Corollary discharge and spatial updating: when the

Corollary discharge and spatial updating: when the
Progress in Brain Research, Vol. 149
ISSN 0079-6123
Copyright ~ 2005 Elsevier BV. All rights reserved
CHApTER 14
Corollary discharge and spatial updating: when the
brain is split, is space still unified?
Carol L. Colby1,*, Rebecca A. Berman1, Laura M. Reiser1 and Richard C. Saunders2
i Department of Neuroscience, University of Pittsburgh, Center for the Neural Basis of Cognition,
Mellon Institute, Room U5, 4400 Fifth Ave., Pittsburgh, PA 15213-2683, USA
2Laboratory of Neuropsychology, National Institute of Mental Health, Room 1B80, Building 49,
49 Convent Drive, MSC 4415, Bethesda, MD 20892-4415, USA
Abstract: How does the brain keep track of salient locations in the visual world when the eyes move? In parietal, frontal
and extrastriate cortex, and in the superior colliculus, neurons update or 'remap' stimulus representations in
conjunction with eye movements. This updating reflects a transfer of visual information, from neurons that encode a
salient location before the saccade, to neurons that encode the lòcation after the saccade. Copies of the oculomotor
command - corollary discharge signals - must initiate this transfer.
We investigated the circuitry that supports spacial updating in the primate brain. Our central hypothesis was that
locations across visual hemifields, from one
the forebrain commissures provide the primary route for remapping spatial
cortical hemisphere to the other. Further, we hypothesized that these commissures provide the primary route for
communicating corollary discharge signals from one hemisphere to the other. We tested these hypotheses using the
double-step task and subsequent physiological recording in two split-brain monkeys. In the double-step task, monkeys
made sequential saccades to two briefly presented targets, Tl and T2. In the visual version of the task, the
representation of T2 was updated either within the same hemifield ("visual-within"), or across hemi:telds ("visualacross"). In the motor version, updating of the visual stimulus was always within-hemifield. The corollary discharge
signal that initiated the updating, however, was generated either within the same hemisphere ("motor-within") or in the
opposite hemisphere ("motor-across"). We expected that, in the absence of the forebrain commissures, both visual-
across and motor-across conditions would be impaired relative to their "within" controls.
In behavioral experiments, we observed striking initial impairments in the monkeys' ability to update stimuli across
visual hemifields. Surprisingly, however, both animals were ultimately capable of performing the visual-across
sequences of the double-step task. In subsequent physiological experiments, we found that neurons in lateral
intraparietal cortex (LIP) can remap stimuli across visual hemifields, albeit with a reduction in the strength of
remapping activity. These behavioral and neural findings indicate that the transfer of visual information is
compromised, but by no means abolished, in the absence of
the forebrain commissures. We found minimal evidence of
impairment of the motor-across condition. Both monkeys readily performed the motor-across sequences of the doublestep task, and LIP neurons were robustly active when within-hemifield updating was initiated by a saccade into the
opposite hemifield. These results indicate that corollary discharge signals are available bilaterally. Altogether, our
findings show that both visual and corollary discharge signals from opposite hemispheres can converge to update
spatial representations in the absence of the forebrain commissures. These investigations provide new evidence that a
unified and stable representation of visual space is supported by a redundant circuit, comprised of cortical as well as
subcortical pathways, with a remarkable capacity for reorganization.
*Corresponding author. E-mail: cco1by(gcnbc.cmu.edu
DOl: 10.1016/80079-6123(05)49014-7
, 187
rip
¡ .
188
Introduction
Updating involves a transfer of visual
and motor signals
We perceive a visual world that is richly detailed,
stable, and continuous. This perception allows us to
In the past two decades, neurophysiological studies
perform a range of spatial behaviors, from reaching
have provided considerable insight into neural
for a cup of coffee to navigating through a busy
mechanisms that contribute to the phenomenon of
street. The ease with which we perform these actions
spatial constancy. Single-unit recording studies in
gives the impression that our sensory experience is
awake, behaving monkeys indicate that several
a direct - and passive - reflection of the world
around us. Our perception, however, is by no means
brain areas participate in updating spatial representa-
a transparent read-out of incoming sensory inputs.
extrastriate cortex, and in the superior collculus,
neurons exhibit a surprising kind of activity, which
exemplifies the important influence of action upon
perception. Neurons in these areas have classical
The active nature of perception is readily appreciated when we consider the nature of the visual
signals that arrive at the periphery. We explore and
analyze the world using the high-acuity center of the
retina, the fovea. In order to direct the fovea toward
visual responses, firing when stimuli appear within
objects of interest we make rapid eye movements,
About every 300 millseconds, the brain receives a
when a saccade brings the receptive field onto a
previously stimulated location - even though no
physical stimulus ever appears within the field (Mays
new image, yet we are oblivious to these nearly
and Sparks, 1980; Goldberg and Bruce, 1990;
continuous displacements of the retinal scene. What
Duhamel et aI., 1992a; Walker et aI., 1995; Umeno
and Goldberg, 1997, 2001; Nakamura and Colby,
2002). This firing, called remapping, is a response to a
called saccades, about three times each second.
we perceive is an internal representation of the
visual world, which seamlessly compensates for our
own movements.
How does the mind construct this stable representation of visual space from such constantly changing
input? In 1866, Helmholtz observed that when he
passively displaced his eye by gently pressing it, the
image of the world was also displaced (Helmholtz,
1866). In contrast, when he displaced his eye by
the receptive field. These neurons also fire, however,
memory trace of the stimulated location, which has
been updated in conjunction with an eye movement.
Remapping provides a dynamic internal representathe visual world that takes our eye movements
into account.
tion of
Remapping requires the communication of visual
generating a voluntary eye movement, the image of
as well as motor signals. When the eyes move, the
visual representation must be transferred from neu-
the world remained stilL. Helmholtz proposed that our
rons that encode the stimulus location before the eye
perception of the visual world is kept stable by the
"effort of wil" associated with making an eye movement. This "effort of wil," placed in the context of
movement, to neurons that wil encode the location
contemporary physiological studies, is a copy of the
command, the corollary discharge signaL. Recent
motor command that generates the saccadic eye
movement. This corollary discharge can support the
i
computations needed to anticipate what the visual
world will look like once the eyes reach their new
location. By using corollary discharge signals, the
brain can update the internal representation of space,
keeping it in register with the incoming retinal signals.
..
tions when the eyes move. In parietal, frontal, and
after the eye movement (Colby and Goldberg, 1999).
This transfer must be initiated by a copy of the motor
studies emphasize that these motor signals contribute
vitally to visual processing and spatial beJiavior
(Guilley and Sherman, 2002a,b; Guilery, 2003;
Sommer and Wurtz, 2004b). Spatial updating is one
such instance in which corollary discharge informa-
\
t
tion must playa role. For example, if the eyes are
In this way, the brain compensates for the retinal
going to move 10° to the right, information about the
impending saccade must be available to visual areas,
f,
displacements caused by eye movements, producing
initiating a transient 10° shift in receptive field
s
a stable representation of objects in the visual world.
locations. The current experiments investigate the
This dynamic process, known as spatial updating,
is the focus of the present study.
circuitry supporting the communication of these
1
visual and oculomotor signals. In the following
a
I
189
sections, we "describe the rationale for these experi-
path for the interhemispheric transfer of both visual
ments and our specific hypotheses. We then present
and oculomotor signals during spatial updating. The
corpus. callosum, with roughly half a bilion fibers,
gical studies
behavioral and physiological evidence that reveals an
intriguing dissociation between pathways that med-
into neural
iate the communication of visual as compared to
tlomenon of
spheric communication (Lamantia and Rakic, 1990;
corollary discharge signals are also presented.
Houzel et aI., 2002), and the anterior commissure
~ studies in
:hat several
1 representa-
frontal, and
r collculus,
tivity, which
action upon
i ve classical
)pear within
re, however,
'ield onto a
though no
: field (Mays
iruce, 1990;
995; Umeno
and Colby,
'esponse to a
1, which has
~ movement.
I representa~ movements
ion of visual
:s move, the
d from neu-
efore the eye
the location
iberg, 1999).
of the motor
~nal. Recent
Is contribute
ial behavior
ilery, 2003;
fating is one
rge informa-
constitutes the most prominent route for interhemi-
provides an immediate link between virtually all
Both visual and motor signals must be
communicated between hemispheres
One of the most noteworthy aspects of remapping is
that, at the time of the eye movement, neurons are
responsive to locations outside their classical receptive fields. Accordingly, neurons must have access to
information from throughout the visual field, even
from the opposite visual hemifield. In the original
experiments on remapping in the lateral intraparietal
cortex (LIP), stimulus representations were updated
from one visual hemifield to another (Duhamel et aI.,
1992a). This neural activity has a behavioral comple-
ment: both humans. and monkeys are capable of
performing spatial tasks that require across-hemifield
remapping (Goldberg et aI., 1990; Duhamel et aI.,
1992b; Li and Andersen, 2001; Jeffries et aI., 2003;
Zivotofsky et aI., 2003). Successful across-hemifield
updating must require a transfer of information
between neurons in opposite hemispheres, as the
representation of visual stimuli is highly lateralized
(Trevarthen, 1990; Medendorp et aI., 2003; Merriam
et aI., 2003). Similarly, physiological and behavioral
studies indicate that corollary discharge signals must
also be transferred between hemispheres. Oculomotor
signals, like visual signals, are highly lateralIzed. Yet
neurons in area LIP exhibit updating activity regardless of saccade direction (Heiser and Colby, 2003),
and updating behavior is accurate when a saccade
into one hemifield initiates updating within the
opposite hemisphere (Heide et aI., 1995). What pathways proVide the substrate for these signals to travel
between hemispheres?
visual areas in the temporal lobes (Jouandet and
Gazzaniga, 1979; Demeter et aI., 1990). Of particular
interest for spatial updating are the extensive callosal
connections between parietal cortices in each hemisphere, and between parietal cortex and areas in the
frontal
lobe (Pandya and Vignolo, 1969; Hedreen and
Yin, 1981; Seltzer and Pandya, 1983; Petrides and
Pandya, 1984; Schwartz and Goldman-Rakic, 1984).
These direct corticocortical connections could support the rapid relay of visual and oculomotor signals
required to influence receptive field properties in
conjunction with saccades. The importance of the
forebrain commissures is further suggested by neuropsychological evidence of their functional role.
Studies of split-brain humans and monkeys have
demonstrated the necessity of the corpus callosum
and anterior commissure for the across-hemisphere
visual and visuomotor processes (Gross
et aI., 1977; Holtzman, 1984; Gazzaniga, 1987; Eacott
and Gaffan, 1989; Trevarthen, 1990; Desimone et aI.,
1993; Corballs, 1995). In light of this anatomical and
integration of
behavioral evidence, we reasoned that the forebrain
commissures are critical for interhemispheric transfer
of the signals involved in spatial updating.
We hypothesized that the forebrain commssures
are necessary for communicating both visual and
corollary discharge signals from one hemisphere to
the other. In Part I, we asked whether the forebrain
commissures are required when visual representations
must be updated from one hemifield to the other. In
Part II, we asked whether these commissures are
required when spatial updating within a single hemi-
field is initiated by a saccade into the opposite
hemifield.
the eyes are
on about the
visual areas,
Hypothesis: Forebrain commissures are necessary
for communication between hemispheres during
~eptive field
spatial updating
Approach
on of these
The forebrain commissures - the corpus callosum
We tested these hypotheses by measuring the behavioral and neural correlates of spatial updating in two
1e following
and anterior commissure - provide the most obvious
rhesus macaques whose forebrain commissures were
vestigate the
190
surgically transected. We measured spatial behavior
A. Double-step sequence
using the double-step task, a classic method for
assessing subjects' ability to localize targets after an
T2\
intervening saccade (Hallett and Lightstone, 1976;
Mays and Sparks, 1980; Goldberg and Bruce, 1990).
The subject must make eye movements to two
successively flashed targets, Tl and T2 (Fig. lA).
The critical feature of this task is that the second
FP
~ T1
target (T2) disappears before the eyes leave the initial
fixation point. If the subject generates the sequence
based only on the retinal location of the T2, the
second saccade wil be incorrect (Fig. lB). For
B. Retinal direction of T2
accurate performance of the sequence, the location
of T2 must be updated in conjunction with the
saccade to Tl (Fig. iC). Subsequent to behavioral
testing, we asked whether neurons in parietal cortex
are active when remapping requires the communication of either visual or corollary discharge signals
between hemispheres.
We evaluated the integrity of spatial updating in
three conditions of the double-step task, illustrated in
Fig. 2. (1) Iú the within condition (Fig. 2A), both
visual and corollary discharge signals are communicated within the same hemisphere. The second target
(T2) must be updated from one location to another
within the same visual hemifield. This condition
therefore requires a transfer of visual information
between neurons in the same cortical hemisphere.
Furthermore, the initiating saccade (the saccade to
the first target, TI) is directed into the same visual
field in which T2 is updated. As a result, the corollary
discharge ,signal is generated by the same hemisphere
in which the transfer of
visual information occurs. We
compared updating in the within-condition to updat-
FP
/
C. Motor direction of T2
\
T1
Fig. 1. Performance of the double-step saccade task requires
ing in two interhemispheric conditions. (2) In the
spatial updating. (A) The double-step sequence. Subjects make
across-hemifield condition (Fig. 2B), T2 is updated
two consecutive saccades, to the first target (T1) and then to the
from one visual hemifield to the other. This condition
second target (T2). The second target appears very briefly, and
is also referred to as the visual-across condition, to
so is visible only when the eyes are at initial fixation (FP). When
the eyes are at fixation, the retina11ocation of T2 is up and to
emphasize that the visual representation ofT2 must
be updated across hemifields. (3) In the motor-across
T1, however, it wil be inaccurate. For accurate completion of
condition (Fig. 2C), like the within condition, the
the sequence (C), the representation of T2 must be updated to
representation ofT2 is updated within the same visual
hemifield. The critical difference is the direction of
the
saccade that initiates spatial updating. In the motor-
take the saccade to TI into account.
the right (B). If the subject generates this retinal saccade from
across condition, the initiating saccade to Tl is
the saccade command, to the hemisphere in which the
directed into the opposite hemifield. The corollary
T2 representation is updated. This condition is
discharge signal therefore must be communicated
interhemispherically, from the hemisphere generating
referred to as motor-across to emphasize that spatial
updating requires an interhemispheric transfer of
191
A
oculomotor signals. We predicted that spatial updat-
Within
ing in the split-brain monkey would be severely
T2
FP_
./
T1
disrupted if not abolished in the visual-across and
motor-across conditions, but not in the within-
condition.
We designed the behavioral paradigms to incorporate controls for sensory, motor, and cognitive
factors, as all training and testing were necessarily
conducted after the commissurotomy to prevent
infection. Once healing was complete, the animals
were trained to perform the double-step task. In the
first stage of training, we used vertical sequences, in
B
which the first saccade was either straight up or
VI",'-acroM I
T2 "
FP-T1
straight down. Updating was therefore always withinhemifield. The monkeys were trained to perform
interleaved vertical sequences at a minimum criterion
of 75% correct, demonstrating a generalized understanding of the task. In the second stage of training,
monkeys learned to perform a central condition in
which the first saccade was horizontal (Fig. 3A, black
lines). In the central sequences, T2 appeared directly
above Tl, so that the updated representation of T2
was available bilaterally and performance did not
C
Motor-across
__T2
T1_FP
require interhemispheric transfer. Once the monkeys
reached criterion on these sequences, we simulta-
neously introduced two novel test conditions: either
within and visual-across (Part I) or within and motoracross (Part II). In each case, the conditions were
matched in saccade amplitude and novelty, and
counterbalanced for direction of the second saccade
(e.g., Fig. 3A). Further, the sequences were randomly
interleaved, so that the monkeys had to rely on an
updated visual representation to complete each triaL.
c requires
ects make
hen to the
riefly, and
ip). When
up and to
::de from
p1etion of
ipdated to
,hich the
iition is
it spatial
.nsfer of
Fig. 2. Comparison of double-step conditions used to determine whether the forebrain commissures are required for
This design isolated the difference of
interest: accurate
interhemispheric transfer of visual and corollary discharge
the monkey's task is to make a
visually-guided saccade to Tl, followed by a memory-guided
saccade to T2. (A) In the within condition, T2 appears in the
right visual field when the eyes are at FP. Its retinal
location is
signals. In each condition,
represented by neurons in the left hemisphere (orange T2).
When the eyes reach T1, T2 itself is gone, but a memory trace
of T2 is stil in the right visual field, encoded by neurons within
the left hemisphere (yellow T2'). Updating therefore involves a
transfer of visual signals between sets of neurons located within
the same cortical hemisphere. The saccade that initiates this
reach TL. Consequently, updating in this condition involves a
transfer of visual information between sets of neurons in
opposite cortical hemispheres. (C) In the motor-across condition, T2 is updated within the same hemisphere, just as in panel
A. The motor-across condition is distinguished by the direction
of the initiating saccade. This leftward movement to Tl is
generated by the opposite hemisphere (white arrow).
Consequently, the corollary discharge signal from the right
hemisphere must be relayed to visual areas in the left
transfer - a rightward saccade - is also generated by the left
hemisphere (white arrow). (B) In the visual-across condition,
hemisphere. It was expected that performance of the visua1-
T2 appears in the right visual field when the eyes are at FP, but
condition, would be impaired in the absence of the forebrain
its memory trace is located in the left visual field once the eyes
commissures.
across and motor-across conditions, but not the within
~
T~2
192
A
T1 FP T1
B First ten trials
EM~lh~AJ~d
CHth~LhdJdJdJ
Central
Within
c
Visual-
Visual-
across
across
Central
Within
First session
20
Oì
Q)
:s
Monkey EM
Monkey CH
. .)
oui0.10
Q)
~
Q)
-¡
c.
:e
Q)
;:
-10
-20
o
-10
o
Horizontal eye position (deg)
o
First session
Angular error
EM CH
30
Distance error
Latency
EM CH
EM CH
c.
Q)
CI
E
\.LL~J..
\.LL~J..
o
\.LL~J.. \.LL~J..
Fig. 3. Initial impairment of visual-across sequences. (A) The six randomly interleaved sequences of the double-step task: trained
central sequences (black), novel within-hemifie1d (green) and visual-across (red) sequences. (B) Eye traces show double-step
performance in the first ten trials of the first testing session, for monkey EM (top row) and monkey CH (bottom). Individual panels
show the eye path for each sequence, in degrees of visual angle; scale bar represents 10°. Dots indicate the locations of the central
fixation point, T1 and T2. The monkeys accurately performed the central and within conditions but demonstrated substantial
impairment on the visual-across condition. (C) Second-saccade (S2) endpoints from the entire first testing session. For monkey EM
(left), impairment on the visual-across condition persisted in both visual fields, throughout the first session. For monkey CH (right),
performance of the visual-across sequence improved during the first session in the left but not the right visual field. (D) Quantitative
measures of double-step performance in initial testing sessions. Each bar represents the mean value (:lSE) of error or latency for
one of the six sequences of the double-step task. In each panel, the first six bars are from monkey EM, second six from monkey CR.
Bars are arranged according to the sequence locations (icons below). Black, central; green, within; red, across. Asterisks indicate
significantly greater error or longer latency for the visual-across sequence as compared to matched central and within sequences,
.-.;
r
193
double-step performance required a transfer of information either within or across hemispheres. We first
asked whether spatial behavior was impaired when
updating involved an interhemispheric transfer of
visual information. Could these split-brain monkeys
perform double-step sequences that required updating from one visual hemifield to the other?
on updating condition (central, within, or visualacross) or direction of the first saccade (right or left).
Of greatest interest was the prospect that individual
sequences of the visual-across condition were significantly impaired. Accordingly, we used post hoc
analyses to compare performance of each visualacross sequence to that of three matched control
sequences: the central sequence in the same visual
field (matched for the first saccade), the within
Part I: The forebrain commissures are the primary
path, though not the only path, for interhemispheric
sequence in the same visual field (matched for novelty
and the first saccade), and the within sequence in the
transfer of visual signals during spatial updating
opposite visual field (matched for novelty and the
Behavioral correlates of visual-across updating
We found that both monkeys exhibited a striking and
selective deficit when performance of the double-step
task required across-hemifield updating of the visual
representation. This impairment is evident in the first
ten trials of the first testing session (Fig. 3B). The
HSD, p ~ 0.05). These data indicate a deficit in the
split-brain monkeys' ability to update spatial loca-
sequences, as expected. Berformance of the within
tions across visual hemifields.
condition was also accurate, even though the
Three supporting lines of evidence demonstrate
sequences were noveL. In contrast, both monkeys
that this visual-across impairment is specific to
made inaccurate eye movements on every trial of the
first ten visual-across sequences. Saccade endpoints
disrupted updating in the absence of the forebrain
testing demonstrate the
commissures intact can perform these sequences
visual-across impairment (Fig. 3C). For the central
(black) and within sequences (green), endpoints are
accurately (Li and Andersen, 2001; Jeffries et aI.,
clustered near the correct T210cations. For the visualacross sequences (red), however, most endpoints are
clustered inaccurately near the central target location.
These endpoint data also reveal an unanticipated
finding: the beginning of recovery is already evident in
the left hemifield of monkey CH.
the split-brain monkeys were not selectively impaired
We assessed the monkeys' initial double-step
performance by analyzing the accuracy and latency
ub1e-step
ial panels
ie central
Ibstantia1
nkey EM
H (right),
antitative
tency for
increased reaction time, or both (Fig. 3D; Tukey's
monkeys were very accurate on the trained central
from the entire first session of
:: trained
second saccade). If performance of the visual-across
sequence was significantly worse than each of the
matched controls, we concluded that the impairment
reflected a deficit in spatial updating. We found
significant impairment for each of the individual
visual-across sequences, manifest in increased error,
commissures. First, humans and monkeys with the
2003; Zivotofsky et aI., 2003). Second, we found that
on single memory-guided saccades to the visualacross T2 locations. Thus, the visual-across impair-
ment could not be attributed to any sensory,
mnemonic, or motor deficits. Third, we determined
that the monkeys could readily perform comparable
double-step sequences when T2 was placed directly on
the midline, and therefore was represented bilaterally.
of the second saccade for the entire first session of
testing (~200 trials per condition, per monkey). We
quantified accuracy using two measures: (1) angular
The monkeys' success in this midline paradigm
error, the angular offset between the actual and target
tion of the second eye movement. These data support
demonstrates that the initial visual-across deficit did
not reflect a general diffculty in reversing the direc-
trajectory, and (2) distance error, the distance
the conclusion that across-hemifield spatial updating
between the saccade endpoint and T2. Saccadic
is disrupted in the absence of the forebrain
latency was computed as the time between the end
commissures.
Despite this initial deficit, both monkeys were able
of the first saccade and the beginning of the
to learn to perform the visual-across sequences.
, indicate
second saccade. We conducted two-way ANOV As,
separately for each monkey, to determine whether
iences.
accuracy and latency measures depended significantly
as demonstrated by the first-session endpoints of
nkey CH.
Improvement occurred quite rapidly in some cases,
194
A
Final session
20
Monkey CH
Monkey EM
Oì
Q)
:2
ui
0
0.
;.
Q)
Q)
10
ro
t;:0
Q)
-10
0
10
-20
20
B
-10
0
10
20
Horizontal eye position (deg)
Horizontal eye position (deg)
Final session
6
30
EM
CH
EM
CH
0
Q)
Ol
Q)
rJ
-0
E
.
0
'-LL~J-l
'-LL~J-l
'-LL~J-l
0
'-LL~J-l
'-LL~J-l
Fig. 4. Visual-across sequences can be learned. S2 endpoints (A) and quantitative data (B) show improved performance for visualacross sequences following multiple testing sessions (64 sessions for monkey EM, 27 sessions for monkey CH).
monkey CH (Fig. 4A). In the left visual field, many
endpoints are clustered near the correct T2 location.
These reflect the monkey's accurate performance,
which emerged during the first 75 trials of this
sequence. After multiple sessions of testing, performance of both visual-across sequences was relatively
accurate for both monkeys (Fig. 4B). These findings
indicate that, although the forebrain commissures
serve as the primary route for interhemispheric
updating, they are not the sole rOute.
neurons are found in the lateral intraparietal area
(LIP). We considered two possibilities. One is that
across-hemifield updating in the split-brain animal
is accomplished using circuitry entirely outside of
parietal cortex. If this were the case, we would expect
to observe no remapping activity in area LIP.
Alternatively, neurons in parietal cortex might stil
be an integral component of the circuitry for the
interhemispheric transfer of visual signals. In this
case, we would expect LIP neurons to exhibit remapping activity for visual-across conditions. This second
possibility seemed more probable in the light of
Physiology of visual-across updating
evidence that parietal cortex is necessary for accu-
rate spatial behavior in the double-step task
The monkeys' ultimate success in performing the
visual-across sequences implies the existence of
neurons that update visual representations across
Andersen,
hemispheres, even in the absence of the forebrain
area LIP in these same monkeys during the single-step
(Duhamel et aI., 1992b; Heide et aI., 1995; Li and
2001).
We tested these possibilities by recording from
commissures. In the second stage of the experiment,
task. This task reveals the neural activity associated
we used single-unit recording to ask whether such
with updating a visual location when the eyes move
195
992a). In each trial, the monkey
(Duhamel et aI., 1
makes a single saccade, bringing the neuron's receptive field onto a location where a stimulus has recently
appeared (Fig. 5A). Critically, no physical stimulus
ever appears in the receptive field. Rather, the neuron
can be driven only by a memory trace of the stimulus,
which has been updated in conjunction with the eye
movement. We recorded from single neurons in LIP
during two conditions of the single-step task, within
and visual-across.
significantly greater for the within as compared to
the visual-across condition (Fig. 6A). Second, do the
two conditions differ in the timecourse of neural
activity? The latency of remapping was significantly
delayed for visual-across as compared to withinconditions (Fig. 6B). These data show that visual
representations can be updated from one hemisphere
to the other in the absence of direct cortico-cortical
lInks. The forebrain commissures, then, are not
the sole mediators of across-hemifield updating.
We found robust neural activity in area LIP for
Nevertheless, the diminished strength and delayed
within-hemifield updating. The neuron shown in Fig.
5 exhibited a strong burst of activity even before the
onset of the eye movement and continued to fire after
evidence that the forebrain commissures are the
completion of the saccade. Remarkably, this same
latency of visual-across remapping provide clear
predominant pathway for updating visual representations across hemispheres.
neuron also fired for visual-across updating, though
this activity began later and was less robust than
within-hemifield activity. This neuron did not
respond in any of corresponding control conditions.
visua1-
It did not fire when the stimulus was presented alone
while the animal fixated centrally (Figs. 5G, H), nor
when the animal generated the saccade alone, with
no stimulus presented (Figs. 5J, K). The activity
observed during the single-step task can be attributed
only to remapping the memory trace of the flashed
stimulus. Thes'e data demonstrate that neurons in
area LIP can stil participate in across-hemifield
updating, despite the absence of the primary link
I area
between the cortical hemispheres.
Part II: The forebrain commissures are not the
primary path for interhemispheric transfer of
motor signals
Behavioral correlates of motor-across updating
Our findings indicate that the forebrain commissures
indeed serve as the primary route for transferring
visual signals between the cortical hemispheres at the
time of an eye movement. Are these same commissures also the primary route for relaying information
about the impending eye movement, in order to
) that
At the population level, as in the single-unit
initiate visuospatial updating? We addressed this
nimal
example, remapping signals were present but reduced
de of
for the visual-across condition. We assessed the
question by testing the monkeys on a configuration
that allowed us to compare performance of
the within
condition to the motor-across condition (Fig. 7 A). In
the motor-across condition, the representation of T2
is updated in the same hemifield, and thus the transfer
:xpect
LIP.
updating activity of 223 visually-responsive LIP
it stil
ning at saccade onset). None of the neurons
)r the
responded in the stimulus-alone task, though some
of visual information is within-hemisphere. The
ri this
neurons exhibited a response in the saccade-alone
corollary discharge signal that initiates the updating,
~map-
task, lIkely due to remapping of the central fixation
however, is thought to arise in the opposite hemi-
econd
~ht of
accutask
point. We adjusted for this activity by computing
the average firing rate in the identical epoch of the
sphere. The conditions of interest, motor-across and
within, were introduced after the monkey reached
criterion of 75% correct on the central training
sequences. We asked whether the monkeys were
j and
from
e-step
cIated
move
neurons during a standard epoch (0-200 ms, begin-
saccade-alone control, and subtracting it from the
average activity in the single-step task. This adjusted
firing rate was computed identically for all conditions
impaired selectively on the motor-across sequences.
and represents the activity attributed to updating of
We found that performance of the motor-across
the memory trace. We used this adjusted firing rate to
ask two questions. First, is updating activity equally
sequences was relatively unimpaired, as shown by the
eye traces from the first ten trials (Fig. 7B). Both
strong for visual-across and within conditions?
monkeys performed this sequence effortlessly, with
We found that the magnitude of remapping was
one exception. Monkey EM made large errors in the
196
A. Within
FP1
l
B. Visual-across
o
C. Motor-across
FP10
FP2'- FP1 0
¡.
.
-
FP2
FP2
a 't+ Ht+
, :.,Jf.
E
Single-step D
F
beginning of saccade
Stimulus alone
G
H
.'
stimulus onset
Saccade alone
J
K
L
"
.A
beginning of saccade
200 ms
Fig. 5. Activity of a single neuron in the single-step and corresponding control tasks. Top panels show the spatial configurations for
the within (A), visual-across (B), and motor-across (C) conditions. Spatial configurations are determined by the neuron's receptive
field, located in the upper right quadrant; the neuron under study was located in the left hemisphere. Cartoons ilustrate the presumed
communication of signals required for spatial updating in each condition. The neuron fired briskly for all three conditions of the
single-step task (D-F). The corresponding control conditions show that activity was minimal when the stimulus appeared alone (0-1)
and when the saccade was generated in the absence of the stimulus (J-L). In each panel, the histogram shows summed activity in 18 ms'
bins. Rasters represent individual trials; each tic mark is a single action potentiaL. The vertical bar to the right of F indicates a firing
rate of 40 spikes per second.
ii
197
I
A
first few motor-across trials in the right visual field.
The monkey nevertheless learned this sequence
rapidly, as is evident in the saccade endpoints from
the entire session (Fig. 7C). Endpoints for the motoracross condition are clustered near the correct T2
N
E.
E ~25
C Q)
location for all sequences, indicating that both
~ OJ
C
.¡:
animals were readily capable of performing the
motor-across sequences as well as the within
i.
sequences.
The monkeys' overall success in performing the
motor-across condition is evident in the measures of
25
o
50
Firing rate (Hz)
saccadic accuracy and latency (Fig. 7D). ANOV As
revealed a significant effect of updating condition for
Visual-across
both monkeys, for both measures of accuracy (all
B 300
p ~ 0.001). The pattern of conditional differences,
~200
.
U)
S
C il
.- U)
É § 100
$_~
.
.
.
experiments, asking whether the accuracy of each
motor-across sequence was significantly worse than
the accuracy of the matched central and within
Q)
o
...
-1~~00
sequences. There was significant impairment for
.
o 100 200
n = 74
300
neural onset (ms)
Visual-across
Fig. 6. Firing rate and neural latency for visual-across as
monkeys' accurate performance, the reaction times
firing rate during the same saccade-aligned epoch of the
saccade-alone control task. Points fallng along the unity line
indicate that both single-step conditions elicited the same
magnitude of remapping activity. Most points fall above the
ie (G-I)
II 18ms'
a firing
(Fig. 7B). The remaining motor-across sequences
were not significantly less accurate than their matched
within-hemifie1ds as compared to across-hemifie1ds. For each
neuron, mean firing rate in the visual-across condition (x-axis)
(y-axis). Firing rate was computed for each neuron using a
s of the
monkey EM had initially performed incorrectly
counterparts.
200 ms epoch, which began at saccade onset; mean firing rate
during the single-step task was adjusted by subtracting mean
eceptive
:esumed
only one of the motor-across sequences (Fig. 7D).
This was the sequence in the right field, which
compared to the within condition. Each point represents a
single cell. (A) LIP neurons fire more strongly for updating
is plotted against mean firing rate in the within condition
ions for
both the central and motor-across conditions. We
conducted post hoc analyses as for the visual-across
::
C
however, did not reflect an overall impairment of the
motor-across sequence. Rather, overall error values
were increased for the within condition relative to
line, indicating that neurons fired more strongly for withinhemifie1d as compared to visual-across updating. (B) LIP
neurons exhibit earlier remapping for the within as compared to
the visual-across condition. For this analysis, we included only
those neurons that met the following two criteria: first, the
latency was definable for both the within and visual-across
conditions; second, there was no significant activity in either
control condition. Most points fall below the line, indicating
that the onset of remapping activity occurred later for the
visual-across condition.
We considered the possibility that, despite the
for the second saccade might stil be slowed for the
motor-across condition as compared to the within
condition. It was found that latencies for the motoracross sequences were either equivalent to, or faster
than, those of the controls; none were significantly
prolonged relative to matched central and within
sequences (p :; 0.05, Tukey's HSD). This finding, in
concert with the accuracy data, indicates that performance of the motor-across double-step task is only
minimally disrupted in the absence of the forebrain
commissures.
Why was overall performance better for the
motor-across sequences than the visual-across
sequences? The most parsimonious explanation is
that the transfer of motor signals, unlike that of
visual signals, is not typically accomplished via the
i
198
T2~T2
T2 T2
A
T1 FP T1
:+,~ I L I FI':tr~ I j I ~I
J
°+__1 LI ~ ~ I) I~I
Within
"
;1
I '.I
Ii
II
c
Within
Central
Motor- Motoracross across
Central
First session
20
Monkey CH
Monkey EM
Õì
Q)
~
ui
o0. 10
;,
Q)
Q)
-æ
u
:¡
~ 0
-20 -10 0
10 20 -20 -10
20
Horizontal eye position (deg)
First session
D
30
EM CH
Latency
Distance error
Angular error
EM CH
6
250
EM CH
OJ
Q)
"0
o
o
..L..-.J.. ..L..-.J..
Fig. 7. Initial performance of visual-across sequences. (A) The six randomly interleaved sequences of the double-step task: trained
central sequences (black), novel within (green) and motor-across (blue) sequences. (B) Eye traces show that performance of motoracross sequences was relatively unimpaired as compared to within sequences. Individual panels show the eye path, in degrees of visual
angle, for the first ten trials of each sequence; conventions as in Fig. 3. Monkey EM made initial errors in the motor-across condition
in the right visual field, but began to adjust the trajectory toward the target as the trials progressed. (C) Second-saccade endpoints from
the entire first testing session for the motor-across condition. (D) Quantitative measures of double-step performance in the initial
motor-across testing session. Each bar represents the mean value (:JSE) of error or latency for one of the six sequences of the doub1estep task. In each panel, the first six bars are from monkey EM, second six from monkey CR. Bars are arranged according to the
sequence locations (icons below). Black, central; green, within; blue, motor-across. Asterisks indicate significantly greater error or
longer latency for the motor-across sequence as compared to matched central and within sequences.
199
forebrain commissures. Before reaching this conclusion, however, we needed to rule out an alternative
explanation, which emerged from the use of different
configurations for the motor-across and visual-across
testing. The motor-across sequence may have been
easier due to the different spatial location of T2 or
possibility by employing a new spatial configuration,
the metrics of the first saccade. We addressed this
We conducted this experiment in monkey EM,
in which the motor-across and visual-across
sequences were directed to the identical T2 location.
The sequences were also matched for the amplitude of
the first saccade, and were interleaved randomly with
the central sequences in the same session (Fig. 8A).
T2 T2
k:
A
T1
T1 FP T1
T1
First ten trials
EM~~~~~~
B
Motor-
Central
across
c
Visual-
Visual-
across
across
Central
Motor-
across
First session
Monkey EM
Cì
20
Q)
:s
..
ui
oa.
~ 10
Q)
ro
u
'E
Q)
:;
o
-20 -10 0 10 20
Horizontal eye position (deg)
First session
D
Angular error
Distance error
Latency
k: trained
of motors of visual
condition
oints from
the initial
lIe doub1e-
ling to the
:r error or
~:ii
~~~
L"LL_LJ": ~:UL
L"LL-".J": L"LL-".J":
Fig. 8. Performance of visual-across sequences is impaired when tested directly against motor-across sequences. (A) The six randomly
interleaved sequences of the double-step task: trained central sequences (black), novel visual-across (red) and motor-across (blue)
sequences. (B) Eye traces show that performance of motor-across sequences was unimpaired as compared to visual-across sequences.
Individual panels show the eye path, in degrees of visual angle, for the first ten trials of each sequence; conventions as in Fig. 3. (C)
Second-saccade endpoints from the entire testing session that directly compared visual-across and motor-across conditions.
(D) Quantitative measures of double-step performance. Asterisks indicate significantly greater error or longer latency for the visualacross sequence as compared to matched central and motor-across sequences.
200
We assessed the strength of remapping in the
who continued to exhibit visual-across impairment
in the standard paradigm prior to testing the new
motor-across condition in a population of 116 LIP
configuration.
The monkey was able to perform the double-step
motor-across conditions wered compared (Fig. 9A),
task accurately for the motor-across but not the
visual-across sequences. This dissociation is evident
in the first ten trials, in the saccade endpoint data from
the entire testing session (Fig. 8B), and in the measures
of accuracy and latency (Fig. 8D). We compared the
accuracy of individual sequences using the standard
post hoc procedure, except that each visual-across
sequence was now compared to its matched central
and motor-across (rather than within) sequences. We
found that both angular and distance error were
significantly greater for the visual-across condition,
for both visual fields. This indicates that the splitbrain monkey could accurately reach the location of
the second target when updating was within-hemi-
field, even though the saccade that initiated updating
was directed into the opposite hemifield. By contrast,
the very same target location was not attained when
updating was across-hemifield. The relative lack of
impairment for motor-across sequences suggests that
the forebrain commissures are not the primary path
for relaying information about an upcoming saccade
locations.
to cortical areas representing visual
Physiology of motor-across updating
:"
Finally, we asked whether LIP neurons are active
when updating requires the interhemispheric transfer
of corollary discharge signals. In our behavioral
experiments, we found that the monkeys were effec-
tively unimpaired when performing the motor-across
condition of the double-step task. We therefore
expected that LIP neurons would exhibit robust
updating activity in the motor-across condition of
the single-step task.
We observed significant updating activity in the
motor-across condition of the single-step task. An
example neuron is shown in Fig. 5. We previously
described this neuron's activity in the within (Fig. 5D)
neurons (Fig. 9). We first compared the within and
plotting the average firing rates against one another
and asking whether activity was greater for the within
condition (y-axis) than the motor-across condition (x-
axis). Most points fall near the unity line, representing
equivalent firing for the two conditions. We nevertheless found a significant diminution of activity in
the motor-across condition (average of 12.3 Hz,
compared to 13.8 Hz in the within condition, adjusted
firing rates, p ~ 0.05, paired t-test). The small difference in firing rate between these conditions (1.5 Hz on
average) indicates a slight yet systematic reduction in
remapping activity for the motor-across as compared
to the within condition. We next asked whether
activity in the motor-across condition differed. significantly from that in the visual-across condition
(Fig. 9B). We found that neural activity was significantly stronger for the motor-across condition
(12.3 Hz, compared to 8.4 Hz in the visual-across
condition; p ~ 0.0001, paired t-test). This difference is
apparent in Fig. 9B, in which a majority of points fall
above the unity line, indicating stronger remapping in
the motor-across condition.
These observations indicate that LIP neurons in
the split-brain monkey exhibit robust remapping
when corollary discharge signals must be relayed
between hemispheres: activity in the motor-across
condition was only slightly diminished relative to
the within condition. Direct comparison of the two
interhemispheric conditions - motor-across and
visual-across - demonstrated that remapping signals
in LIP were significantly stronger in the motor-across
condition. We concluded that the interhemispheric
transfer of corollary discharge signals, unlike that of
visual signals, is relatively unaffected by the absence
of the forebrain commissures.
Subcortical and cortical areas contribute to
remapping
and visual-across conditions (Fig. 5E), noting that it
exhibited remapping in both conditions, though
Transfei' of visual signals
activity was reduced in the visual-across condition.
This same neuron had strong and significant activity
Our behavioral and physiological findings indicate
in the motor-across condition (Fig. 5F).
that the across-hemispheric updating of visual
201
60
A
ig in the
representations is compromised in the absence of the
forebrain commissures. In behavioral experiments, we
found that split-brain monkeys exhibited an initial
impairment in performance of double-step sequences
that required updating across visual hemifields. In
f 116 LIP
iithin and
(Fig. 9A),
N
ie another
i:
the within
6
~
idition (x-
physiological experiments, we found that remapping
Q)
£ ~
30
activity in LIP was less robust when visual informa-
OJ
i:
'i:
presenting
We never-
tion had to be transferred across hemifields as
ü:
compared to within. These deficits indicate that the
forebrain commissures provide a principle, direct
route for visual information to be updated from one
activity in
12.3 Hz,
hemisphere to the other. Despite these clear deficits,
1, adjusted
30
iall differ-
60
however, remapping was not abolished as was
(1. Hz on
Firing rate (Hz)
expected. Instead, we observed an ultimate recovery
duction in
Motor-across
of function in spatial behavior, as measured by the
compared
double-step task. This behavioral success was paral-
60
B
f whether
leled by our discovery in subsequent recording studies
ffered sig-
that parietal neurons exhibited significant remapping
condition
when across-hemisphere transfer of visual informa-
tion was required. Additional pathways must be
vas signifi-
condition
mal-across
ifference is
. points fall
napping in
.
en
en
N
tli 6
recruited to transmit information from one hemisphere to the other.
What brain structures partiCipate in across-hemifield updating in the absence of the forebrain commissures? The superior collculus (SC) likely plays an
important role. Neurons in the intermediate layer of
0
Q)
¿ ~
0 OJ
Õ 'i:c
:: ü:
30
the SC demonstrate remapping activity (Walker et aI.,
rreurons in
1995). In the normal monkey, this activity is thought
remapping
be relayed
30
)tor-across
Firing rate (Hz)
relative to
Visual-across
of the two
Lcross and
iing signals
::tor-across
Lemispheric
like that of
he absence
to
60
to be a reflection of signals generated in parietal
cortex, which is considered to be critical for spatial
updating (Duhamel et aI., 1992b; Heide et aI., 1995;
Fig. 9. Remapping activity in the single-step task, for the
motor-across condition as compared to the within condition
(A) and visual-across condition (B). When average remapping
Quaia et aI., 1998; Li and Andersen, 2001). In the
activity in the motor-across condition (x-axis) is plotted against
more circuitous route, and imposed on the sc.
Alternatively, remapping activity in LIP may reflect
activity in the within condition (y-axis), the distribution of
points is primarily centered on the unity line. More points
fall above the line, indicating slightly higher firing rates for
the within condition (mean = 13.8 Hz for within, 12.3 Hz for
motor-across, p .c 0.05). In B, when remapping activity in the
motor-across condition (y-axis) is compared to that of the
visual-across condition (x-axis) most points fall above the unity
line (mean = 12.3 Hz for motor-across, 8.4 Hz for visual-across,
p .c 0.0001). These data show that, on average, the motoracross condition elicited remapping activity that was slightly
diminished in magnitude compared to that of the within
condition, and substantially greater than that of the visualacross condition.
split-brain monkey, the updated visual representa-
tion may still be constructed in LIP, by use of a
processes that originate in the Sc.
The SC, via the intertectal commissures, is an
obvious candidate for supporting interhemispheric
visual transfer in the absence of the forebrain
commissures (Moschovakis et aI., 1988; Munoz and
Istvan, 1998, OlIvier et aI., 1998). Yet it is one among
many structures that may participate in across-hemifield remapping. The pulvinar nucleus of the thalamus
is another such structure. The pulvinar is thought to
contribute to visual, oculomotor, and attentional
.gs indicate
functions (Robinson, 1993), and has been implicated
of visual
as a conduit for interhemispheric transfer in the
~~
III"'!
202
absence of the forebrain commissures (Corballs,
1995). Remapping has not yet been investigated in
the pulvinar, but it is interconnected with areas that
exhibit remapping (Hardy and Lynch, 1992; Lynch et
aI., 1994). Furthermore, its visual responses can be
modulated by extraretinal signals - possibly corollary discharge signals - related to saccades
(Robinson and Petersen, 1985). These findings
demonstrate that the functional properties of the
púlvinar are consistent with contributing to spatial
updating.
In contrast to our observations on transfer of visual
signals, our behavioral and physiological findings
the forebrain commissures
has only a minimal effect on the communication of
indicate that transection of
the corollary discharge signals that initiate spatial
updating. Both monkeys easily performed the condi-
main ways in which it could transmit information
tion of the double-step task that required an inter-
for remapping. First, the pulvinar may act as an
hemispheric transfer of this oculomotor information.
Likewise, neurons in area LIP had strong remapping
activity in this condition.
How are corollary discharge signals readily transmitted between the two hemispheres in the absence of
the forebrain commissures? Studies in split-brain
cortex (Benevento and Fallon, 1975; Hardy and
Lynch, 1992; Clower et aI., 2001; Stepniewska et aI.,
2000). This ascending route has long been consid-
ered as a second visual pathway for visual sensory
signals to reach the cortex (Diamond and Hall,
1969), and physiological studies have emphasized
the notion ¡¡hat it conveys cognitive signals, partic-
,~
Transfer of motor signals
The connectivity of the pulvinar suggests two
ascending link between superior colliculus and
ii
visual representations when the predominant pathways - the forebrain commissures - are absent.
humans have suggested that the disconnected hemispheres are capable of generating eye movements in
both directions (Holtzman, 1984; Hughes et aI.,
ularly those related to visual attention (Robinson,
1992). This claim is further supported by the observa-
1993; Bender and Youakim, 2001; Wurtz et aI., this
tion that hemispherectomy patients can make bidir-
volume). Second, the pulvinar may provide a
ectional saccades (Sharpe et aI., 1979). Hughes et al.
transthalamic link between cortical areas involved
in remapping. This compellng possibility emerges
postulated that, in split-brain subjects, there is either
an ipsilateral representation of oculomotor com-
from studies indicating that much of the pulvinar
mands at the cortical level, or a subcortical transfer
receives its driving input from cortical rather than
of information. If ipsilateral saccade representations
subcortical structures (Bender, 1983; Feig and
are present in the cortex of the split-brain monkey,
Harting, 1998; Guilery et aI., 2001; Van Horn and
they provide a ready explanation for the relative ease
of updating observed in our experiments. In effect,
Sherman, 2004). This connectivity has led to the
proposal'that the pulvinar is a higher-order thalamic
updating in this condition would be accomplished
nucleus: it does not simply relay information from
easily because the tránsfer of corollary discharge
subthalamic regions to the cortex, but rather, is
signals is intrahemispheric. Physiological studies
primarily involved in transmitting and modifying
complex signals between cortical areas (Guilery,
1995; Guillery and Sherman, 2002a). It is intriguing
to consider how these cortico-thalamocortical paths
provide scant evidence, however, that the cortical
eye fields represent ipsiversive saccades.
may contribute to spatial updating, both in the
charge signals required for spatial updating. One of
the most promising is the ascending path from the
normal and in the split brain.
A growing body of evidence favors a role for
subcortical pathways in relaying the corollary dis-
Our discussion of the circuitry for transferring
superior colliculus to the frontal eye field (FEF).
visual signals in the split-brain monkey has focused
Anatomical and microstimulation studies have shown
on the superior collculus and pulvinar, but this is by
no means an exhaustive consideration of possible
project to the FEF via the mediodorsal thalamus
pathways. In all likelihood, a broad network of
(Lynch et aI., 1994; Sommer and Wurtz, 1998). These
regions, both cortical and subcortical, must work
projections are predominantly ipsilateraL. In other
words, information about a rightward saccade is
together to carry out interhemispheric remapping of
that neurons in the intermediate layer of the SC
203
path:nt.
visual
ndings
issures
tion of
spatial
condil interIlation.
appmg
represented in the left hemisphere, at the level of SC
commissures - are differentially involved in the
and at the level of cortex.
communication of visual as compared to motor
Of direct relevance to our results, stimulation
studies have recently identified a crossed pathway
from the SC to the FEF. In a population of FEF
neurons receiving input from the SC, roughly 20% of
the cells received projections from the contralateral
signals in spatial updating. These commissures are
likely the primary conduit for the transfer of visual
signals from one hemisphere to the other. They are
not, however, strictly necessary. Our findings indicate
ally, in
that subcortical pathways play an important role in
the recovery of function, and are suffcient to support
across-hemifield updating when the direct corticocortical paths have been removed. In contrast, the
both the normal and the split-brain monkey. In
forebrain commissures do not appear to be the
other words, a copy of the command to make a
primary route for the transfer of motor signals in
SC (M. Sommer, personal communication). This
crossed ascending path could serve to transmit a
corollary discharge signal interhemispheric
rightward saccade - generated in the left SC - could
spatial updating. It appears that subcortico-cortical
be sent to the right FEF. This corollary discharge
circuits can readily communicate corollary discharge
command could then act upon visual representations
in the right cortical hemisphere, potentially in area
, trans-
LIP. An alternate possibility is that the corollary
ence of
t-brain
L hemi-
discharge signal generated in one SC could cross at
the level of the intertectal commissures, then travel
via the uncrossed ascending tecto
cortical pathway.
,en ts in
Either route could support the accurate updating
signals to visual areas in both cortical hemispheres.
Together, these conclusions emphasize the idea that
spatial updating is sub
served by a network of cortical
and subcortical structures, in which motor signals can
act upon visual information to create a stable
representation of the external world.
et aI.,
observed in the present study. The anatomical basis
bserva -
of corollary discharge signals is an active area of
~ bidir-
research (Guilery and Sherman, 2002b; Guillery,
References
s either
2003; Sommer and Wurtz, 2002, 2004a,b). These
signals may arise from many brain structures, both
Bender, D.E. (1983) Visual activation of
r com-
cortical and subcortical. Our findings indicate that
:ransfer
itations
ionkey,
ive ease
l effect,
iplished
scharge
studies
cortical
information about our impending eye movements is
:s et al.
'ole for
uy disOne of
~om the
(FEF).
e shown
the SC
readily available to modify visual representations in
each of the cortical hemispheres.
Bender, D.E. and Youakim, M. (2001) Effect of attentive
fixation in macaque thalamus and cortex. J. Neurophysiol.,
85: 219-234.
Clower, D.M., West, R.A., Lynch, J.e. and Strick, P.L. (2001)
The inferior parietal lobule is the target of output from
Conclusion
the superior collcu1us, hippocampus, and cerebellum.
J. Neurosci., 21: 6283-6291.
The phenomenon of remapping, in which visual
representations are updated in conjunction with
saccades, demonstrates the influence of motor signals
on perceptual processing. When a command is issued
to move the eyes, a corollary discharge of this
command is sent to visual areas, inducing a transfer
of visual information from neurons that represent a
salient location before the saccade, to those that wil
represent the location after the saccade. We have used
behavioral and physiological methods to investigate
ialamus
). These
the circuitry underlying the transfer of these visual
n other
The central implication of our findings is that
direct cortico-cortical paths - the forebrain
;cade is
neurons in the primate
pulvinar depends on cortex but not collicu1us. Brain Res.,
279: 258-261.
and motor signals between hemispheres.
Colby and Goldberg (1999) Space and attention in parietal
cortex. Annu. Rev. Neurosci., 22: 319-349.
Corballs, M.e. (1995) Visual integration in the split brain.
Neuropsycho10gia., 33: 937-959.
Demeter, S., Rosene, D.L. and Van Hoesen, G.W. (1990) Field
of origin and pathways of the interhemispheric commissures
in the temporal lobe of macaques. J. Compo Neurol., 302:
29-53.
Desimone, R., Moran, J., Schein, S.J. and Mishkin, M. (1993)
A role for the corpus callosum in visual area V4 of the
macaque. Vis. Neurosci., 10: 159-171.
Diamond, I.T. and Hall, W.e. (1969) Evolution of neocortex.
Science, 164: 251-262.
Duhamel, J.R., Colby, C.L. and Goldberg, M.E. (1992a) The
updating of the representation of visual space in parietal
cortex by intended eye movements. Science, 255: 90-92.
II'
204
Duhamel, J.R., Goldberg, M.E., Fitzgibbon, E.J., Sirigu, A.
and Grafman, J. (1992b) Saccadic dysmetria in a patient with
a right frontoparietal lesion. The importance of corollary
Heiser, L.M. and Colby, e.L. (2003). Is remapping universal?
Soc. Neurosci. Abstr., 29: 386.14.
visual learning in monkeys with intact optic chiasm. Exp.
Helmholtz, H. (1866). Treatise on Physiological Optics. Dover,
New York.
Holtzman, J.D. (1984) Interactions between cortical and subcortical visual areas: evidence from human commissurotomy
patients. Vision Res., 24: 801-813.
Brain Res., 74: 348-352.
Houze1, J.C., Carvalho, M.L. and Lent, R. (2002)
discharge for accurate spatial behaviour. Brain, 115:
1387-1402.
Eacott, M.J. and Gaffan, D. (1989) Interhemispheric transfer of
Feig, S. and Harting, J.K. (1998) Corticocortica1 communica-
Interhemispheric connections between primary visual areas:
tion via the thalamus: ultrastructural studies of corticotha-
beyond the midline rule. Braz. J. Med. BioI. Res., 35:
lamic projections from area 17 to the lateral posterior nucleus
of the cat and inferior pulvinar nucleus of the owl monkey.
J. Compo Neurol., 395: 281-295.
Gazzaniga, M.S. (1987) Perceptual and attentiona1 processes
following callosal section in humans. Neuropsychologia, 25:
119-133.
Goldberg, M.E., Colby, e.L. and Duhamel, J.R. (1990)
Representation of visuomotor space in the parietal lobe of
the monkey. Cold Spring Harb. Symp. Quant. BioI., 55:
729-739.
Goldberg, M.E. and Bruce, e.J. (1990) Primate frontal eye
fields. III. Maintenance ora spatially accurate saccade signaL.
J. Neurophysiol., 64: 489-508.
Gross, e.G., Bender, D.B. and Mishkin, M. (1977)
ContributitJns of the corpus callosum and the anterior
commissure to visual activation of inferior temporal neurons.
1441-1453.
Hughes, H.e., Reuter-Lorenz, P.A., Fendrich, R. and
Gazzaniga, M.S. (1992) Bidirectional control of saccadic
eye movements by the disconnected cerebral hemispheres.
Exp. Brain Res., 91: 335-339.
Jeffries, S.M., Kusunoki, M., Cohen, I.S. and Goldberg, M.E.
(2003). Localization error in a double-step saccade task are
qualitatively explained by peri-saccadic response patterns in
LIP. Soc. Neurosci. Abstracts, 386.13..
Jouandet, M.L. and Gazzaniga, M.S. (1979) Cortical field of
origin of the anterior commissure of the rhesus monkey. Exp.
Neurol., 66: 381-397.
Lamantia, A.S. and Rakic, P. (1990) Cytological and quanti-
tative characteristics of four cerebral commissures in the
rhesus monkey. J. Compo Neurol., 291: 520-537.
Li, e.S. and Andersen, R.A. (2001) Inactivation of macaque
Brain Res., 131: 227-239.
Guillery, R.W. (1995) Anatomical evidence concerning the role
of the thalamus in corticcortical communication: a brief
review. J. Anat., 187: 583-592.
Guilery, R.W. (2003) Branching thalamic afferents link action
and perception. J. Neurophysiol., 90: 539-548.
Guillery, R.W., Feig, S.L. and Van Lieshout, D.P. (2001)
Connections of higher order visual relays in the thalamus: a
study of corticothalamic pathways in cat. J. Compo Neurol.,
438: 66-85.
Guilery, R.W. and Sherman, S.M. (2002a) Thalamic relay
functions and their role in corticocortica1 communication:
generalizations from the visual system. Neuron., 33: 163-175.
Guilery, R.W. and Sherman, S.M. (2002b) The thalamus as a
monitor of motor outputs. Phi10s. Trans. R. Sco. Lond. B
BioI. Sci., 358: 1809-1821.
Hallett, P.E. and Lightstone, A.D. (1976) Saccadic eye move=1
ments to flashed targets. Vision Res., 16: 107-114.
lateral intraparietal area delays initiation of the second
saccade predominantly from contralesiona1 eye positions in a
double-saccade task. Exp. Brain. Res., 137: 45-57.
Lynch, J.e., Hoover, J.E. and Strick, P.L. (1994) Input to the
primate frontal eye field from the substantia nigra, superior
collcu1us, and dentate nucleus demonstrated by transneurona1 transport. Exp. Brain Res., 100: 181-186.
Mays, L.E. and Sparks, D.L. (1980) Dissociation of
visual and
saccade-related responses in superior collicu1us neurons. 1.
Neurophysiol., 43: 207-232.
Medendorp, W.P., Goltz, H.e., Vilis, T. and Crawford, J.D.
(2003) Gaze-centered updating of visual space in human
parietal cortex. J. Neurosci., 23: 6209-6214.
Merriam, E.P., Genovese, C.R. and Colby, C.L. (2003) Spatial
updating in human parietal cortex. Neuron, 39: 361-373.
Moschovakis, A.K., Karabe1as, A.B. and Highstein, S.M.
(1988) Structure-function relationship in the primate superior
pulvinar neurons that project to two subregions of the
collicu1us. i. Morphological classification of efferent neurons.
J. Neurophysiol., 60: 232-262.
inferior parietal lobule in the macaque. Cereb. Cortex., 2:
Munoz, D.P. and Istvan, P.J. (1998) Lateral inhibitory
217-230.
Hedreen, J.e. and Yin, T.e. (1981) Homotopic and heterotopic
callosal afferents of caudal inferior parietal lobule in Macaca
interactions in the intermediate layers of the monkey superior
collicu1us. J. Neurophysiol., 79: 1193-1209.
Hardy, S.G. and Lynch, J.C. (1992) The spatial distribution of
mu1atta. J. Compo Neurol., 197: 605-621.
Heide, W., Blankenburg, M., Zimmermann, E. and Kompf, D.
(1995) Cortical control of double-step saccades: implications
for spatial orientation. Ann. Neurol., 38: 739-748.
Nakamura, K. and Colby, e.L.(2002) Updating of the visual
representation in monkey striate and extrastriate cortex
during saccades. Proc. Natl. Acad. Sci., 99: 4026-031.
Olivier, E., Porter, J.D. and May, P.J. (1998) Comparison
of the distribution and somatodendritic morphology of
I.
¡
I
205
iniversa1?
tectotectal neurons in the cat and monkey. Vis. Neurosci., 15:
903-922.
cs. Dover,
tells the frontal cortex. i. Oculomotor signals sent from
Pandya, D.N. and Vigno10, L.A. (1969) Interhemispheric
projections of the parietal lobe in the rhesus monkey.
I and sub;surotomy
t. (2002)
mal areas:
Res., 35:
Petrides, M. and Pandya, D.N. (1984) Projections to the frontal
cortex from the posterior parietal region in the rhesus
f saccadic
berg, M.E.
Ie task are
patterns in
~a1 field of
,nkey. Exp.
,nd quantiires in the
thalamus. J. Neurophysiol., 91: 1381-1402.
monkey. J. Compo Neurol., 228: 105-116.
stem tells the frontal cortex. II. Role of the SC-MD-FEF
pathway in corollary discharge. J. Neurophysiol., 91:
1403-1423.
Quaia, c., Optican, L.M. and Goldberg, M.E. (1998) The
Stepniewska, I., Qi, H.-X., Kaas, J.H. (2000) Projections of the
maintenance of spatial accuracy by the perisaccadic remap-
superior collicu1us to subdivisions of the inferior pulvinar
visual receptive fields. NeuraL. Netw., 11: 1229-1240.
in New World and Old World monkeys, Vis. Neurosci.,
Robinson, D.L. (1993) Functional contributions of
the primate
pulvinar. Prog. Brain Res., 95: 371-380.
mispheres.
superior colliculus to frontal eye field via mediodorsal
Sommer, M.A. and Wurtz, R.H. (2004b) What the brain
Brain Res., 15: 49-65.
ping of
R. and
Sommer, M.A. and Wurtz, R.H. (2004a) What the brain stem
17: 529-549.
Trevarthen, C. (1990) Integrative functions of the cerebral
Robinson, D.L. and Petersen, S.E. (1985) Responses of
commissures. In: Boller, F.G.J. (Ed.), Handbook of
pulvinar neutrons to real and self-induced stimulus move-
Neuropsychology, Elsevier Science Publishers, B.V.
ment. Brain. Res., 338: 392-394.
Schwartz, M.L. and Goldman-Rakic, P.S. (1984) Callosal and
intrahemispheric connectivity of the prefrontal association
cortex in rhesus monkey: relation between intraparietal and
principal sulcal cortex. J. Compo Neurol., 226: 403-420.
Seltzer, B. and Pandya, D.N. (1983) The distribution of
posterior parietal fibers in the corpus callosum of the rhesus
monkey. Exp. Brain Res., 49: 147-150.
Sharpe, J.A., Lo, A.W., and Rabinovitch, H.E. (1979) Control
of the saccadic and smootJ¡ pursuit systems after cerebral
hemidecortication. Brain, 102: 387-403.
(Biomedical Division), pp. 49-83.
Umeno, M.M. and Goldberg, M.E. (1997) Spatial processing in
the monkey frontal eye field. i. Predictive visual responses.
J. Neurophysiol., 78: 1373-1383.
Umeno, M.M. and Goldberg, M.E. (2001) Spatial processing
in the monkey frontal eye field. II. Memory responses.
J. Neurophysiol., 86: 23442352.
Van Horn, S.c. and Sherman, S.M. (2004) Differences in
projection patterns between large and small corticothalamic
terminals. J. Compo Neurol., 475: 406-15.
Walker, A.V., Fitzgibbon, E.J. and Goldberg, M.E. (1995)
.f macaque
Sommer, M.A. and Wurtz, R.H. (1998) Frontal eye field neu-
Neurons in the monkey superior colliculus predict the visual
the second
rons orthodromically activated from the superior collicu1us.
result of impending saccadic eye movements. J. Neuro-
isitions in a
J. Neurophysiol., 80: 3331-3335.
Sommer, M.A. and Wurtz, R.H. (2002) A pathway in primate
7.
physiol., 73: 1988-2003.
Zivotofsky, A.Z., Tzur, R., Caspi, A. and Gordon, C.R. (2003).
nput to the
brain for internal monitoring of movements. Science, 296:
Evidence for co-processing of orthogonal compared to
ra, superior
1480-1482.
co-linear saccades. Soc. Neurosci. Abstracts, 441.6.
y transneuI
f visual and
neurons. J.
wford, J.D.
I
I
in human
i
(03) Spatial
361-373.
stein, S.M.
Ii
ate superior
mt neurons.
Ii
! inhibitory
II
key superior
)f the visual
riate cortex
~6-031.
Comparison
rpho10gy of
l,i
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Related manuals

Download PDF

advertisement