arm movement.
A contribution of tactile input at the fingertip to the appreciation of arm position
is consistent with recent observations demonstrating that when pointing movements
are made to targets on a surface, the pattern of shear force stimulation of the fingertip
provides a spatial coding of hand position with respect to the body.1 The region of
the fingertip stimulated also indicates far vs. near and left vs. right. The shear reaction
force vectors generated on the fingertip during the first 30 msec after impact are
hand position and actual hand position. Thus, full endpoint adaptation cannot be
achieved.
In this experimental situation, the adaptation takes place automatically if
terminal contact is allowed. With repeated reaches, the subject becomes progres-
sively more accurate without awareness of what he or she is doing. Moreover, after
adaptation is complete subjects no longer perceive the presence of the Coriolis forces
generated by their reaching movements. Their movements seem completely natural
and normal, and the Coriolis forces, although still generated during movements,
seem absent. Space does not permit pursuit of this important fact and its implications
for perception of the forces encountered during self-generated movements but this
issue is discussed further elsewhere by Lackner and DiZio.1
These studies demonstrate the importance of somatosensory cues from the hand
in the calibration and maintenance of limb position sense and in the adaptive
recalibration of arm movement control. A final series of studies will be described
to illustrate the contribution of haptic cues from the hand to balance.
all. Touch stabilizes posture even though the levels of applied force at the fingertip
are far below those adequate to provide mechanical stabilization. Nearly all subjects
spontaneously adopt about a 40 g level of applied force at their fingertip. This value
is interesting because it corresponds to the region of maximum dynamic sensitivity
of the somatosensory receptors in the fingertip as measured by Westling and Johans-
son.32 Variations about this level give the greatest modulation of somatosensory
afferent activity. Analyses of the relationship among the force changes at the fin-
gertip, EMG activity in the leg muscles controlling sway, and the displacement of
the body center-of-pressure reveal that the fingertip is not being used as a passive
probe to detect sway, but rather is being actively controlled. Changes in the leg EMG
signals follow changes at the fingertip by about 150 msec, and changes in body
position occur another 150 msec later (see Figure 5.11).33
Individuals without labyrinthine function cannot stand more than a few seconds
in the dark without falling if they are in a heel-to-toe stance. Nevertheless, if allowed
fingertip contact, they can balance as stably as normal subjects who are standing
without finger touch but with sight of their surroundings.34 This observation dem-
onstrates the great functional significance for spatial orientation of haptic inputs and
the possibility of substituting them for vestibular input in some circumstances. Touch
cues can also override proprioceptive misinformation about body orientation. If the
leg muscles of standing subjects are vibrated to elicit a reflexive contraction through
a tonic vibration reflex, they will be greatly destabilized and may fall over. However,
light touch of the index finger completely overrides the destabilizing influence of
leg muscle vibration.35 In fact, subjects are more stable during leg muscle vibration
in the dark if they are allowed touch than they are in the absence of vibration without
5.8 SUMMARY
Mechanical contact on the body surface is an omnipresent feature of terrestrial life.
We have shown here that such contact provides spatial information about the orien-
tation of the body with respect to the environment. During reaching movements, the
pattern of forces generated on the reaching hand when it contacts a surface provides
a spatial mapping of hand location relative to the body. This spatial coding of terminal
hand position is also extremely important for allowing adaptive changes in motor
control to be implemented. In fact, in the absence of vision and contact cues, reaching
accuracy rapidly degrades. Contact of the hand with the body itself provides a means
REFERENCES
1. Lackner, J. R. and DiZio, P. Aspects of body self-calibration. Trends Cog. Sci., 4,
279, 2000.
2. Alexander, R. M. Locomotion of Animals. Blackwell, Oxford, 1982.
3. Winter, D. A. Biomechanics and Motor Control of Human Movement, 2nd Ed. John
Wiley & Sons, New York, 1990.
4. Bigby, J. V. Some problems of postural sway. In Ciba Foundation Symposium on
Myotactic Kinethetic and Vestibular Mechanism, A. V. S. de Reuck and J. Knight,
Eds., Churchill, London, 1967.
5. Maxwell, S. S. Labyrinth and Equilibrium. Lippencott, Philadelphia, 1923.
6. Lackner, J. R. and Graybiel, A. Parabolic flight: Loss of sense of orientation. Science,
206, 1105, 1979.
7. Lackner, J. R. and Graybiel, A. Perceived orientation in free fall depends on visual,
postural, and architectural factors. Aviat, Space Environ. Med., 54, 47, 1983.
8. Lackner, J. R. and Graybiel, A. Postural illusions experienced during Z-axis recum-
bent rotation and their dependence on somatosensory stimulation of the body surface.
Aviat. Space Environ. Med., 49, 484, 1978.
9. Lackner, J. R. and Graybiel A. Some influences of touch and pressure cues on human
spatial orientation. Aviat. Space Environ. Med., 49, 798, 1978.
10. Lackner, J. R. and DiZio, P. Some efferent and somatosensory influences on body
orientation and oculomotor control. In Sensory Experience, Adaptation, and Percep-
tion, L. Spillman and B. R. Wooten, Eds., Erlbaum Associates, Clifton, NJ, 281, 1984.
11. Graybiel, A. and Hupp, D. I. The oculogyral illusion: a form of apparent motion
which may be observed following stimulation of the semicircular canals. J. Aviat.
Med., 17, 3, 1946.
12. Evanoff, J. N. and Lackner, J. R. Some proprioceptive influences on the spatial dis-
placement component of the oculogyral illusion. Perception and Psychophysics, 43,
526, 1988.
13. Lackner, J. R. and Zabkar, J. Proprioceptive information about target location sup-
presses autokinesis. Vision Res., 17, 1225, 1977.
14. Winterson, B. J., Steinman, R. M. Proprioceptive information neither improves fixa-
tion stability nor reduces autokinesis. Vision Res. 19, 1289, 1979.
15. Lackner, J. R., Fox, C., and DiZio, P. Autokinesis is suppressed by eye-lid contact
cues related to eye position. Submitted, 2001.
16. Levine, M. S. and Lackner, J. R. Some sensory and motor factors influencing the
control and appreciation of eye and limb position. Exp. Brain Res., 36, 275, 1979.
17. Goodwin, G. M., McCloskey, D. I., and Matthews, P. B. C. The contribution of muscle
afferents to kinaesthesia shown by vibration induced illusions of movement and by
the effects of paralysing joint afferents. Brain, 95, 705, 1972.
CONTENTS
6.1 Introduction
6.2 Cells vs. Systems
6.2.1 Historical Development
6.3 The Effect on Locomotion of the Loss of Sensation
6.4 Receptors Involved in the Control of Locomotion
6.4.1 Mechanoreceptors
6.4.2 Vision, Vestibular Input, and Hearing
6.5 Structure and Response Properties of Proprioceptors
6.5.1 Muscle Spindles: Structure
6.5.2 Passive Response Properties of Spindle Afferents
6.5.3 Fusimotor Action
6.5.4 Spindle Models
6.5.5 Tendon Organs: Structure and Response Properties
6.5.6 Tendon Organ Models
6.5.7 Mechanoreceptors in Joints, Ligaments, and Skin
6.5.8 Overview of Proprioceptive Firing during Locomotion
6.6 Simple Locomotor Reflexes
6.6.1 The Stretch Reflex
6.6.2 Flexion and Extension Responses, Reflex Stepping, and
Reflex Walking
6.7 Central Pattern Generators and Sensory Feedback
6.7.1 Cyclical Motor Patterns Generated without Sensory Input
6.7.2 Interaction between CPGs and Sensory Feedback
6.7.3 Human Locomotion
6.7.4 Robots
6.7.5 Virtual Animals
6.7.6 IfThen Rules Governing Phase Switching and the Selection
of Hazard Responses
6.1 INTRODUCTION
Dr. P.R. Burgess organized a Society for Neuroscience symposium in 1992 to discuss
his contention that You can only control what you sense. The question is, what is
being controlled in locomotion and which of the many sensory inputs to the CNS
are the main players? There are numerous ways of answering this, each implying a
different level of control and different neural systems. For example, at one level,
that which is controlled is support and movement of the body with respect to uneven
terrain. The control problem at this level is to cope with support surfaces of variable
orientation, consistency, stability, friction, and compliance. At another level, that
which is controlled is movement of the body with respect to a moving target (e.g.,
as in the hunting of prey). At this level, the problem is to anticipate future positions
of the target and to control and adapt ones own trajectory accordingly, taking into
account obstacles and hazards in the way.
1. The basic ability to produce voluntary force and move limbs is preserved
after deafferentation. However, movements are generally uncoordinated
and inaccurate, especially when visual guidance is absent.
2. Coordination of the different segments of the primate hand in precision
tasks is particularly impaired. The accuracy of spatial orientation, frac-
tionated movements, and anticipatory pre-shaping of the hand is reduced,
and writing may be severely affected.
Edin 2001) and it has been argued that they probably contribute to the sense of
position and motion of the extremities. There are massive numbers of skin receptors
in the limbs. For example, it has been estimated that there are about 17,000 skin
mechanoreceptors with myelinated afferent fibers on the surface of the human hand
(Johansson and Vallbo 1979), compared to about 4000 muscle spindles, 2500 tendon
organs, and a few hundred mid-range joint receptors in the whole arm (Voss 1971;
Hulliger 1984). Type I skin receptors and hair follicle receptors are responsive to
rapidly varying skin stimuli. Slowly adapting type II and III receptors respond
sensitively to stretching of the skin and continue to signal maintained stretch (Horch
et al. 1977). In two recent studies, it was shown that stretching of the skin overlying
finger joints produced illusions of movement of the fingers, reinforcing the idea of
a proprioceptive role for skin receptors (Edin and Johansson 1995; Collins and
Prochazka 1996). Figure 6.3 shows the firing rate of a slowly adapting type III
cutaneous receptor located at D in the left panel recorded in a human subject with
microneurography. The afferent clearly provided information about knee joint dis-
placement and velocity. Its firing profile was extremely well fitted with a first order
transfer function similar to that used in the simpler models of group Ia transduction.
Figure 6.5 also illustrates another aspect of proprioceptive reflexes that was quite
unexpected from a control theoretical point of view, and that has only recently been
understood analytically: positive force feedback. To the surprise of many in the field,
it was discovered in 1987 that signals from Ib afferents, which normally inhibit
homonymous and synergistic alpha motoneurons (MNs) in the absence of locomo-
tion, switched to exciting these motoneurons during fictive locomotion in the acute
spinal cat (Conway et al. 1987). This has been confirmed in the cat by several other
groups since (Pearson and Collins 1993; Guertin et al. 1995; Angel et al. 1996),
though the evidence for this switch in humans is equivocal (Dietz 1998; Stephens
and Yang 1999; Capaday 2000; Misiaszek et al. 2000; Pang and Yang 2000). MN
excitation causes muscles to generate more force, further increasing Ib firing, which
in turn would produce yet more excitation, i.e., positive feedback. Would this not
result in an explosive increase to maximal force? The surprising answer is that
because the force response of muscles to increments in activation declines as the
muscle gets shorter, an equilibrium length is reached at which the force-related
excitation does not cause further muscle shortening. Thus in Figure 6.5, for three
different levels of positive force-feedback gain, the muscle resisted stretching with
a much higher stiffness than could be achieved with spindle Ia feedback. Not only
As mentioned above, the neural basis of flexion and extension reflexes was first
studied in spinal and decerebrate dogs and cats in the late 19th Century. These
reflexes were elicited by aversive stimuli applied to the skin or cutaneous nerves of
the ipsilateral or contralateral limb. The movement synergies thus evoked were
similar, though not identical, to those in the flexion and extension phases of loco-
motion. Reflex stepping was studied in decapitated dogs by Freusberg (Freusberg
1874) and then in decerebrate and spinal cats by Sherrington and Brown (Sherrington
1910; Brown 1911). An hour or so after spinalization, locomotor movements could
be triggered by lifting the animal and dropping one hindlimb from a semi-flexed
starting position. An important observation of both Sherrington and Brown was that
position and loading of the limbs were crucial in triggering and maintaining reflex
stepping. If the downward motion of the hindlimb was again stopped in semi-flexion,
stepping movements ceased in this limb, as well as in the other, free limb. These
results have been verified in numerous experiments since (Rossignol 1996; Orlovsky
et al. 1999).
Reflex stepping could also be elicited by kneading, squeezing, or continuous
electrical stimulation of the skin of the tail, perineum, back, neck, or pinna. It could
also be elicited by continuous electrical stimulation applied to the transected spinal
cord at cervical level. Sherrington struggled to explain how these remote, continuous
stimuli applied to exteroceptive sensory receptors interacted with phasic sensory
signals from the limb, which he argued were mainly proprioceptive, to produce
cyclical movement. He suggested a scheme whereby the remote stimuli evoked a
2002 CRC Press LLC
primary reflex (causing either extension or flexion, depending on the site of
stimulation), which evoked proprioceptive stimuli in the moving limb. These elicited
a secondary reflex, which alternated with or interrupted the primary reflex to
produce cyclical activity. One flaw in this argument was preliminary evidence that
deafferentation did not necessarily abolish the rhythm. Sherrington wondered
whether the central rebound effect described a few years earlier by Magnus and von
Uexkuell (Umkehr) could be contributing. A year later, Sherrington presented a
paper to the Royal Society by his student Graham Brown (Brown 1911), confirming
the deafferentation results in decerebrated and spinalized cats. After a low thoracic
transection, the deafferented lumbosacral spinal cord could still generate rhythmical
movements in an isolated pair of hindlimb muscles in the absence of descending
input or sensory input from either hind limb. Graham Brown dubbed the underlying
mechanism the intrinsic factor, a term that has now been superceded by the more
specific term Central Pattern Generator (CPG) (Grillner and Zangger 1975).
Literally hundreds of studies in which sensory input has been reduced or abolished
by deafferentation have since demonstrated beyond doubt that isolated neuronal
networks in the CNS can generate the basic rhythmical motor patterns involved not
only in walking but also in the activities of breathing, chewing, swimming, flying,
scratching, paw-shaking, and autonomic functions such as micturition and sexual
reflexes (Grillner 1975; Prochazka 1996b; Kiehn et al. 1998; Orlovsky et al. 1999).
There has been much discussion about the ability of CPGs, in the absence of
sensory input, to generate complex coordinated patterns of muscle activity such as
those required for overground locomotion. After deafferentation in the cat, many
subtle features of normal activation sequences can still be seen (e.g., the small burst
of activity in knee flexor muscles at the end of the swing phase in cat locomotion
[Grillner and Zangger 1975]). However, the locomotor rhythm is generally more
labile (Grillner and Zangger 1975; Wetzel et al. 1976; Goldberger and Murray 1980;
Grillner and Zangger 1984; Giuliani and Smith 1987; Koshland and Smith 1989)
and above all, there is an inability to compensate for the changes in loading or terrain
(Allum et al. 1998).
This latter defect is more serious than it may sound. If information on joint
angles, body posture, loading, and displacement of the extremities are all unavailable
to the central controller, the amplitude and timing of the cyclical motor output can
only be set to some default level and cannot be matched to the varying requirements.
To quote Brown (1911):
6.7.4 ROBOTS
Recently there have been two interesting developments in technology that may
provide insight into these issues. The first is the design of walking robots. Some of
the most advanced work in this area is of a corporate nature (e.g., the humanoid
robots developed by the Honda company), and the information available on the
control strategies used tends to be sketchy. Nonetheless, it is quite clear that the
most versatile robots, including the extraordinary Honda Asimo P4, rely heavily on
sensory input to generate locomotion over uneven terrain. The robot designers
carefully reviewed the essential aspects of biological locomotor control in insects,
quadruped mammals, and humans and then implemented the most promising aspects
in their machines. The instructive thing here is to consider which control strategies
turned out to be effective. Two hexapod robots, a stick insect and a cockroach,
developed in Cleveland, provide useful information in this regard (Espenschied et al.
1996; Nelson and Quinn 1999). In both cases, the robots have six two- or three-
segment limbs. Each joint has a passive spring for compliance, an actuator, and a
position sensor. The actuator is under proportional position feedback control of
variable gain, providing a controllable stiffness that adds to the passive inherent
spring stiffness. The stiffness properties of each joint therefore mimic the intrinsic
properties of biological muscles under stretch reflex control. Locomotion is achieved
by a mixture of processes local to joints and legs, and two governing (global)
algorithms. The local processes include the active stiffness control just mentioned,
as well as IfThen control rules based on end-point position for stance-swing and
swing-stance transitions, and special rules for adaptive responses to tripping (stumble
reaction) and foot-in-hole (Gorassini et al. 1994). Figure 6.6 shows the reactions
of a leg of the hexapod robot in three situations, perturbed stance, placing reaction,
and foot-in-hole. For comparison, we have included a panel showing the searching
movements of a locust in which the leading limb enters a hole (at 2), searches, and
is eventually placed (at 3) (from a movie kindly provided by K. G. Pearson).
Regarding the two global algorithms, one adjusts leg trajectories to distribute
force equally among weight-bearing legs, and to match leg lengths according to
the terrain (i.e., to keep the body horizontal in the face of slopes and other unevenness
of ground support). Interestingly, leg length (the distance from the end of the paw
to the hip joint) has recently been identified as an emergent variable that accounts
best for the firing behavior of dorsal spinocerebellar tract neurons in the cat, even
when individual joints are constrained (Bosco and Poppele 2001). These authors
suggest that the spinocerebellar system may be viewed as the end-point of processing
of proprioceptive sensory information in the spinal cord. The other global algorithm
quoted encourages stance legs to lift into their swing phases in a coordinated
manner, swing forward, and transition to stance (Quinn and Ritzmann 1998).
Like the hexapod robots, the Honda robot also utilizes local active and passive
joint compliances, IF-THEN phase transitions, special adaptive reactions similar to
the ones just described and a novel global strategy of setting a moving target of
ground reaction force as the command for forward or backward locomotion. The
actual ground reaction force is continually computed from the sensor signals. The
difference between this vector and the target vector is referred to as the falling
moment. This falling moment is minimized by reflexes to the joint actuators,
which are presumably synergistically coupled.
Notice that in neither of the above robots is it easy to extract the notion of an
autonomous CPG from the various global algorithms. Rather, the central controllers
respond to external requirements by issuing general commands to move in particular
directions (e.g., by proposing a virtual trajectory for the ground reaction force),
selecting sensory rule bases appropriate to the task and context, and evaluating perfor-
mance for predictive adjustments. Hazard rules are also computed (e.g., the Honda
robot resists lateral imposed forces by stiffness control, but when sway exceeds the
imbalance point, it yields and takes a step in the direction of the imposed force).
velocity (Figure 6.8B). It does not help just to add stretch reflexes, because these
merely augment the muscle intrinsic stiffnesses without affecting step cycle phase-
switching. On the other hand, IfThen rules are useful in coping with variations of
this type. In Figure 6.8B, the IfThen rules governing the transitions from stance to
swing and back (see below) allowed the model to adapt to the increased EMG read-
out rate and hence increased velocity, though it did not fare so well with the lowered
read-out rate.
6.10 APPENDIX
6.10.1 STRETCH REFLEX MODEL
Appendix Figure 1 shows a Matlab Simulink model originally developed to test the
effect of excitatory Ib input to homonymous and heteronymous muscles (positive
force feedback). In this slightly modified version, we have added a feedback pathway
and product element marked by the A and B, to take into account the fact that
transmission from group Ia afferents to MNs scales with the number of MNs
recruited. The digital version of this model can be downloaded from the authors
website, or obtained upon request.
ACKNOWLEDGMENT
We thank Dr. Vivian Mushahwar for help with the graphics, Dr. K.G. Pearson for
providing us with the locust film, and Mr. Allen Denington for digitizing parts of
this film. This work was supported by the Canadian Institutes for Health Research
and the Alberta Heritage Foundation for Medical Research.
REFERENCES
Ach, N. ber die Willensttigkeit und das Denken. Gttingen 1905.
Adrian, E.D. and Zotterman, Y. The impulses produced by sensory nerve endings. Part 2. The
responses of a single end-organ. J. Physiol. 61, 151, 1926.
Akazawa, K., Aldridge, J.W., Steeves, J.D., and Stein, R.B. Modulation of stretch reflexes
during locomotion in the mesencephalic cat. J. Physiol. (Lond.) 329, 553, 1982.
Allum, J.H., Bloem, B.R., Carpenter, M.G., Hulliger, M., and Hadders-Algra, M. Propriocep-
tive control of posture: a review of new concepts. Gait Posture 8, 214, 1998.
Allum, J.H. and Honegger, F. Interactions between vestibular and proprioceptive inputs trig-
gering and modulating human balance-correcting responses differ across muscles.
Exp. Brain Res.121, 478, 1998.
Alnaes, E. Static and dynamic properties of Golgi tendon organs in the anterior tibial and
soleus muscles of the cat. Acta Physiol. Scand. 70, 176, 1967.
Anderson, J.H. Dynamic characteristics of Golgi tendon organs. Brain Res. 67, 531, 1974.
CONTENTS
7.1 Introduction
7.2 Candidate Models for the Distribution of Learning-Related
Modifications
7.3 Involvement of Sensory Cortex in Perception and Memory
7.4 Tactile Learning in a Topographic Framework
7.5 Evidence from Learning with the Whiskers
7.6 Evidence from Learning with the Fingers
7.7 Topographic Learning: Shared Principles across Species and
across Modalities
7.8 Conclusions
Acknowledgments
References
7.1 INTRODUCTION
One of the many ambitious goals set in our era of neuroscience is to understand
the brain modifications that underlie learning first, to identify the relevant neu-
ronal circuits and second, to describe the cellular changes occurring within those
circuits. For nearly three decades1 long-term potentiation and long-term depression
have been investigated, and synaptic plasticity of this sort is now believed to
constitute the cellular change underlying most types of learning.2 But progress in
the first of the two issues has been slower: we still know relatively little about the
locus of brain modifications during learning of even the simplest tasks. Exceptions
do exist: the cerebellar cortex and interpositus nucleus clearly form a locus of
plasticity during certain precisely-timed conditioned reflexes3 and the amygdala, a
critical locus during certain types of emotional conditioning.4 Yet, consider how
little is known about what parts of the brain are involved when a monkey learns to
remember visual stimuli across a delay. Lesion experiments point to an important
role of the hippocampus,5 but even the claim of the hippocampal locus can be called
into question.6 Thus, while the hippocampus is the most systematically studied
fMRI reported that subjects who knew they were about to receive a tickle-inducing
stimulus showed increased activity in the same cortical regions, including SI and
SII, that showed activation during actual tickling.22
modality23-25 and we have used the same strategy to uncover the distribution of
learning in the tactile modality.26-29 Our experiments, reviewed below, show that
when animals or human subjects learn a tactile discrimination task, the neural
modifications associated with the learning are distributed according to the spatial
arrangement of the cortical map.
With the prosthetic whisker in place, they were immediately tested. The main
result was that rats gap-crossed without delay only if the prosthetic whisker was re-
attached to the stub of the previously trained whisker. For any other whisker trans-
position, a period of relearning was required: indeed, the number of trials necessary
to reacquire the task increased systematically as a function of the distance along
rows and arcs between the trained and the prosthetic whisker. Rats showed no benefit
from their previous training if the prosthetic whisker was attached two or more
positions away from the site of the original whisker.
Learning with a single whisker might seem to be a special case, but additional
experiments demonstrated that the topographic learning rule was the same for multi-
whisker learning. When rats were trained with a set of four adjacent whiskers, the
learned ability quickly transferred to any of the whiskers immediately surrounding
the trained ones, but transferred less readily to non-adjacent whiskers. Moreover,
the topographic rule held up across the midline: rats could rapidly transfer learning
to whiskers symmetrically opposite the trained whiskers (e.g., left C3 to right C3),
but required additional training before successfully transferring to non-symmetric
opposite whiskers. The corpus callosum, which connects homotopic sites in the left
and right barrel field cortex,32 could be the neural substrate for this transfer.
From large-scale electrophysiological recordings using 100-microelectrode
arrays implanted in cortex (Figure 7.3), we found that the extent to which learning
transferred across whisker positions was perfectly correlated with the degree of
test surfaces were unfamiliar. They did not: there was no advantage for the trained
fingertip. This indicates that the tasks involved topographically distributed memory
for specific information.
If we accept the topographic learning rule, then the finding that the precise
pattern of the learning distribution depended upon the features of the acquired
information (Figure 7.4) should not be surprising. Since sensory neocortex is par-
celled into multiple areas, each specialized to process specific types of stimuli,37 the
spatial distribution of stored information should reflect the somatotopy present in
the cortical area that processes and stores that type of stimulus. Although a detailed
review is beyond the limits of this chapter, a survey of the literature gives a straight-
forward indication as to which cortical areas might be the critical substrates for the
experiments described above. Area 3b, containing neurons with rapidly adapting
response properties, single-digit receptive fields, and no callosal connectivity, is the
best candidate for vibration frequency learning.17,18,38-45 For roughness and punctate
pressure learning, the functional organization of area 43 (also known as SII) appears
to be well suited.46-53 For example, the dense callosal connectivity of SII is likely to
be the substrate for the transfer of roughness and punctate pressure learning to the
opposite, symmetric fingertip.
7.8 CONCLUSIONS
Behavioral and physiological studies suggest that the internal, map-like organization
of sensory cortical areas shapes the way sensory information is learned. When we
assert that information is stored within a topographic framework, we must clarify
the sort of information to which we are referring. In most situations, a large part of
what is learned is the behavioral strategy (consider, for example, the controversy
as to whether the Morris water maze best measures behavioral strategy learning as
opposed to spatial map learning58). An animal placed in an apparatus to make specific
sensory discriminations must first solve problems concerning the overall context of
the situation. One would hardly expect the learning of behavioral strategy to be
localized to a few cortical columns or even to a few cortical processing regions.
Indeed, Lashley is famous for showing (inadvertently) that learning of behavioral
strategy is poorly localized in the brain.59An additional complicating factor is that,
during regimes of sensory training extending across weeks or months, the distribution
of learning may change. The idea of a subcorticalization of motor procedures
during long periods of practice is common, and similar redistributions may occur
in sensory learning.
ACKNOWLEDGMENTS
Supported by the Telethon Foundation, the J.S. McDonnell Foundation, the European
Community, the Ministero dellUniversit e della Ricerca Scientifica e Tecnologica
(MURST), Consiglio Nazionale delle Ricerche, and the Regione Friuli Venezia
Giulia.
REFERENCES
1. Bliss, T.V. and Lomo, T., Long-lasting potentiation of synaptic transmission in the
dentate area of the anaesthetized rabbit following stimulation of the perforant path,
J. Physiol., 232, 331, 1973.
2. Bailey, C.H., Bartsch, D., and Kandel, E.R., Toward a molecular definition of long-
term memory storage, Proc. Natl. Acad. Sci. U.S.A., 93, 13445, 1996.
3. Thompson, R.F., Swain, R., Clark, R., and Shinkman, P., Intracerebellar
conditioning Brogden and Gantt revisited, Behav. Brain Res., 110, 3, 2000.
CONTENTS
8.1 Introduction
8.2 Psychophysical Studies of Tactile Attention
8.2.1 Ascending or BottomUp Mechanisms of Attention
8.2.2 Descending Mechanisms of Attention
8.2.3 Focus of Attention
8.3 Neurophysiological Studies of Tactile Attention
8.3.1 Anatomy
8.3.2 Effects of Attention on Neuronal Responses
8.3.2.1 Sub-Cortical Areas
8.3.2.2 SI Cortex
8.3.2.3 Memory in SI
8.3.2.4 SII Cortex
8.3.2.5 Prefrontal Cortex
8.3.3 Mechanisms of Attention
8.4 Attention and Imaging in Humans
8.5 Neglect Syndrome in Humans
8.6 Summary and Conclusions
Acknowledgments
References
8.1 INTRODUCTION
In this chapter, we review the psychophysical and neurophysiological basis of tactile
selective attention. As in other sensory systems, attention plays a major role in the
way that sensory inputs are processed and perceived. The importance of attention
in touch can easily be demonstrated by switching ones focus of attention to different
locations on the body. For example, if you switch your focus of attention to your
foot, you immediately become conscious of sensations arising from receptors in
your foot that were non-existent a moment earlier. This simple observation demon-
strates the power of selective attention and emphasizes two important aspects of
sensory processing. First, it shows that we have a limited information processing
8.3.2.2 SI Cortex
Initial studies suggested that attention had minimal effects on the response properties
of neurons in SI cortex.18 A later study by Hyvarinen et al.25 suggested that attention
had a small effect. They trained animals to perform both a vibration detection task
and to sit passively while the same stimuli were presented to the hand. They found
that about 8% of the neurons in area 3b and about 22% of the neurons in area 1
were affected by the animals focus of attention. More recently, two studies have
shown that attention has much greater effects on neurons in SI cortex.
Hsiao et al.24 trained animals to perform a tactile letter discrimination task and
a visual light detection task. During the recording sessions, the animals focus of
attention was switched back and forth every 5 to 10 minutes between the two tasks
while the same tactile stimulus was scanned across the distal pads of the animals
hand. The effects of attention were assessed by comparing a neurons responses to
the embossed letters for the two tasks. They reported that about 50% of the neurons
in areas 3b and 1 of SI cortex had increased firing rates when the animal performed
the tactile task; this study differed from the first two studies in that it required the
animal to perform a visual distracting task as the control. Having the animal perform
this task served two functions. One was to control for arousal effects since the animal
received rewards at about the same rate during both tasks. The other was to draw
the animals focus of attention away from the tactile task, which may explain why
these authors found greater effects in SI cortex.
An example of the response of a typical SA neuron in area 3b is shown in
Figure 8.3B. In this figure, the letters above the raster plots represent the letters of
the alphabet that were scanned across the animals finger and the target letter.
Although attention affected fewer neurons and had smaller effects on neurons in SI
than SII cortex, there was a clear enhancement of the neural responses when the
animal performed the tactile task relative to when it performed the visual task. The
enhanced responses were not only observed on the target letters but also on non-
reward letters that did not directly follow a target letter (e.g., see the response to the
letter O and the second letter A following the second target letter L in Figure 8.3A).
Attentional effects were not observed during two types of periods. One was time-
out periods that were triggered by false-positive responses. The other was reward
periods, which corresponded to non-target letters following target letters in which
the animal correctly identified the letter. The most likely explanation for the lack of
attention effects on these letters is that the animal learned that certain letters were
behaviorally irrelevant (e.g., two target letters never appeared in succession on the
drum).
The effects of attention on the mean impulse rates evoked by a target letter and
the following letter, while the animal was performing the tactile and visual tasks is
shown in Figure 8.4. Figure 8.4A shows a response that was typical of neurons in
SI and SII. Neurons in these areas showed a vigorous response to non-target letters
(e.g., letter P) following misses and minimal responses to these same letters following
hits. Figure 8.4B does not show the responses during tactile misses since the animal
had too few misses during these trials.
This study demonstrates several aspects concerning the effects of attention on
neurons in SI cortex. Besides showing that the neural response of a large fraction
of neurons in SI are affected by attention, it demonstrates that attention has large
effects on the responses of neurons and that these effects cannot be explained by
simple changes in neuronal gain. For example, the response to the target letter L
(Figure 8.4B) became weaker when the same animal performed the tactile task and
almost non-existent when the same letter passed over the receptive field and the
animal was performing the visual task. This suggests that attention modifies not
only the sensitivity of the neuron but also the spatiotemporal form of the response.
Second, this study demonstrates that attention is turned on and off quickly. In
Figure 8.4, the small arrows represent the mean time that the animal turned the
response switch. These responses suggest that the effects of attention on the neural
responses are engaged and disengaged within 100 msec, which indicates that the
8.3.2.3 Memory in SI
There is evidence that neurons in SI may play a role in the memory of tactile stimuli.
Zhou et al.50,51 trained animals to perform either a tactile-tactile matching task or a
cross-modal visual-tactile matching task. In these studies, animals were given either
a tactile cue (e.g., reach out and touch a bar containing horizontal bars) or a visual
cue (e.g., display with horizontal bars). Following a short delay, the animal was
trained to touch a bar that either contained a matching (horizontal bars) or non-
matching (vertical bars) stimulus. They found that, during the delay period between
the stimuli, many neurons in all four areas of SI cortex showed elevated rate changes.
While these rate changes could be attributed to attentional effects, they are also
indicative of neurons that may play a role in the short-term memory of tactile stimuli.
There have been several attention studies in SII cortex and all of them have shown
that selective attention has profound effects on the responses of neurons. In the initial
studies, Poranen and Hyvarinen39 trained animals to detect the cessation of a vibra-
tion and reported that, while attention had minimal effects on the responses of
neurons in SI, all of the neurons in SII were greatly affected. Similarly, Hsiao et al.,24
recording from animals performing the same task described earlier, reported that
more than 80% of the neurons in SII were affected by attention and that the attention
to the tactile stimulus caused both increases (58%) and decreases (22%) in the firing
rates of SII neurons. In general, the effect of attention on the responses of neurons
was similar in form but greater in magnitude to what was seen in SI cortex. Examples
of four neurons in SII are shown in Figure 8.3. The neurons shown in Figures 8.3A
8.3D, and 8.3E showed enhanced responses during the tactile tasks while the neuron
shown in Figure 8.3C showed suppressed responses. Figure 8.4A shows that the
spatial structure of the responses to the target letter X during hits and misses is
not a simple multiplicative gain from the response recorded while the animal per-
formed the visual task. These results suggest that attention alters the responses in
more complex ways than by simply changing the neural gain of the signal.
Recently, Hsiao and his colleagues found that about 30% of the neurons in SII
cortex show orientation-tuned responses and that the tuning is greatly affected by
attention.17 In their studies, they trained animals to perform a visual task and a match-
to-sample tactile orientation discrimination task. In the tactile task, the animal was
presented with a small bar at one of eight orientations on the distal finger pad for
500 msec. Then, one second later, the animal was presented with the bar at either
the same orientation or at the orthogonal orientation. The animals task was to push
a switch with its foot if the orientations of the two bars were the same and to pull
the switch if the orientation of the two bars differed. In the visual task, the animal
pulled the switch with its foot when a white square dimmed on a video screen.
Figure 8.5 shows an example of how attention affects the tuning properties of neurons
in SII cortex. In this example, the neuron shows minimal orientation tuning during
the visual task but shows strong tuning during the tactile task. This figure also shows
that neurons in SII exhibit attention modulated responses during the delay period
suggesting that SII may also play a role in tactile memory. For this neuron, the delay
period response shows the same tuning characteristics as the stimulus-driven
responses.
In another study of attention in SII cortex, Burton et al.5 studied animals per-
forming a vibration detection task similar to the one described earlier. They found
that 45% of the neurons in SII cortex were affected by the animals focus of attention.
Recently, Romo et al.42 have shown that neurons in the prefrontal cortex are affected
by the animals focus of attention. In their study, they trained the animals to perform
a vibrotactile match-to-sample task in which the animals were presented with two
sequential vibrations to the finger pads. The animal was required to press a key
indicating whether the second vibration was at a higher or lower frequency than the
first. They found that neurons showed a continuous discharge during the delay period
(up to 6 seconds) and that 65% of the neurons had discharge rates during the delay
period that varied as a monotonic function of the frequency of the base stimulus.
These results suggest that neurons in prefrontal cortex play a role in tactile working
memory.
ACKNOWLEDGMENTS
We would like to thank Dr. Dianne Pawluk for her insightful comments and John
Lane and Sapna Prasad for technical assistance. Supported by NIH R01-NS34086
and Robert Wood Johnson Foundation, grant number: 037218.
REFERENCES
1. Binkofski F, Buccino G, Stephan KM, Rizzolatti G, Seitz RJ, and Freund HJ (1999)
A parieto-premotor network for object manipulation: evidence from neuroimaging.
Exp Brain Res 128: 210-213.
2. Buchner H, Reinartz U, Waberski TD, Gobbele R, Noppeney U, and Scherg M (1999)
Sustained attention modulates the immediate effect of de-afferentation on the cortical
representation of the digits: source localization of somatosensory evoked potentials
in humans. Neurosci Lett 260: 57-60.
3. Burton H, Abend NS, MacLeod AM, Sinclair RJ, Snyder AZ, and Raichle ME (1999)
Tactile attention tasks enhance activation in somatosensory regions of parietal cortex:
a positron emission tomography study. Cereb Cortex 9: 662-674.
4. Burton H and Sinclair RJ (2000) Tactile-spatial and cross-modal attention effects in
the primary somatosensory cortical areas 3b and 12 of rhesus monkeys. Somatosens
Mot Res 17: 213-228.
5. Burton H, Sinclair RJ, Hong SY, Pruett JR, and Whang KC (1997b) Tactile-spatial
and cross-modal attention effects in the second somatosensory and 7b cortical areas
of rhesus monkeys. Somatosens Mot Res 14: 237-267.
6. Bushnell MC, Duncan GH, Dubner R, Fang L, and He LF (1984) Activity of trigem-
inothalamic neurons in medullary dorsal horn of awake monkeys trained in a thermal
discrimination task. J Neurophysiol 52: 170-187.
7. Bushnell MC, Duncan GH, Dubner R, Jones RL, and Maixner W (1985) Attentional
influences on noxious and innocuous cutaneous heat detection in humans and mon-
keys. J Neurosci 5: 1103-1110.
8. Carli G, LaMotte RH, and Mountcastle VB (1971) A simultaneous study of somatic
sensory behavior and the activity of somatic sensory cortical neurons. Fed Proc 30: 664.
9. Craig JC (1985) Attending to two fingers: two hands are better than one. Percept
Psychophys 38: 496-511.
10. Crick F (1984) Function of the thalamic reticular complex: The searchlight hypoth-
esis. Proc Natl Acad Sci U S A 819: 4586-4590.
11. Dorsch, AK, Hsiao, SS, Johnson, KO, and Yoshioka, T (2001) Tactile attention:
Subjective magnitude estimates of roughness using one or two fingers. Society for
Neuroscience Abstracts.
CONTENTS
9.1 Introduction
9.2 What Studies Can Digital Video Enable?
9.3 How Is Digital Video Implemented?
9.4 How Are the Neurophysiological Data Analyzed?
9.4.1 Spike Separation
9.4.2 Burst Analysis
9.4.3 Spike Rasters and PSTHs
9.4.4 Statistical Analyses of Firing Rates
9.4.5 Receptive Field Mapping
9.5 How Are Hand Behaviors Quantified Using Digital Video?
9.6 What Instrumentation Is Recommended?
9.6.1 Video Equipment
9.6.2 Computer Hardware and Software
9.7 What Can We Learn from Imaging Actions of the Hand?
9.8 Useful Websites (listed alphabetically)
Acknowledgments
References
9.1 INTRODUCTION
The brains own body image is normally expressed in both a cognitive and pragmatic
manner. By projecting the indentation patterns impressed upon the skin onto external
objects that touch its surface, the tactile sense in the hand serves a cognitive function.
Information derived from tactile receptors about an objects intrinsic properties its
neurons. In this manner, we could accurately map receptive fields and qualitatively
assess whether the neuron under study responded to the behavioral task.
In addition, and most important for these studies, the camcorder also recorded
the spike trains of cortical neurons on its audio channels, providing a simple method
for linking neurophysiological and behavioral data. The video images and simulta-
neously recorded spike trains were stored together on videocassettes for subsequent
analysis. Although we could play back these videotapes at a later time, simply
viewing tapes of the animal and listening to the spike trains (or displaying them on
an oscilloscope) yielded little beyond the same qualitative information that could be
gleaned during the experiment itself. In order to make direct frame-by-frame quan-
titative correlations between video images of behavior and spike data, we digitized
both sets of recordings.
Digital editing software was used to download these files to a computer and to
examine details of the hand kinematics in digitized images displayed on a computer
monitor. The hand behavior could be viewed in real time, at high speed, or by
selected individual frames. Forward and backward bracketing of sequential frames
was particularly useful for visualizing how the hand or the stimulus changed position
over time. As each image was labeled during the initial data acquisition with time
taneously to provide different views of the subject. One can allow the cameras to
roll continuously as electrodes are repositioned to record what might be unexpected,
but interesting, behaviors as the session proceeds.
File Format) files. We therefore make a duplicate copy of the QuickTime audio
track(s) using the File Export command in the editing software, selecting audio only
and AIFF format.* Our protocols use Superscope II, as its built-in functions auto-
matically implement many of the desired analyses. AIFF files from each clip are
imported into a Superscope instrument and subdivided into one-minute fragments,
which are analyzed independently to facilitate interactive spike separation. This
improves processing efficiency when setting spike thresholds or cluster boundaries,
as a single channel sampled at 32 kHz yields nearly 4 MB of data per minute. The
entire data trace is reconstructed afterwards by linking sequential data fragments.
* Adobe Premiere (v. 5) only supports AIFF file export in Mac OS; Final Cut Pro (v. 2) supports both
Mac OS and Windows for audio file export.
of continuous recordings were synchronized, analyses performed for one set of clips
could be matched to views obtained from the other cameras.
The usefulness of the burst analysis technique for comparing responses to dif-
ferent behaviors is illustrated in Figure 9.3. These continuous records show the first
minute of recording from another clip derived from our studies of prehension
behaviors in monkeys. In order to study prehension in a reproducible manner, we
trained the animals to perform a task in which objects are grasped and manipulated
in response to visual cues. The objects tested were rectangular, round, and cylindrical
knobs that protruded from a box placed directly before the animal (shape box,
Figures 9.1 and 9.2). The animal was reinforced for grasping a visually-cued knob,
and lifting it vertically, 58 mm, until an upper stop was contacted; the knob had to
be returned to the lowest position before the next trial was initiated. We divided the
task into 8 stages 1: Approach. 2: Contact. 3: Grasp. 4: Lift. 5: Hold. 6: Lower.
7: Relax. 8: Release. Stages 13 occurred during object acquisition; stages 4 and 5,
during manipulation; and stages 68, upon relaxation and release of grasp. Start
times of the 8 task stages were determined by visual inspection of the digitized
(dark blue). Firing dropped precipitously at the onset of holding (light blue), and
did not resume until after grasp was released (cyan). Significant activity was
observed immediately upon release only when it was followed by approach toward
another object.
Although this neuron responded to tactile stimulation of the hand, the onset of
activity 250 ms before contact suggests a role in motor planning of grasping. A
further increase in firing rates during grasping indicates that tactile cues relevant to
task performance were enhanced during active acquisition of objects, while irrelevant
stimuli may be suppressed by hand movements (gating studies reviewed in Chapman
et al., 1996). The sensory responses of cortical neurons illustrated in Figure 9.4 may
match expected and actual sensations to guide subsequent hand actions.
of movement of different parts of the hand as the action progressed. In the example
shown in Figure 9.6, the animal picked up a raisin from the top of the shape box.
The preshaping of the fingertips for precision grip is clearly visualized, and the grip
aperture can be measured. Forearm movements were stabilized as the fingers scooped
up the food morsel and enclosed it in a fist; finger movements were clearly resolved
at 30 fps because they are relatively slow compared to reaches to distant targets.
FIGURE 9.7 (previous pg.)Kinematics of the prehension task traced from a series of sequential
video frames. A. Approach trajectory. Drawings of three frames spaced 66 ms apart outline
the hand posture during movements between the right and left rectangular knobs. The dashed
lines project the center of gaze, which moved from the computer monitor that provided location
cues (image 1) to the selected grasp site (image 2); gaze was maintained at the grasp site
until the hand was placed there (image 3). B. Grasp and manipulation. After contact, the
hand was slid forward along the knob surface as grasp was initiated, until the inner edge
reached the base of the fingers (image 4). Gaze was lifted back toward the screen as grasp
was tightened, and the object lifted by the hand (image 5). The hand posture in these drawings
is nearly identical to that shown in Figure 9.2 from another session, demonstrating the
stereotyped behavior adapted by each animal when it performed the task.
Motion trajectories can also be constructed from the images. For example, in
Figure 9.8, we drew oval-shaped markers at the wrist in a series of aligned frames,
and then placed all of the markers in one layer. This allowed us to visualize the
entire path of hand movement between knobs 4 and 1 (right panel), and from knob
2 to a raisin placed near the right edge of the shape box (left panel); two trials are
illustrated for each movement. As in previous kinematic studies using expensive
infrared (IRED) monitoring systems (Jeannerod, 1984; Jeannerod et al., 1995; Roy
et al., 2000), these reach-to-grasp movements were executed in a smooth trajectory
in which the peak velocity of the wrist occurred midway between start and end
locations. Furthermore, although the 30 fps sample rate of DV is an order of
magnitude slower than that of IRED systems, this was sufficient to outline the reach
path. The DV frame rate is adequate for measuring manipulatory movements of the
fingers, especially when the proper view angle is selected (Figures 9.2, 9.6, and 9.7).
REFERENCES
Babu, K. S., Debowy, D. J., Ghosh, S., Hu, E. H., Harris, A., Natiello, M., and Gardner, E. P.,
Spike burst analysis: a tool for analysing natural prehension behaviors recorded with
digital video. Soc. Neurosci. Abstr. 26, 2202, 2000.
Bovik, A., Ed. Handbook of Image and Video Processing. San Diego, CA: Academic Press,
2000.
Chapman, C. E., Tremblay, F., and Ageranioti-Blanger, S. A., Role of primary somatosensory
cortex in active and passive touch. In: Wing, A. M., Haggard, P., and Flanagan, J.R.,
Eds. Hand and Brain. Academic Press, San Diego, CA, pp. 329, 1996.
Debowy, D. J., Babu, K. S., Hu, E. H., Ghosh, S., Harris, A., Natiello, M., and Gardner, E. P.,
Spike burst analysis: a tool for analysing trained prehension behaviors recorded with
digital video. Soc. Neurosci. Abstr. 26, 2202, 2000.
Debowy, D., Ghosh, S., Ro, J. Y., and Gardner E. P., Comparison of neuronal firing rates in
somatosensory and posterior parietal cortex during prehension. Exp. Brain Res. 137,
269, 2001.
Edin, B. B. and Abbs, J. H., Finger movement responses of cutaneous mechanoreceptors in
the dorsal skin of the human hand. J. Neurophysiol., 65, 657, 1991.
Edin, B. B. and Johansson, R. S., Skin strain patterns provide kinaesthetic information to the
human central nervous system. J. Physiol. Lond., 487, 243, 1995.
Fenton, A. A. and Muller, R. U., Using digital video techniques to identify correlations
between behavior and the activity of single neurons. J. Neurosci. Meth.,70, 211, 1996.
Gardner, E. P., Ro, J. Y., Debowy, D., and Ghosh, S., Facilitation of neuronal activity in
somatosensory and posterior parietal cortex during prehension. Exp. Brain Res., 127,
329, 1999.
Hulliger, M., Nordh, E., Thelin, A. E., and Vallbo, .B., The responses of afferent fibers from
the glabrous skin of the hand during voluntary finger movements in man. J. Physiol.
Lond., 291, 233, 1979.
Jeannerod, M., The timing of natural prehension movements. J. Mot. Behav., 16, 235, 1984.
Jeannerod, M., Arbib, M. A., Rizzolatti, G., and Sakata, H., Grasping objects: the cortical
mechanisms of visuomotor transformation. TINS, 18, 314, 1995.
Johansson, R. S., Sensory control of dexterous manipulation in humans. In: Wing, A. M.,
Haggard, P., and Flanagan, J. R., Eds., Hand and Brain. Academic Press, San Diego,
CA, pp. 381, 1996.
Lederman, S. J., Browse, R. A., and Klatzky, R. L., Haptic processing of spatially distributed
information. Percept. Psychophys., 44, 222, 1988.
CONTENTS
10.1 Introduction
10.2 Ways of Representing the Code
10.2.1 Recent Considerations of the Temporal Aspects of Input
Representations
10.2.2 Spike Train Quantification
10.2.3 Average Firing Rate after Initial Responses
10.2.4 Percentage Entrainment of the Response to the Stimulus
10.2.5 Synchronicity of the Response with the Stimulus
10.2.6 Phase and Mean Phase of the Response
10.2.7 Interspike Intervals, Joint Interval Plots,
and Autocorrelograms
10.3 What the Code Can Reveal
10.3.1 Changes in Sensory Responsiveness
10.3.2 Sensory Responsiveness Changes with Changes
in Expectation
10.3.3 Sensory Responsiveness during Guided Movements
10.4 Summary
Acknowledgments
References
10.1 INTRODUCTION
Sir Charles Sherrington, in the Hughlings Jackson Lecture given on January 29,
1931, paid homage to his predecessor by stating a number of comments that are no
less attractive today. He suggested that Jackson would have been pleased with the
thought that The nervous system is indeed both a form and a series of events. These
Considerable effort by others also has been devoted to understanding firing patterns
of neurons within the sensorimotor cortices and relating the time-variant patterns of
CNS activity to the actual pattern of the stimuli being delivered and the behaviors
being produced (some recent studies include References 919). Mechanoreceptive
afferents respond relatively consistently over time.8,20 Experience-dependent
response plasticity in sensorimotor cortical neurons, however, may be continually
modulated by mechanisms that probably involve changes in the threshold that allow
calcium entry, are NMDA receptor dependent, and are GABA regulated.21-23 A recent
report has suggested that experience-dependent cortical plasticity may result from
at least two mechanisms.24 The first appears to be a change in the excitatory con-
nectivity of cortical elements; the second may be a release from inhibition by cortical
interneurons. The structure of spike trains that represent peripheral stimulus features
may degrade as the information ascends in the CNS. By the time that it reaches the
cortex, the fidelity with which the temporal pattern of activity is synchronized with
peripheral events may have diminished.2 The activity patterns of single cortical
neurons may not exhibit as faithful a representation of stimuli as ensembles of
neurons. Single neuron changes in firing rate may not convey as much information
about stimulus characteristics as populations of neurons, either by singular or cor-
related activity (see References 2527).
delivered to a location on the skin at time zero. Arrayed below are four possible
responses, shown in the form of rasters, which differ in the extent to which they
represent entrainment to the stimulus. From above, the first represents several trials
in which, although stimulation evoked an increase in firing rate, this response was
not visibly affected by stimulus periodicity. The second shows what might happen
if a neuron received a light stimulus, and consequently was lightly entrained to it.
With moderately intense peripheral stimulation, visible neural entrainment would
be expected in certain neurons. However, if the stimulus is intense, as previously
observed, neurons may fire more than one spike per stimulus cycle and therefore
exhibit a second, relatively consistent vertical band in raster displays.
The datasets used to construct the panels were derived based on the following
assumptions. First, there is a fixed probability of spike occurrence in a given stimulus
cycle, equivalent to mean firing rate. Second, each spike is phase-locked to the
stimulus with a certain precision. Accordingly, a spike was assigned to every 3 out
of 4 stimulus cycles. The timing of each spike relative to the start of the stimulus cycle
of the great benefits of using flutter vibration as a peripheral stimulus is that it and
neuronal activity associated with it can be described as a function of stimulus phase.
Provided that stimulus phase is consistent as a function of presentation time, then
spike occurrences translate directly to the phase of the stimulus at that time. If then,
the response latency (i.e., the conduction time from periphery to the central location)
remains relatively constant, shifts in phase plots may indicate changes in represen-
tational fidelity that accompany changes in the sensory environment. Figure 10.3
shows the phase plots that were derived from the datasets previously described above.
Several measures can be calculated using phase plots.
Two decades ago, Ferrington and Rowe29 used an entrainment index they called
Percentage Entrainment (PE) to quantitatively describe the distribution of spike
occurrences relative to the phase angle of the sinusoidal stimuli delivered to the
sensory periphery. PE is calculated by plotting the distribution of spike occurrences
as a function of stimulus phase, binning the results in some reasonable way. By
then, calculating the number of spikes in each half-cycle and comparing that ratio
with the total, the PE, which ranges between 0.5 and 1.0 can be expressed.
Figure 10.3 illustrates what happens if the 40 trials from the sample datasets are
plotted in this manner. It is easy to see that PE correlates fairly well with increases
vibratory entrainment, rather than merely by firing rates. Following this line of
reasoning, changes in the degree of variability in spike occurrences may indicate
changes in fidelity for that subpopulation. Shifts in the distribution of spikes without
significant changes in variability may indicate changes in threshold for response,
assuming that neurons spike once per stimulus cycle. These shifts could be associated
where r is the mean vector length, is the frequency of vibration, ti represents the
time of spike occurrences, and n is the number of spikes. The value of r ranges from
0.0 to 1.0, where 0.0 corresponds to the uniform cycle distribution, and 1.0 corre-
sponds to discharges at a constant phase relative to the stimulus cycle. It is this
phase-locking or entrainment that we said earlier was an indication of the fidelity
of the central representation of peripheral sensory events. In addition to the mean
vector length, the angular deviation, the polar version of the standard deviation, can
be calculated and several determinations can be made from that. One of the most
important calculations that then must be done is to determine if the phase distribution
of neuronal discharges is non-uniform (i.e., perhaps indicating vibratory entrain-
ment), by using Rayleigh Test (Reference 31; pp. 54-55). In this test, a z-score is
calculated according to the formula:
z = nr2
where r is the mean vector length and n is the number of elements in the parent
distribution. The values of z greater than 4.6 were taken as satisfying the criterion
of statistical significance, which corresponded to < = 0.01. Another is to re-
represent the vector distribution, by doubling the phase angle, if the initial distribu-
tion appears bimodal.31 This is sometimes necessary if a spike train contains spikes
centered around both a phase and its antiphase. The resultant mean vector angle
would tend to misrepresent the actual distribution since it would be at approximate
right angles to both arms of the distribution and the mean vector length would be
uncharacteristically short (see below). Data from an area 1 somatosensory cortical
neuron, plotted in polar coordinates, is displayed in Figure 10.4F. The length of the
mean vector, 0.88, indicates that the neuron is tightly coupled to the sinusoidal
stimulus presented to its peripheral receptive field.
are doubled and replotted (Panel E), the length of the mean vector (r) improves
somewhat. That is, the value is greater on a scale from 0 to 1. The center column,
that for the 57 Hz trials, illustrates both the strengths and the weaknesses of each
of these ways of visualizing and quantifying the data. It is clear from the histogram
that the post-stimulus firing rate is lower than that for the 27 Hz trials. It is also
evident that there are gaps in which there is little or no spiking. These occur
immediately after stimulus onset and just prior to movement onset (open squares).
FIGURE 10.7 (See color Figure 10.7 in color insert.) Population response patterns of four types of area 1 cue-related neurons as a function of
trial type and behavioral time. Scatterplots constructed by binning all ISIs (5 ms 1 ms) and normalizing by spikes/trial to give probability of spike
occurrences in a bin. Probability indicated by attributed color. Tick marks to the right indicate time corresponding to the period of the 57 Hz vibratory
stimulus. Response types as in Figure 10.6. Nomenclature for trial types as in text.
in the VIS-only, and COM paradigms. To date, we have examined the data for only
10 movement amplitudes from one monkey. All correct movements were rewarded
in these self-paced tasks. The animal was not penalized for variations in RTs, MTs,
or intertrial intervals. Complete records (all three paradigms) were obtained for
403/691 task-related neurons; 156 were located in pre-central and 247 in post-central
cortical areas. As shown in Figure 10.8B, reconstruction of the electrode penetrations
indicated that the recordings in the post-central cortex were mainly in areas 3a, 3b, and 1.
The general behavior of the monkey was quantified by examining records from
the post-central cortical neurons. RTs, during the VIS-only task, were greater than
those for VIB-only and COM for both wrist extension and flexion movements (p <
0.01). MTs for wrist flexion were longer than that of wrist extension. However, there
is no significant difference for MTs across the three paradigms as a function of
direction.
Analyses were conducted for 74/247 (~30%) post-central cue-related neurons
using analytical procedures developed in the previously described study. These
neurons have been grouped, by visual inspection, into C (n = 25), Q (n = 17), S (n =
16), and V (n = 16) neurons, based on their responses to vibratory stimuli.
Figure 10.8C shows the responses of a V neuron which exhibited no cue-related
discharges in VIS-only trials, but was entrained to the vibratory stimulus in both
VIB-only and COM-cue paradigms. These paradigms will subsequently be men-
tioned collectively as the vibratory paradigms (or trials). Activity onsets, BKG,
QAMFR, SAMFR were measured. Mean vector length was calculated for all neurons
with cue-related firing rate increases during the vibratory paradigms.
The animal behaved similarly during the vibratory paradigms, but there were
some notable differences in response patterns. During post-central neuron recording,
RTs for the vibratory trials averaged 275 11 ms and were significantly shorter than
those for VIS-only trials (297 15; p<.0001; repeated measures ANOVA). BKG was
not different across the three paradigms. Comparing the firing rates across the
vibratory paradigms yielded no statistically significant differences in QAMFR and
SAMFR, although firing rates during these two epochs, as well as during BKG,
tended to be higher in VIB-only trials. However, across these paradigms, there were
significant differences in mean vector length for the V neurons but not for C and Q
neurons. V neurons were better entrained to the premovement part of the 57 Hz
vibratory cue in the COM-cue trials (mean 0.393 .24) than in the VIB-only trials
(mean 0.275 .17; p<.0001; repeated measures ANOVA). Quantifying the entrain-
ment during movement to the targets was somewhat difficult because the frequency
of the vibration is time-variant (by experimental design) and the MTs, while rela-
tively uniform, vary by times equal to several periods of the final 127 Hz vibratory
stimulus. These variations can be seen in the population plots presented in
Figure 10.9.
From this figure, several initial observations, later confirmed statistically, were
noted. It appeared that in the vibratory paradigms, both C and Q neurons encoded
the vibratory stimuli to some extent. While the general probability of spike occur-
rence in any given bin was higher in the the VIB-only paradigm for these classes
of neurons, the spread of the distribution was also wider, suggesting that entrainment
may not be as good as in the COM-cue paradigm. It appeared that V neurons may
be more poorly entrained to vibratory stimuli prior to movement but better entrained
during movement and once the target was achieved and the new position held. This
suggestion is based on the intensity of the spike occurrences at multiple periods of
the final (127 Hz) stimulus frequency at times (~400900 ms) when the monkeys
held the handle still at the new position. In several cases, activity patterns immedi-
ately after go-cue onset were qualitatively different when changes in vibratory
frequency specified target location. In these instances, cessations in activity were
seen that did not occur in either VIS-only or COM trials.
Appropriate quantification methods of the entrainment during and after move-
ments are still being developed. Conceptually, the latter seems easier to accomplish
since the records could be aligned in a time corresponding to the beginning of a
vibratory cycle in which the end of the movement occurs. Thus, the records would
maintain their temporal association with the 127 Hz stimulus, independent of movement
time. To initially see if this might be the case, data can be plotted using one of the
methods described above. Figure 10.10 shows the re-plotting of one of the population
scattergrams from Figure 10.9, only this time with the data centered on the onset of
wrist movement. In this case, it is the lower left panel which depicts the ISIs for the
area 1 V neurons. Although mean vector analyses have not been conducted as yet for
these data, several things are clear. First, these neurons initially show entrainment to
the 57 Hz vibratory stimulus that acted as the go-cue to trigger the movements. Second,
during the movement (from movement onset to arrival at target), there is apparently a
smooth transition in the ISIs from the initial to the new (127 Hz) vibratory frequency
that reflects the experimental design. Finally, once the target, and hence the new handle
position is attained, these neurons appear to be entrained to the high frequency and not
suppressed by it as was the neuron whose records are illustrated in Figure 10.5. The
distribution of ISIs, once the target is reached, is quite narrow, suggesting that these
neurons closely follow the frequency of the vibration of the handle at the target zone.
These observations suggest that when movement endpoints are specified by
somatosensory cues alone, initial increases in firing rates may be accompanied by
degradations in the fidelity of responses. Prior to movement, ongoing responses to
vibration are suppressed for some types of neurons. Some neurons are more poorly
entrained by vibratory stimuli in trials without explicit visual targeting cues. It
appears from population plots that this entrainment may improve significantly once
movements, guided by vibration alone, begin. If so, it would be consistent with the
idea that suppression of activity (or correlated firing) occurs before instances of
highest activity/movement coupling to improve signal-to-noise ratios. Certainly,
more studies are needed to examine suppression and facilitation of responses to
10.4 SUMMARY
This chapter began by asking how one might go about determining the fidelity with
which peripheral stimuli are represented by sensorimotor cortical neurons. To start
to answer this question, the responses to vibratory stimuli were considered because
these stimuli, by nature, have a temporal signature. Neurons that spike with relatively
consistent ISIs corresponding to the frequency of known peripheral stimuli are well
documented.7 What historically has been difficult to do is to find the means of
quantifying the fidelity of signal representation in the responses of neurons to those
signals. Several ways of doing so have been presented here. Perhaps none is sufficient
alone. Taken together, however, these analytical tools provide a means of suggesting
answers to several questions about how cortical representations work when animals
are working.
As stated earlier, what is presented here is by no means an exhaustive review
of all the ways in which data may be represented to address the question of response
fidelity. That was not the intent. The intent was to describe some forms of illustrating
pertinent data in forms that are readily tractable to those not necessarily in the sub-
discipline and measurements that can accompany these forms of data representation.
Some changes in sensory responsiveness are far from subtle. For those, differences
in mean firing rates may be sufficient to test the hypotheses under consideration.
Certainly, stimulus intensity is well correlated with firing rates over much of the
dynamic range of sensorimotor cortical neurons. Stimulus location in this system is
represented in a number of topographic, though distorted, representations of the
sensory periphery.4 However, some influences on sensation of peripheral events and
on behaviors that may accompany these sensations are subtle, such as attention,
intention to move, expectation of goal attainment, and Knowledge of Result (KR).
It is therefore reasonable to think that firing pattern changes associated with the
alterations just mentioned may themselves be subtle. These changes may be in the finer
aspects that represent fidelity and not necessarily as mean firing rate changes. Some of
the data analyses and data representations discussed above are useful in deciphering
the code that the periphery sends the central nervous system as the brain deals with the
sensory environment and its own body image.
ACKNOWLEDGMENTS
The original work described herein and the formulation of this manuscript was
funded by research grant R01-NS36860 from the National Institute of Neurological
Disorders and Stroke to Randall J. Nelson. We thank John M. Denton for his
assistance in drafts of the manuscript.
CONTENTS
* However, a model being developed by Burnod and colleagues (1999) illustrates that functional models
can also capture these trends while providing some pragmatic rationale for them.
* The classification of area 4 cells used here is based on Kalaska et al. (1989).
The first six of these hypotheses are described in detail by Bullock et al. (1998),
where evidence for them is reviewed. The seventh hypothesis is discussed by Cisek
et al. (1998) in the context of various proprioceptive illusions induced by muscle
tendon vibration, as well as movements in the presence of elastic loads and Coriolis
forces. In this chapter, I will focus primarily on the first two hypotheses, those
concerned with the computations proposed to occur in the posterior parietal cortex.
Below, I attempt to bring together diverse sources of experimental data on the
functional role of the parietal cortex. Perhaps some of this data is being inappropri-
ately forced into my particular perspective. The parietal cortex is a complex brain
region, with many functional roles. Its study is made difficult by a lack of consensus
on what regions in the human are homologous with those in the monkey, and even
by a lack of a consistent nomenclature for specific recording sites. Therefore, the
following discussion should merely be taken as well-intentioned speculation.
* This very important observation led Iacoboni (1999) to conclude that reaching movements are fully
planned before movement onset. However, I believe that this takes the data too far. I question that
conclusion below, and present several alternative hypotheses which are consistent with available data.
In the original implementation of the FAVITE model (Bullock, Cisek, & Gross-
berg, 1998), the only source of directional information for the precentral DVV was
the area 5 DV, generated online using the area 5 PPV. However, suppose that a
second source of directional information comes from a forward model, which has
learned to anticipate the temporal profile of the DV for some reasonably familiar
movement context (Figure 11.2). During performance, the input to the pre-central
DVV may come solely from the forward model, and the parietal DV computation
may only be used when sudden unexpected perturbations (such as a change in target
location) disrupt the course of movement.
Suppose that the transcranial magnetic stimulation used by Desmurget et al.
(1999) disrupted the PPV and DV computations occurring in the parietal cortex.
This means that the forward model has become the only source for guiding move-
ment, and all perturbations, including changes in target location, are ignored. The
accuracy of movement is compromised, but the hand does at least hit the right
ballpark. In other words, one can hypothesize that in addition to an online DV in
area 5, there are other sources of directional information available to update the pre-
central descending command, including perhaps a forward model of the kinematics.
A number of recent studies support the hypothesis that the posterior parietal
cortex is primarily involved in the online guidance of movements using currently
available (and attended) sensory information. During tasks requiring the recalibration
of sensorimotor transformations, whether adapting to displacing prisms (Clower,
Hoffman, Votaw, Faber, Woods, & Alexander, 1996) or to a rotated reference frame
(Ghilardi, Ghez, Dhawan, et al., 2000), positron emission tomography reveals activ-
ity changes in the posterior parietal cortex. Lacquaniti et al. (1997) found increased
activation in the inferior parietal lobule in an immediate-pointing task, as compared
with a task in which pointing movements were made to a previously memorized
location. Consistent with these results, the accuracy of memory-guided saccades is
The extrinsic PPV (PPVe) can be computed in one of two ways: either using
visual feedback of the hand, if that is available, or otherwise through a transformation
of the PPV in intrinsic coordinates (PPVi). The DV in intrinsic coordinates (DVi)
can also be computed in one of two ways: either through a transformation of the
extrinsic DV (DVe), if available, or otherwise by subtracting the PPVi from a motor
TPVi. In each case, the computation based on visual information takes precedence
over the computation using proprioceptive intrinsic coordinate signals.
Visual presentation of a target activates the TPVe population and visual feedback
of the position of the hand activates the PPVe. With these two sources of input, the
DVe can be computed in extrinsic spatial coordinates and transformed into a DVi
in intrinsic arm coordinates, which is then used to update the pre-central descending
command. Note that the DVi is computed based on the available DVe, and not by
subtracting the PPVi from the TPVi. Thus, during reaching movements to visual
targets with full visual feedback of the arm, the PPVi and TPVi populations are not
strongly involved, since the PPVe and TPVe are available. Assuming that the trans-
formation from DVe to DVi is accurate (as it should be after years of practice), such
conditions allow very accurate movement.
When vision of the arm is prevented, the PPVi comes into play. The PPVi is
computed based on efference copy and spindle feedback (as in the original FAVITE
model), and transformed into the PPVe through another well-practiced and accurate
coordinate transformation. Thus, the difference vector is again computed in spatial
coordinates (DVe) and then transformed into intrinsic arm coordinates (DVi). The
reaching movement is less accurate under these conditions, because it relies on an
imperfect estimate of the position computed using efference copy and proprioceptive
feedback.
Under other conditions, the target may not be defined through the visual modality
at all. Suppose instead that a somatosensory target is presented to the system, as in
the case of scratching an itch. In this case, the TPVi is defined but the TPVe is not, and
so the difference vector computation cannot occur in extrinsic spatial coordinates.
Instead, it occurs directly in intrinsic coordinates (DVi) using the available TPVi
and the somatosensory PPVi. In other words, all the activity during proprioceptively
guided movement is confined to the bottom row of the Figure 11.3b circuit.
In summary, the circuit of Figure 11.3b allows very flexible use of various
sources of information regarding current arm position and target location to allow
reaching movements under many different conditions. I introduce it here because it
also provides a functional context for interpreting the results of Desmurget et al., as
discussed below.
In the Desmurget et al. (1999) experiment, there are essentially four cases: 1)
movements to stationary targets without TMS; 2) jumping targets without TMS; 3)
stationary targets with TMS; and 4) jumping targets with TMS. In all of these cases,
the target itself is presented visually, but there is no visual feedback of the moving
hand. Figure 11.4a illustrates what happens in the first two cases. Because the target
is presented visually, it activates the TPVe population. However, since the hand is
out of view, the only position information comes from efference copy and peripheral
feedback, activating the PPVi population. Thanks to a learned transformation, the
PPVi is converted into a PPVe, which then allows the computation of a DVe in
extrinsic spatial coordinates. This is then coverted into a DVi, which controls the
movement. During movement, the DVe is continuously recomputed based on the
current PPVe and TPVe, and thus automatically corrects for target jumps. However,
because the computation of the PPVi is not perfectly accurate, the movement tra-
jectories exhibit some endpoint inaccuracies.
Consider now what happens in the cases with transcranial magnetic stimulation
(Figure 11.4b). When the target is first visually presented at the beginning of a trial,
it activates TPVe, which in turn generates TPVi activity through a learned coordinate
* In a system with redundant degrees of freedom, this transformation is non-trivial. Because each extrinsic
target position (TPVe) can be mapped onto many intrinsic target arm configurations (TPVi), the mapping
is ill-posed. To address this difficulty, one might propose that the system simply has one preferred arm
configuration for every extrinsic position.
11.5 CONCLUSIONS
As stated in the introduction, the usefulness of a functional model is that it provides
a pragmatic context for interpreting data. It leads one beyond questions of how
cellular activities co-vary with various movement-related information and toward
questions of how that information is actually used in a mechanism for guiding
movement. There are two reasons why this is important. First, the statement that
some variable affects some pattern of activity is merely an observation and not an
hypothesis. Testing that observation only confirms whether it is accurate, but does
not leave us with an explanation. Second, by considering the pragmatic concerns
facing a system such as the cortical circuits for movement control, one is led toward
those issues which motivated the evolution of that system. Motor control is not about
describing actions but about performing actions, and accuracy of a description does
not always go hand in hand with accuracy of a movement. For example, a veridical
representation of the current position of the limb is not always the most useful
element for a successful control scheme. In the FAVITE model, the gain of peripheral
feedback is deliberately reduced during fast movements in order to avoid the insta-
bilities inherent in a feedback system with conduction delays (Cisek, Grossberg, &
Bullock, 1998). This results in a system that does not represent position accurately,
but performs better than a system which makes veridical position sense a priority.
There are other examples. The Difference Vector is fundamentally a variable
useful for directing movement and determining its completion. In essence, all that
a good DV has to do is point in the right direction and shut off when the hand
reaches the target. The precise representation of distance is of lesser importance, as
long as other (possibly pre-central) mechanisms take care of shaping the speed
profile. This means that the DV can combine other sources of information without
disrupting movement performance, as discussed in the example below.
Consider what happens when several possible targets for a movement are avail-
able. Based on data from the oculomotor system (Glimcher & Sparks, 1992; Platt
& Glimcher, 1997) and the arm movement system (Cisek & Kalaska, 1999), one
can speculate that the brain can begin to prepare several potential directional signals
simultaneously before selecting one to guide overt execution. Kinematic trajectories
of movements made in the presence of distractors support this proposal (Tipper,
Howard, & Houghton, 1998; Tipper, Lortie, & Baylis, 1992). If that is the case, then
how does the brain decide between different actions? One candidate mechanism is
a competition between representations of different movement directions in the pos-
terior parietal cortex. Suppose that different potential movements are represented in
parietal cortex as different DVs, each of which represents direction but not distance.
REFERENCES
Ashby, W.R. An Introduction to Cybernetics. Chapman and Hall, 1956.
Ashby, W.R. Design for a Brain: The Origin of Adaptive Behaviour. London, Chapman and
Hall, 1965
Ashe, J., and Georgopoulos, A.P. Movement parameters and neural activity in motor cortex
and area 5. Cerebral Cortex, 4, 590, 1994
Bhushan, N., and Shadmehr, R. Computational nature of human adaptive control during
learning of reaching movements in force fields. Biological Cybernetics, 81, 39, 1999.
Bioulac, B., and Lamarre, Y. Activity of post-central cortical neurons of the monkey during
conditioned movements of a deafferented limb. Brain Research, 172, 427, 1979.
Bizzi, E., Accornero, N., Chapple, W., and Hogan, N. Posture control and trajectory formation
during arm movement. Journal of Neuroscience, 4, 2738, 1984.
Bizzi, E., Hogan, N., Mussa-Ivaldi, F.A., and Giszter, S.F. Does the nervous system use
equilibrium-point control to guide single and multiple joint movements? Behavioral
and Brain Sciences, 15, 603, 1992.
Bracewell, R.M., Mazzoni, P., Barash, S., and Andersen, R.A. Motor intention activity in the
macaques lateral intraparietal area. II. Changes of motor plan. Journal of Neurophys-
iology, 76, 1457, 1996.
CONTENTS
12.1 Introduction
12.2 The Feedforward Model of Touch
12.3 The Potential Role of Corticothalamic Feedback Projections
in Tactile Information Processing
12.4 Sensory Gating of Neural Responses during Active
Tactile Exploration
12.5 The Computational Argument in Favor of a Recurrent View
of Touch
12.6 Conclusions
Acknowledgments
References
12.1 INTRODUCTION
For the past five decades, neurophysiological theories aimed at accounting for the
exquisite tactile perceptual capabilities of mammals have been dominated by the view
that the somatosensory system, the network of subcortical and cortical neurons special-
ized in processing somatic information, relies primarily on feedforward (FF) computa-
tions to generate a broad spectrum of sensations (e.g., fine touch, thermo sensation,
pain, etc.).29 Although there are many nuances and variations of this general
model,29,57,78,79,121 for the sake of simplicity, throughout this chapter, we will refer to this
view as the feedforward model of touch. It is important to emphasize at the outset that
the central assumptions made by the proponents of this classical model of touch are
almost identical to those used in an equivalent feedforward model of vision. Conse-
quently, some of the arguments presented here have also appeared in recent studies that
propose interactive models of vision (for a discussion in vision see Reference 20).
The wide acceptance of the FF model of touch can be hardly overstated. One
needs only to inspect the somatosensory chapters of recent medical textbooks to
FIGURE 12.1 (next pg.) (See Color Figure 12.1 in color insert.) Schematic diagram of the
rat trigeminal somatosensory system. Whiskers on the rats snout are labeled according to the
row and column in which they are located. Whisker columns are labeled from 1 to 5, caudal
to rostral, while whisker rows are labeled A to E, dorsal to ventral. Peripheral nerve fibers
innervating single whisker follicles have their cell bodies located in the trigeminal ganglion
(Vg). Here, only the projections from Vg neurons to two main subdivisions of the trigeminal
brainstem complex, the principal trigeminal nucleus (PrV) and the spinal trigeminal nucleus
(SpV), are illustrated. Proponents of the feedforward model of touch usually divide these
projections into rapidly adapting (RA) and slowly adapting (SA) fibers, according to their
physiological responses to tactile stimuli (see text). Each of these categories contains further
subdivisions, which are not described here. Neurons located in these two brainstem nuclei
give rise to parallel excitatory projections to the ventroposterior medial nucleus (VPM) of
the thalamus. Neurons in VPM give rise to projections to layer IV of the primary somatosen-
sory cortex (SI). A collateral of these thalamocortical projections reach the reticular nucleus
(RT), whose neurons provide the main source of GABAergic inhibition to the VPM. Descend-
ing excitatory corticothalamic projections, originating in layer VI of the SI cortex, reach the
VPM and the reticular nucleus of the thalamus (RTn). The assumed topographic arrangement
of these projections in the VPM and the RT are illustrated in the scheme. Feedback corticofugal
projections originated in layer V of the SI cortex also reach the trigeminal brainstem complex,
targeting primarily the SpV subdivision.
In addition to these findings, previous studies have shown that the presence of
one stimulus can alter the ability of cortical and subcortical neurons to respond to
a subsequent stimulus for a period of time.116,117 Evidence from our experiments and
other studies9,10 suggests that the ability of one tactile response to modulate the
magnitude of a subsequent one is substantially decreased during motor activity
(Figure 12.5). For example, when two tactile stimuli were presented with an inter-
stimulus interval of 2575 ms in the absence of any whisker movements (i.e., awake
immobility), the response to the second stimulus was significantly reduced
(Figure 12.5). However, during periods in which the same rats produced whisker
movements, the response of VPM and SI neurons to the second stimulus was not
statistically different in magnitude from the first at any inter-stimulus interval tested
(Figure 12.5). Further examination of these results indicated that these effects par-
alleled a change in the amount of post-excitatory inhibition that follows the first IO
stimulus in different behavior states (immobility vs. whisker movements). Thus, in
the absence of any movement of the whiskers, we observed the occurrence of a long
period of reduced firing, following the presentation of the first tactile stimulus.50,116,117
However, this period is substantially shorter (in VPM) or non-existent (in SI) during
the presence of exploratory whisker movements (Figure 12.4), suggesting that motor
activity-related changes in post-stimulus inhibition could account for the differential
responses to paired stimuli we observed in different behavioral conditions.
Overall, the results of these experiments suggest that during different behavioral
states, different types of thalamocortical transmission may occur (in this case, awake
immobility vs. whisking) and that these different modes of transmission may serve
different perceptual purposes. Thus, differences in cortical and subcortical tactile
response characteristics, from periods of whisker immobility to periods of whisker
movements, suggest that the somatosensory system can shift from a state of high
FIGURE 12.6 (Left Panel) A prototypical ART model circuit adapted from Grossberg.47 A
short-term memory (STM) activation pattern evoked across feature detectors in level F1 by
ascending input is transmitted to level F2 where a compressed STM of the ascending patterns
is created. Learned expectations are matched through topdown pathways against bottomup
input evoked patterns in F1; the difference in this matching, relative to the vigilance param-
eter, , is used to initiate a search for a more appropriate or novel recognition code. (Right
Panel) Grossbergs ART model provides a theoretic framework that fits well our view of the
functioning of the trigeminal, somatosensory system. The appropriateness of this comparison
extends beyond the superficial similarities illustrated in the figure between the ART models
input layer, non-specific inhibitory gain control, F2, and F1, layers and the trigeminal systems
input, reticular nucleus, thalamus, and cortex. In addition to these structural similarities, it is
the dynamic interplay between ascending spatiotemporal input patterns and descending expec-
tations, captured in both systems to generate adaptive memories that is complementary and
appealing.
12.6 CONCLUSIONS
Although the classic feedforward model of touch has provided a fundamental blue-
print for the development of somatosensory research in the last five decades, a variety
of experimental findings and theoretical arguments demonstrate that this model no
longer offers a useful account on how tactile perception may emerge in the mam-
malian brain. Instead, anatomical, physiological, and computational arguments favor
the hypothesis that tactile perception emerge through interactive and recurrent inter-
actions between the multiple cortical and subcortical levels that define the mamma-
lian somatosensory system. Central to this recurrent model of touch is the experi-
mental demonstration that the massive corticofugal projections, that originate in the
neocortex and reach most of the subcortical structures that form the somatosensory
system, may play as relevant a role in tactile information processing as the parallel
feedforward pathways of this system. In addition, we propose that a recurrent model
of touch has to acknowledge unequivocal inseparability of the somatosensory and
motor systems, which together serve as the true neuronal substrate of tactile percep-
tion. Finally, a recurrent model of touch offers a more biologically relevant account
of the computations required for learning to occur throughout life.
REFERENCES
1. Adams, N.C., Lozsadi, D.A., and Guillery, R.W., Complexities in the thalamocortical
and corticothalamic pathways, European Journal of Neuroscience, 9, 204, 1997.
2. Albe-Fessard, D., Condes-Lara, M., Kesar, S., and Sanderson, P., Tonic cortical
controls acting on spontaneous and evoked thalamic activity. In G. Macchi, A. Rus-
tioni, and R. Spreafico, Eds., Somatosensory Integration in the Thalamus: A Reeval-
uation Based on New Methodological Approaches, Elsevier, Amsterdam, 1983.
3. Andersen, P., Junge, K., and Sveen, O., Cortico-thalamic facilitation of somatosensory
impulses, Nature, 214, 1011, 1967.
4. Andersen, P., Junge, K., and Sveen, O., Cortifugal facilitation of thalamic transmis-
sion. Brain, Behavior, Evolution, 6, 170, 1972.
5. Anderson, P., Eccles, J.C., and Sears, T.A., Cortically evoked depolarization of pri-
mary afferent fibers in the spinal cord, Journal of Neurophysiology, 27, 63, 1964.
6. Bishop, C.M., Neural Networks for Pattern Recognition. Clarendon Press, Oxford,
1995.
7. Bourassa, J., Pinault, D., and Deschenes, M., Corticothalamic projections from the
cortical barrel field to the somatosensory thalamus in rats: a single-fibre study using
biocytin as an anterograde tracer, European Journal of Neuroscience 7, 19, 1995.
8. Carvell, G.E. and Simons, D.J., Biometric analyses of vibrissal tactile discrimination
in the rat. Journal of Neuroscience, 10, 2638, 1990.
9. Castro-Alamancos, M.A. and Connors, B.W., Short-term synaptic enhancement and
long-term potentiation in neocortex. Proceedings of the National Academy of Sciences
U.S.A., 93, 1335, 1996.
10. Castro-Alamancos, M.A. and Connors, B.W., Spatiotemporal properties of short-term
plasticity in sensorimotor thalamocortical pathway of the rat. Journal of Neuroscience,
16, 2767, 1996.
11. Chapin, J.K., Modulation of cutaneous sensory transmission during movement: pos-
sible mechanisms and biological significance. In S.P. Wise, Ed., Higher Brain Func-
tion: Recent Explorations of the Brains Emergent Properties, John Wiley & Sons,
New York, 181, 1987.
12. Chapin, J.K., Sadeq, M., and Guise, J.L.U., Corticocortical connections within the
primary somatosensory cortex of the rat, Journal of Comparative Neurology, 263,
326, 1987.
13. Chapin, J.K. and Woodward, D.J., Modulation of sensory responsiveness of single
somatosensory cortical cells during movement and arousal behaviors, Experimental
Neurology, 72, 164, 1981.
14. Chapin, J.K. and Woodward, D.J., Somatic sensory transmission to the cortex during
movement: gating of single cell responses to touch, Experimental Neurology, 78, 654,
1982.
15. Chapin, J.K. and Woodward, D.J., Somatic sensory transmission to the cortex during
movement: phasic modulation over the locomotor step cycle, Experimental Neurology,
78, 670, 1982.
CONTENTS
13.1 Introduction
13.2 Topographic Hierarchy
13.2.1 Dominant Inputs
13.2.2 Latent Inputs
13.2.3 Other Inputs
13.3 Stages of Plasticity
13.3.1 Changes that Appear Rapidly
13.3.2 Changes that Develop Over an Intermediate Time Course
13.3.3 Changes that Require a Long Time Course
13.4 Mechanisms of Plasticity
13.4.1 Adjustments in Synaptic Efficacy
13.4.2 Sprouting
13.4.3 Integration across Multiple Levels of the Pathway
13.5 Conclusions
References
13.1 INTRODUCTION
Over the past decade, the literature on plasticity has grown and there now are diverse
and manifold descriptions of somatosensory plasticity. In some studies, plastic
changes are invoked by modifications in behavior, through training or alterations in
the environment. However, in the vast majority of cases, evidence for plasticity
comes from deafferentation studies, where peripheral sensory afferents are either
silenced or eliminated. Both experimental paradigms produce shifts in the balance
of activity from the peripheral sensory inputs, and a common assay is to measure
changes in sensory maps, or the map substrates. At first pass, the breadth and
diversity of the descriptions may seem overwhelming, but after consideration of the
data, what emerges is a sense that some basic features of reorganization are common
across experimental paradigms, levels of the system, and species.
FIGURE 13.2 (next pg.) Schematic showing patterns of topographic organization in area
3b of macaque monkeys after amputation of part or all of the forelimb. A. The topographic
organization of the hand representation in area 3b typical of normal macaque monkeys.10,11
Large numerals 15 refer to the representations of digits 15, and thin lines indicate borders
between representations. B. The representation of the hand in a monkey that lost the middle
and distal phalanges of digit 2. The top panel shows the approximate extent of the denervated
region in area 3b (black fill). The bottom panel is a schematic rendering of the pattern of
reorganization that emerged. Arrows indicate that adjacent representations expanded into the
deprived zone. Adapted from Manger et al.22 C. After amputation of all the digits on one
hand for treatment of injury, the region of area 3b where the digits are normally represented
is taken over by a large representation of the palm of the hand. Conventions are the same as
described for B. Adapted from Florence et al.25 D. After amputation at the level of the wrist,
the region where the hand would be represented in nornal monkeys is taken over by repre-
sentations of the wrist. Conventions are the same as described for B. Adapted from Florence
and Kaas.26 E. After amputation at the mid-humeral level of the arm, the deprived region is
taken over by an expanded representation of the stump of the arm medially and the face
laterally. Conventions are the same as described for B. Adapted from Reference 25. C, caudal;
D, dorsum of the hand (dors.); F, face; FA, forearm; M, medial; P, palm; p2, proximal phalange
of digit 2.
virtually all cases where somatosensory deactivation has been followed over
extended time frames.
In primates, much of the information on time course of reactivation comes from
studies of the effects of median nerve transection, to denervate the lateral half of
the palmar hand. The emergence of new receptive fields on the dorsal surface of the
hand may require many weeks before the full extent of potential change is expressed
(Figure 13.5). Initially, neurons only along the borders of the deactivated zone are
responsive to cutaneous stimulation of adjacent intact parts of the hand.2,18,34,44
13.5 CONCLUSIONS
In the somatosensory system, some principles of reorganization are common to
nearly all descriptions of plasticity after peripheral sensory deprivation. The recur-
rence of these same principles, despite different manipulations used to induce reor-
ganization and differences in outcomes, suggests that the system adapts to peripheral
modulations using a only few simple strategies.
Three common principles are discussed in the present chapter based on evidence
from studies in the primate somatosensory system. The principles are summarized
below.
REFERENCES
1. Merzenich, M. M. et al., Topographic reorganization of somatosensory cortical areas
3b and 1 in adult monkeys following restricted deafferentation, Neuroscience, 8, 33,
1983.
2. Merzenich, M. M. et al., Progression of change following median nerve section in
the cortical representation of the hand in areas 3b and 1 in adult owl and squirrel
monkeys, Neuroscience, 10, 639, 1983.
3. Wall, J. T., Nepomuceno, V., and Rasey, S. K., Nerve innervation of the hand and
associated nerve dominance aggregates in the somatosensory cortex of a primate
(squirrel monkey), Journal of Comparative Neurology, 337, 191, 1993.
4. Merzenich, M. M. et al., Somatosensory cortical map changes following digit ampu-
tation in adult monkeys, Journal of Comparative Neurology, 224, 591, 1984.
5. Jain, N., Catania, K. C., and Kaas, J. H., Deactivation and reactivation of somatosen-
sory cortex after dorsal spinal cord injury, Nature, 1997.
6. Florence, S. L., Wall, J. T., and Kaas, J. H., The somatotopic pattern of afferent
projections from the digits to the spinal cord and cuneate nucleus in macaque mon-
keys, Brain Research, 452, 388, 1988.
CONTENTS
14.1 Introduction
14.2 The Problem of Detecting Brain Plasticity
14.3 Reorganization after Deafferentations
14.4 Motor Cortex Reorganizations
14.5 Reorganizations of Somatosensory Cortex Following Changes
in Sensory Stimulation and Experience
14.6 Mechanisms of Neural Plasticity
14.7 The Perceptual and Motor Consequences of Plasticity
Acknowledgments
References
14.1 INTRODUCTION
The developing and mature nervous system must solve two major problems. The
brain must be able to function properly, or at least well, and it must be able to
change. We will see that these two problems are aspects of the same task. I am
reminded of a 1931 painting by Salvador Dali called The Persistence of Memory.
In the painting, the faces of clocks had become rather fluid, and they hung loosely
and folded over tree branches. The plastic clocks reminded me that we do retain our
memories over long periods of time, but the memories change and gradually become
distorted. While we clearly do have brain mechanisms for storing information, or
ACKNOWLEDGMENTS
This paper was prepared while the author was a Fellow at the Center for Advanced
Study in the Behavioral Sciences. I am grateful for the financial support provided
by The John D. and Catherine T. MacArthur Foundation, and research support by
NIH Grant NS16446.
REFERENCES
Arckens, L., Eysel, U. T., Vanderhaeghen, J. J., Orban, G. A., and Vandesande, F., Effect of
sensory deafferentation on the GABAergic circuitry of the adult cat visual system,
Neuroscience, 83, 381, 1998.
Brenman, J. E. and Bredt, D. S., Synaptic signaling by nitric oxide, Current Opinion in
Neurobiology, 7, 374, 1997.
Brodmann, K., Vergleichende Lokalisationslehre der Grosshirnrinde, Barth: Leipzig, 1909.
Brown, T. H., Chapman, P. F., Kairiss, E. W., and Keenan, C. L., Long-term synaptic poten-
tiation, Science, 242, 724, 1988.
Buonomano, D. V. and Merzenich, M. M. Cortical plasticity: From synapses to maps, Annual
Review of Neuroscience, 21, 149, 1998.
Byl, N. N., Merzenich, M. M., and Jenkins, W. M., A primate genesis model of focal dystonia
and repetitive strain injury: I. Learning-induced dedifferentiation of the representation
of the hand in the primary somatosensory cortex in adult monkeys, Neurology, 47,
508, 1996.
Calford, M. B. and Tweedale, R., Immediate and chronic changes in responses of somatosen-
sory cortex in adult flying-fox after digit amputation, Nature, 332, 446, 1988.
Chino, Y. M., Receptive-field plasticity in the adult visual cortex: Dynamic signal rerouting
or experience-dependent plasticity, Seminars in the Neurosciences, 9, 34, 1997.
Chino, Y. M., Kaas, J. H., Smith, III, E. L., Langston, A. L., and Cheng, H., Rapid organization
of cortical maps in adult cats following restricted deafferentation in retina, Vision
Research, 32, 789, 1992.
Cohen, L. G., Bandinelli, S., Findley, T. W., and Hallett, M., Motor reorganization after upper
limb amputation in man: A study with focal magnetic stimulation, Brain, 114, 615, 1991.
Cohen, N. J. and Eichenbaum, H., Memory, Amnesia, and the Hippocampal System, M. I. T.
Press: Cambridge, MA, 1993.
Darian-Smith, C. and Gilbert, C. D., Axonal sprouting accompanies functional reorganization
in adult cat striate cortex, Nature, 368, 737, 1994.
Darian-Smith, C. and Gilbert, C. D., Topographic reorganization in the striate cortex of the
adult cat and monkey is cortically mediated, Journal of Neuroscience, 15, 1631, 1995.
Dinse, H. R., Recanzone, G. H., and Merzenich, M. M., Alterations in correlated activity
parallel ICMS-induced representational plasticity, Neuroreport, 5, 173, 1993.
Donoghue, J. P., Suner, S., and Sanes, J. N., Dynamic organization of primary motor cortex
output to target muscles in adult rats: II. Rapid reorganization following motor nerve
lesions, Experimental Brain Research, 79, 492, 1990.