Primate Rhinal Cortex Participates in Both Visual

Primate Rhinal Cortex Participates in Both Visual Recognition and
Working Memory Tasks: Functional Mapping With 2-DG
LILA DAVACHI AND PATRICIA S. GOLDMAN-RAKIC
Section of Neurobiology, Yale University School of Medicine, New Haven, Connecticut 06511
Received 24 November 1999; accepted in final form 22 January 2001
Davachi, Lila and Patricia S. Goldman-Rakic. Primate rhinal cortex participates in both visual recognition and working memory tasks:
functional mapping with 2-DG. J Neurophysiol 85: 2590 –2601, 2001.
The rhinal cortex in the medial temporal lobe has been implicated in
object recognition memory tasks and indeed is considered to be the
critical node in a visual memory network. Previous studies using the
2-deoxyglucose method have shown that thalamic and hippocampal
structures thought to be involved in visual recognition memory are
also engaged by spatial and object working memory tasks in the
nonhuman primate. Networks engaged in memory processing can be
recognized by analysis of patterns of activation accompanying performance of specifically designed tasks. In the present study, we
compared metabolic activation of the entorhinal and perirhinal cortex
during the performance of three working memory tasks [delayed
response (DR), delayed alternation (DA), and delayed object alternation (DOA)] to that induced by a standard recognition memory task
[delayed match-to-sample (DMS)] and a sensorimotor control task in
rhesus monkeys. A region-of-interest analysis revealed elevated local
cerebral glucose utilization in the perirhinal cortex in animals performing the DA, DOA, and DMS tasks, and animals performing the
DMS task were distinct in showing a strong focus of activation in the
lateral perirhinal cortex. No significant differences were evident between groups performing memory and control tasks in the entorhinal
cortex. These findings suggest that the perirhinal cortex may play a
much broader role in memory processing than has been previously
thought, encompassing explicit working memory as well as recognition memory.
The role of the medial temporal lobes in various forms of
memory function has been a subject of interest in both animal
and human studies (Murray and Mishkin 1986; Squire and
Zola-Morgan 1985, 1991). A favored approach to modeling the
human amnesiac syndrome has been through lesions of medial
temporal lobe structures in the nonhuman primate (Meunier et
al. 1993; Mishkin 1978; Murray and Mishkin 1984; ZolaMorgan and Squire 1985). Medial temporal lobe lesions most
often are associated with deficits on declarative memory tasks
such as the delayed matched-to-sample (DMS) stimulus-recognition task. However, impairment on traditional working
memory tasks, such as delayed alternation (DA), has also been
reported (Correll and Scoville 1967; Mahut 1971; Mahut and
Cordeau 1963; Zola-Morgan and Squire 1985).
Working memory has been defined as the “on-line” or “ac-
tive” maintenance of information over a short period of time
(Goldman-Rakic 1987). Declarative memory has been defined
as the recall and/or recognition of past events (Squire and Zola
1996). These two “types” of memory function are not easily
dissociable either empirically or conceptually and may instead
exist on a continuum of time and circumstance. For example,
the initial processing of a stimulus may involve short-term
maintenance of that stimulus while related associations and
functions are being accessed through long-term memories related to that stimulus. However, a distinction between these
two memory systems has traditionally been based on evidence
from human neuropsychology and lesion studies in animals.
These studies have demonstrated that damage to the prefrontal
cortex causes impairments in working memory tasks with short
delays while damage to the medial temporal lobe is associated
with impairments with longer delay periods (Alvarez et al.
1994). Working memory tasks also differ from recognition
memory tasks in placing a greater demand on “active” maintenance of information, as indicated, for example, by the higher
level of proactive interference usually observed in working
memory tasks than in recognition memory tasks.
Although a distinction between frontally mediated working
memory processes and medial-temporal-lobe-dependent longterm memory processes has been widely accepted, recent neuroimaging studies in humans have clearly shown that the
prefrontal cortex is activated during the performance of episodic memory tasks (Kapur et al. 1994; Tulving et al. 1994).
Conversely, medial temporal lobe lesions, in nonhuman primates, including selective lesions of the perirhinal cortex have
been shown to impair memory performance with relatively
short delays (Gaffan and Murray 1992; Meunier et al. 1993;
Zola-Morgan et al. 1989). To further explore this issue of
delay-dependent involvement of medial temporal lobe memory
system, we set out to examine whether the rhinal cortex,
consisting of the entorhinal and perirhinal cortex, is metabolically activated in animals performing different memory tasks
with short delay intervals.
A major difficulty in using lesions to localize memory functions is the inevitable variations in the size of the lesion and
reactive compensatory mechanisms, such as neuronal sprouting, that have been documented following lesions to medial
temporal lobe structures (Leonard et al. 1995; Lynch et al.
1972). Thus there are distinct advantages to studying the intact
Present address and address for reprint requests: L. Davachi, Dept. of
Brain and Cognitive Sciences, Massachusetts Institute of Technology,
NE-20 Rm. 343, 77 Massachusetts Ave., Cambridge, MA 02139 (E-mail:
lila@psyche.mit.edu).
The costs of publication of this article were defrayed in part by the payment
of page charges. The article must therefore be hereby marked ‘‘advertisement’’
in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
INTRODUCTION
2590
0022-3077/01 $5.00 Copyright © 2001 The American Physiological Society
www.jn.org
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
brain during performance of memory tasks as is being done in
studies of human brain imaging. Positron emission tomography
and functional magnetic resonance imagine (MRI) techniques
have shed much light on the functional anatomy of the human
brain but lack the spatial resolution to differentiate between
cytoarchitectonically distinct areas within the medial temporal
cortex, such as the entorhinal and perirhinal cortices. The
2-deoxyglucose imaging technique (2-DG), on the other hand,
has excellent spatial resolution in animal subjects and is sufficiently sensitive to differentiate cytoarchitectonic areas and
cortical laminae during the performance of different memory
tasks.
In the present study, we examined activation in the rhinal
cortex (entorhinal and perirhinal areas) while animals performed either object or spatial working memory tasks [delayed
object alternation (DOA), DA, or delayed response (DR)] or a
recognition memory task (trial-unique DMS), all with similar
short delay intervals (12–30 s). Given our previous findings
that hippocampal and thalamic structures known to be important for recognition memory also exhibit elevated local cerebral
glucose utilization (LCGU) during working memory performance (Friedman and Goldman-Rakic 1988; Friedman et al.
1990), we hypothesized that LCGU in the rhinal cortex would
also be elevated in animals performing working memory tasks
as well as recognition tasks. Also, based on the involvement of
the perirhinal cortex in tasks that require object identification
and its abundance of cortical input from visual association
cortex (Buckley and Gaffan 1998b; Suzuki and Amaral 1994;
Thornton et al. 1997), we expected that the perirhinal cortex
would be activated preferentially in tasks requiring the memory
of object features.
2591
move cardboard plaques or objects to retrieve a food reward from
either of the two wells. The training on their respective tasks progressed depending on the individual monkeys’ performance from very
short delays (1 or 2 s) until proficiency was demonstrated (85–100%
correct within a session) to longer delays with an increased number of
trials per session. This was continued over a period of weeks and
months until the desired criterion (90% correct) was achieved at the
prescribed delay in a 45- to 50-min continuous test session. The 2-DG
test date was then scheduled. The procedures for some of the behavioral tests have been described previously (Friedman and GoldmanRakic 1988; Pribram and Mishkin 1956) and are reviewed in the next
section.
Tasks
DELAYED RESPONSE. Three monkeys were trained on this task. One
food well was baited with a food reward in full view of the monkey,
and then both wells were covered with identical cardboard plaques.
The screen was lowered for 12 s to obscure the monkey’s view of the
experimenter and the testing tray. Following the delay, the screen was
raised and the monkey chose a well. For the animal to succeed, the
monkey had to choose the previously baited well. The position of the
food reward varied in a random order according to Gellerman (1933).
Four monkeys were trained on
this task. Food wells were baited with the screen lowered. On the first
trial, both wells were baited with a food reward and covered with
identical cardboard plaques, and the monkey was allowed to displace
either plaque to obtain a reward. On the following trial, only the well
that was not selected on the previous trial was baited, and subsequent
trials proceeded in this fashion. Therefore the location of the previous
reward had to be remembered over the delay (12 or 30 s) for the
monkey to select the alternate well and obtain the reward on the new
trial. Two monkeys were trained with a delay of 12 s and two with a
delay of 30 s on this task.
DELAYED SPATIAL ALTERNATION.
Three monkeys were trained to
perform this task. This paradigm required that objects as opposed to
spatial locations be remembered over the delay to achieve a food
reward. A blue cube (6.5 cm square ⫻ 3 cm high) and a green cylinder
(6.5 cm diam ⫻ 8 cm high) were the objects used to cover the food
wells. Monkeys were first taught to do a simple object discrimination
task using a criterion of 90% correct in 60 trials before reversing the
reward contingencies. Thus one object (either the blue cube or the
green cylinder) was always rewarded until the animal reached criterion, after which the other object was rewarded. The number of trials
to reversal was then gradually decreased from 60 to 30, 15, 10, and 5
trials until 1-trial alternation was achieved, and the intertrial interval
was gradually increased to 12 s. The animals were now choosing the
alternate object on every trial with the spatial position of the reward
was being randomly varied (Gellerman 1933). Once again, as in DOA,
the preceding response had to be remembered over the delay for the
monkey to select the correct object and obtain a reward.
DELAYED OBJECT ALTERNATION.
METHODS
Subjects
Seventeen male rhesus monkeys (Macaca mulatta, 2.0 – 6.0 kg)
were trained to perform either one of four memory tasks (n ⫽ 14) or
a control task (n ⫽ 3). Monkeys were housed individually and fed a
diet of monkey chow and fruit, adjusted to maintain body weight at
90% of free-feeding weight. Monkeys were given free access to water.
Water was available ad libitum. Data for the rhinal cortex of the
medial temporal lobe was obtained from 11 animals that performed
the DOA (n ⫽ 3), DA (n ⫽ 4), and DR (n ⫽ 3) tasks and the
sensory-motor task (SMC, n ⫽ 1) in our previous 2-DG studies of
LCGU in the prefrontal cortex (Friedman and Goldman-Rakic 1994)
and hippocampus (Friedman and Goldman-Rakic 1988). The DMS
(n ⫽ 4) group and two animals from the SMC group are new additions
to this experiment. All behavioral methods used in the new cases were
identical to those employed previously. The data analysis for all
animals was performed de novo and in the same manner for this study.
Apparatus and testing procedures
Monkeys were seated in a primate chair while being tested in a
modified Wisconsin General Test Apparatus (WGTA), which included a test tray with two or three food wells (depending on the task,
see following text) where stimuli and rewards were provided to the
animal and a manually manipulated screen for imposing delays. The
testing room was darkened and sound attenuated. A white-noise
generator supplied a constant level of background noise during testing
(90 dB).
Training began after the monkeys became acclimated to the chair,
testing room, and WGTA. Initially, monkeys were only required to
DELAYED MATCH-TO-SAMPLE. Four monkeys were trained to perform this visual recognition memory task. To maximize reliance on
recognition memory processes, trial unique objects were used in the
administration of the DMS task. Stimuli consisted of “junk” objects
varying in size, color, and shape. A sample object was presented over
the middle well covering a food reward. The monkey was required to
displace this object after which a delay of 12 s ensued. After this
delay, the sample object and a novel object were placed over the two
side wells and the correct choice (i.e., the object that was baited) was
the object that “matched” the sample. The novel object was not baited.
If the monkey chose the novel object, the trial was terminated, and an
intertrial interval of 15 s ensued before the beginning of the next trial.
Monkeys performing this task performed an average of 80 trials per
session and thus were exposed to an average of 160 novel objects
during the 2-DG experiment.
2592
L. DAVACHI AND P. S. GOLDMAN-RAKIC
Three monkeys were trained to perform this task in which both wells are baited and identical plaques
were placed on the testing board. Monkeys were required to displace
a plaque and retrieve a food reward. The opaque screen was lowered
for a delay of 12 s between trials, but the monkey did not have to
remember anything during this time because a reward was given on
every trial. The sensory stimuli and motor responses used in this task
were similar to those used in all other tasks.
SENSORY-MOTOR CONTROL.
tem is comprised of an NEC Express5800 computer equipped with a
200-MHz Pentium Pro processor, a high-resolution PCI graphics card,
a Dage MTI CCD72 video camera, and a color video monitor. Pixel
gray values were computed by optical density and were then translated into 14C radioactivity levels. These were then translated into
LCGU rates using the integrated plasma specific activities obtained
for each monkey (Sokoloff et al. 1977).
Data analysis
2-DG experiment
The quantitative 2-DG method developed by
Sokoloff et al. (1977) was followed. Seventeen monkeys received
arterial and venous catheters of the femoral vessels under halothane
and nitric oxide anesthesia prior to the 2-DG test (1 monkey received
a catheter into the saphonous artery). Four monkeys (2 DA and 2
DOA) were catheterized 24 h prior to the 2-DG injection to further
promote alert testing performance and were therefore given ketamine
(5 mg/kg) in addition to the gas anesthesia. In these cases, the
catheterization was done under aseptic conditions. Sterile catheters
were inserted and tied to the arteries and veins, and the exposed ends
were sealed and secured about the sutured wound with bandages for
protection. These monkeys were kept in their home cages overnight.
The free ends of the catheters were opened prior to the 2-DG experiment. The other 13 monkeys were catheterized the morning of the
2-DG experiment at least 2 h prior to testing. They received only gas
anesthesia and were allowed ample time to recover from anesthesia
before beginning the experiment and receiving the 2-DG injection.
Subcutaneous lidocaine was applied during the 24-h catheterization
experiments, and topical anesthetics were applied liberally during
both procedures. Analyses of data showed that there was no apparent
correlation between the type of catheterization and subsequent task
performance.
PREPARATION.
About 3–5 min into the test session, 14C2-DG (100 ␮Ci/kg in 1 ␮Ci/10 ␮l sterile saline, 50 – 60 mCi/mM;
American Radiolabeled Chemicals) was injected followed by a saline
flush. Blood samples were then taken over the 45-min test period
at timed intervals. At the end of the test session, the monkey was
killed with an overdose of sodium pentobarbital. The monkey was
then lightly perfused through the heart with a solution of buffered
3.3% paraformaldehyde (1.5–3.0 l, pH 7.4). The brain was quickly
removed and sectioned into blocks before immersion in cold isopentane (⫺40°C). Blocks were stored at ⫺70°C.
EXPERIMENTAL SESSION.
TISSUE PROCESSING. Brain blocks were cut at 20-␮m thickness on
a cryostat (Hacker Instruments) at ⫺22°C. In most cases, 4 of every
20 serial sections was saved throughout the entire extent of the brain.
One section in this series was mounted on a glass slide for cresyl
violet staining while the adjacent three sections were mounted on
glass coverslips and dried on a hot plate for autoradiography. These
coverslips were taped to 7 ⫻ 8-in cardboard, and the sections were
then exposed to X-ray film (SB5 or Biomax, Kodak) for 4 –10 days
together with a set of polymethylmethacrylate 14C standards (0 –1.08
␮Ci/g, Amersham). Films were then processed in developer and
fixative (GBX, Kodak) according to packaged instructions.
BLOOD GLUCOSE AND 14C LEVELS. Blood samples collected at fixed
intervals during the experiment were immediately centrifuged. Plasma
samples (20 ␮l) were then analyzed for glucose (Beckman Glucose
Analyzer 2) and for 14C concentration using a liquid scintillation
counter (Beckman scintillation counter). Integrated arterial plasma
specific activities were derived from the blood concentration curves,
and these were used to convert tissue 14C concentrations to LCGU as
described by Kennedy et al. (1978) for each monkey.
QUANTIFICATION OF AUTORADIOGRAMS. Autoradiograms of selected brain sections along with the 14C standards for every page of
film were digitized using a computerized video imaging processing
system (MCID, M2, Imaging Research, Ontario, Canada). This sys-
Data were analyzed using two methods. The first method was a
region of interest analysis (ROI), which was employed to compare
LCGU between groups over the extent of any given cortical area, in
this case, the entorhinal and perirhinal cortex. This method has been
used in previous 2-DG experiments (Friedman and Goldman-Rakic
1994) and is described in more detail in the next section. In addition,
a local maxima analysis was employed to reveal common and disparate patterns of activation within these same regions. This method
highlights areas or subregions that reach a specified criterion that can
then be compared between groups (see following text).
ROI analysis
ENTORHINAL CORTEX. The entorhinal cortex (EC) is located ventral
to the amygdala and hippocampal formation and is bounded medially
by the parasubiculum and laterally by the fundus of the rhinal sulcus
(Fig. 1). Medial EC was defined as the cortex bounded medially by the
parasubiculum and laterally by the lip of the rhinal sulcus while lateral
EC was defined as the cortex on the medial bank of the rhinal sulcus
ending in the fundus of the rhinal sulcus (Saleem and Tanaka 1996;
Suzuki and Amaral 1994). LCGU was measured in the EC of the right
hemisphere starting at the appearance of the rhinal sulcus and extending throughout its rostrocaudal extent. ROIs were sampled using an
adjustable sampling tool (a “ribbon” tool) that allowed measurement
of the entire mediolateral extent of the cortex and layers II–V regardless of the width of the cortex. Layers I and VI were not sampled
because LCGU rates were very low in these layers. Density values
were summed across the length of the ribbon tool, thus one value
representing the mean LCGU in the EC per section was obtained. Data
from animals in the DA, DR, or DOA groups were sampled by
defining ROI in three serial sections at 400-␮m intervals throughout
the entire extent of the area examined (Friedman and Goldman-Rakic
1994; Friedman et al. 1989). The cutting schedule was modified for
the DMS and two control animals. As a consequence, data from these
animals were obtained by defining regions of interest in four equally
spaced sections every 400 ␮m. An average of 73 sections, hence 73
measures, per animal was used to determine the mean LCGU per ROI.
Across groups an average of 83, 87, and 62 sections were analyzed for
the DMS, SMC, and working memory groups, respectively. One mean
LCGU rate was then calculated for the entire EC and these means
were then transferred into SYSTAT (Sherman, IL) for statistical
analysis (see following text).
The perirhinal cortex (PC) was analyzed beginning at the level of the rhinal sulcus as seen in coronal sections and
therefore the rostral portion (area 36d and rostral 35 and 36r) of this
area was not included in our analysis. Caudally, the analysis of the PC
extended to the end of the rhinal sulcus. Medial PC was defined as
cortex within the lateral bank of the rhinal sulcus and lateral PC was
defined as PC lateral to medial PC and bordering inferotemporal
cortex. The lateral border of PC was defined as the midpoint between
the lateral lip of the rhinal sulcus and the medial lip of the anterior
medial temporal sulcus (AMTS). This border was obtained by applying a straight-edge ruler to the screen displaying the image to be
sampled. It should be noted that this border is largely in keeping with
Saleem and Tanaka’s (1996) designations. In anterior sections where
the AMTS ended, the lateral border was defined as a third of the
distance from the lateral lip of the rhinal sulcus to the medial lip of the
PERIRHINAL CORTEX.
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
2593
Local maxima analysis
FIG. 1. Cresyl-violet stained sections through the antero-posterior (AP)
levels examined of the rhinal cortex. These are sections through the entorhinal
and perirhinal cortices with lines showing borders between (medial to lateral)
the entorhinal cortex and the perirhinal cortex and the perirhinal cortex and
inferotemporal cortex. Hipp, hippocampus; sub, subiculum; EC, entorhinal
cortex; rs, rhinal sulcus; PC, perirhinal cortex; TE, inferotemporal cortex.
superior temporal sulcus (see Fig. 1 for details). There is some debate
as to the true border between PC and inferotemporal cortex in the
literature. Recent anatomical findings have located this border considerably more lateral (closer to the AMTS) (Suzuki and Amaral
1994); however, other reports have located this border medial to that
(Saleem and Tanaka 1996). The exact border between PC and inferotemporal cortex was not crucial in this study, so we chose to be
conservative in our approximations primarily because LCGU is much
higher in visual association cortex and would potentially contaminate
our data collection. Thus we chose this conservative estimate of the
lateral border to ensure that we were sampling from the PC alone.1
The same sections were used for analysis of the PC as were used for
analysis of the EC. Data were collected on the same schedule and in
the same manner as described in the preceding text for the EC.
All animals were exposed to white
noise (90 dB) during the 2-DG test session to block out environmental
sounds. Because of this, the medial geniculate body was chosen to
serve as a control for individual differences in glucose consumption.
Animals received the same auditory input allowing us to use the
LCGU of this region as a covariate in our statistical analyses. Consecutive sections, comprising 2 mm in the rostrocaudal axis through
the medial geniculate body were analyzed from each animal. LCGU
for the entire structure was obtained. A mean LCGU value for the
medial geniculate body was then calculated and used in the statistical
analyses described in the following text.
MEDIAL GENICULATE BODY.
1
On the suggestion of a reviewer, we have also analyzed the data using
published criteria of the entorhinal/perirhinal boundary and the perirhinal/TE
boundary (Suzuki and Amaral 1994) for three of the five groups examined
(DA, DMS, and SMC). We found the same results using both criteria.
To obtain a more detailed view of focal regions activated within the
EC and the PC and to be certain that regional variations were not
being diluted by the ROI analysis, a local maxima analysis was
performed. This consisted of highlighting pixels on each section that
reached a specific density and size criterion. First, mean LCGU for the
rhinal cortex within a single section was determined. Pixels that were
at least 1.5 SD above this mean were then highlighted. This resulted
in hundreds of contiguous pixels being highlighted (Fig. 2B). A size
filter was then applied such that only activations that were at least 2
SD above the average size of activated pixels were included. This
cutoff was used to eliminate small activations that may be artifactual.
This yielded the most informative and manageable picture of activity
within a section (Fig. 2C). This thresholding procedure was applied to
each section independently. Thus variations in tissue thickness did not
affect the outcome. The same sections that were used for the region of
interest analysis were used for the local maxima analysis. Thus in all
an average of 73 sections per animal was analyzed for local maxima.
Local maxima were then characterized according to where they
were located, their intensity (measured by the LCGU rates), and their
size (measured by the number of consecutive sections that contained
that activation). All animals displayed activations in the fundus of the
rhinal sulcus, and these activations often crossed the cytoarchitectonic
borders of the EC and the PC, both within a section and across
sections. Since these activations spanned both the entorhinal region of
the fundus as well as the perirhinal region of the fundus, they will be
referred to as “fundus” activations instead of belonging to one or
another cortical region. The size and intensity of local maxima were
compared between groups. The peak level of glucose utilization
within local maxima was compared using an analysis of covariance.
Statistical analysis
An analysis of covariance was used to control for individual differences in general brain metabolism (Friedman and Goldman-Rakic
1988). The medial geniculate body was used as the covariate for
reasons described in the preceding text. Adjusted mean LCGU was
then derived from the analysis of covariance and is the dependent
variable used in all data presentations. The analysis was done using a
computer-based statistics program (SYSTAT, Sherman, IL).
RESULTS
General results
Performance on the 2-DG test date ranged from 73 to 100%
correct. The animals that performed memory tasks obtained a
mean of 88% correct. Within tasks, mean performance was
80% or better over the 45-min test period (Table 1). Tasks
could be ranked in order of difficulty based on the number of
trials to criterion in initial learning. By this criterion, the DOA
task was judged to be the most difficult since training required
more sessions on average than any other task and percent
correct performance was lower than any other task. By the
same criterion, the DA task was the penultimate in difficulty
followed by the DMS, DR, and SMC tasks, in that order.
General pattern of glucose uptake in the rhinal cortex
The general pattern of glucose uptake in the rhinal cortex is
illustrated in Fig. 3. Glucose uptake was always visibly lower
in the medial temporal regions compared with neighboring
cortical areas (Fig. 3, A–C). The EC was characterized by less
activation in medial portions compared with lateral portions
and LCGU increased in the fundus of the rhinal sulcus possibly
2594
L. DAVACHI AND P. S. GOLDMAN-RAKIC
FIG. 2. Local maxima thresholding method. An
example of the outcome of the thresholding method
used in this study. A: example of an autoradiographic image from 1 animal. B: same image as in A
but with pixels highlighted which meet the density
criterion [1.5 SD above the mean local cerebral
glucose utilization (LCGU) for rhinal cortex].
C: same image as in A and B but now with an
additional size criterion imposed. Only pixels which
meet a density criterion (1.5 SD above the mean
LCGU for the rhinal cortex) and a size criterion (2
SD above the mean area of contiguous pixels which
meet the density criterion) are highlighted here.
Note that over 100 contiguous pixels are highlighted
in B while only 2 are highlighted in C.
due to the compression of cortical layers. Laminar patterns of
uptake in medial EC changed somewhat in the antero-posterior
(AP) axis. Specifically, the middle cortical layers of medial EC
were more evenly activated across layers in anterior portions of
the EC, while a superficial and deep band of activity emerge at
more caudal levels. Conversely, a uniform pattern of LCGU
across cortical layers was seen in lateral EC at both anterior
and posterior levels.
Within the medial PC, the highest glucose utilization was
localized to deep layer III/IV, whereas in the lateral PC, it
spanned layers II–V. Local maxima activations (see Fig. 7 for
examples) were usually located within the middle layers of
cortex where LCGU always appeared to be the strongest. For
example, lateral PC activations were usually localized within
layers III–V in all animals.
1. Trials to criterion (85% correct in 100 trails) in initial
training percent correct and number of trials performed on the
2-DG test day
TABLE
Performance 2-DG Test Day
Task Monkey
Percent correct
Trials
DOA1
DOA2
DOA3
Mean DOA animals
DA1
DA2
DA3
DA4
Mean DA animals
DMS1
DMS2
DMS3
DMS4
Mean DMS animals
DR1
DR2
DR3
Mean DR animals
SMC1
SMC2
SMC3
Mean SMC animals
82
86
73
80
89
91
76
97
88
93
91
87
88
90
95
84
96
92
N/A
N/A
N/A
127
140
169
145.33
160
128
62
87
109.25
80
81
81
80
80.50
131
128
143
134.00
130
140
149
139.67
Initial Training
(Trials to Criterion)
1560.00
3100.00
2275.00
2311.67
410.00
1205.00
619.00
1208.00
860.50
699.00
900.00
337.00
420.00
589.00
240.00
150.00
240.00
210.00
N/A
N/A
N/A
DOA, delayed object alternation; DA, delayed alternation; DMS, delayed
match to sample; DR, delayed response; SMC, sensory-motor control; 2-DG,
2-deoxyglucose.
The only exceptions to this were in the fundus of the rhinal
sulcus where activations often spanned across more than three
cortical layers. An example of the general pattern of activation
in a magnified image of the fundus of the rhinal sulcus is
displayed in Fig. 4. The pattern seen in an anterior section
(where the fundus predominantly lies in perirhinal region area
35) and a posterior section (where the fundus lies within the
boundaries of the EC) demonstrates that regardless of whether
the fundus lies within the entorhinal or perirhinal region, it
emerges as a region of elevated LCGU. Again, it can be seen
more clearly from this view that the lateral entorhinal region is
more uniformly activated than the medial perirhinal region
(lining the rhinal sulcus) where deep layer III/IV is activated
more than the surrounding layers.
The general pattern of activity is also displayed on a flattened map of LCGU through the anterior posterior extent of the
EC and PC in a representative case (Fig. 3D). This twodimensional flattened map through the middle layers of cortex
(layers III–V) is based on 78 different AP levels to demonstrate
the pattern of glucose uptake across the anterior-posterior
dimension. Again, this figure demonstrates that anterior regions of EC were less metabolically active than posterior
portions and lateral EC is more active than medial regions
across most AP levels. Furthermore these regional differences
within the EC withstood statistical analyses showing that,
across all animals, lateral EC was more active than medial EC
(F ⫽ 5.935, df ⫽ 1, P ⬍ 0.05) and that within the lateral EC,
posterior regions were more active than anterior regions (F ⫽
7.252, df ⫽ 1, P ⬍ 0.02). Also evident in this representative
image are peaks of activity (shown in red) localized around the
fundus (in area 28 or 35 depending on which peak). Such
peaks, which were observed in all animals, are analyzed further
in the following text.
It is important to note that the general patterns of glucose
utilization displayed in Figs. 3 and 4 were observed in all
animals regardless of the type of task they were performing.
The striking similarity between control and experimental
groups in patterns of glucose uptake has been noted in previous
studies of 2-DG imaging (Friedman and Goldman-Rakic 1988,
1994). As in prior studies, group differences in glucose uptake
emerged only when we applied quantitative autoradiographic
methods to areas of interest (ROI analysis) and local maxima
analyses (see METHODS).
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
FIG. 3. General pattern of LCGU observed in rhinal cortex. A: an example of an autoradiographic image from 1 animal showing
that overall LCGU is lower in medial temporal lobe regions compared with surrounding cortical areas. Inset: higher power view
of this image showing details of the pattern of LCGU within the rhinal cortex (see RESULTS). B: pseudo-colored (according to
calibration of LCGU in top right corner) image of same section shown in A. C: cresyl-violet stained section adjacent to that seen
in A. Areas are labeled in inset. D: a flattened map through the entorhinal and perirhinal cortices of a representative animal. Sections
are flattened and stacked from most anterior (bottom) to most posterior (top) with medial to left and lateral to the right. The map
is pseudo-colored according to the calibration bar at the right representing LCGU values. Anatomical borders corresponding to
medial and lateral EC (BA 28) and medial (BA 35) and lateral (BA 36) PC are drawn on the map with white dotted lines, labeled
at the top of the map. This map was drawn based on the cytoarchitectonic borders described by Suzuki and Amaral (1994). Distance
calibrations are included at bottom and left sides of the map. Three autoradiographic sections are included at the left to give an idea
of the AP level of sections. A, amygdala; rs, rhinal sulcus; AMTS, anterior medial temporal sulcus.
2595
2596
L. DAVACHI AND P. S. GOLDMAN-RAKIC
FIG. 4. General pattern of activation on
an anterior (A–C) and posterior (D–F) section through the rhinal cortex. A and D are
autoradiographic sections displaying the AP
level of magnified sections B, C, E, and F. B
and E are magnified sections adjacent to
those displayed in A and D stained with the
Nissl stain. C and F are magnified pseudocolored images of glucose utilization in the
same sections as A and D. The highest level
of glucose utilization is displayed in red.
Dotted lines correspond to the cytoarchitectonic boundaries between entorhinal (EC)
and perirhinal (area 35 and area 36) cortices.
LCGU rates in the rhinal cortex: ROI analysis
Figure 5 displays LCGU for all groups in the rhinal cortex.
The ROI analysis revealed that animals performing the DMS,
DA, and DOA tasks exhibited significantly elevated LCGU in
the rhinal cortex as a whole compared with controls. A 2
(area) ⫻ 5 (task) analysis of covariance (ANCOVA) revealed
a significant task effect (F ⫽ 5.367, df ⫽ 4, P ⫽ 0.001 and
Fisher’s LSD post hoc comparisons confirmed DMS ⬎ SMC,
P ⬍ 0.02; DA ⬎ SMC, P ⬍ 0.01; DOA ⬎ SMC, P ⬍ 0.01).
The difference between the DR and SMC groups was not
significant.
Separate analyses performed on the EC and PC revealed that
the differences noted in the overall rhinal analysis arose out of
task differences in the PC, not the EC (Fig. 6). A one-way
ANCOVA performed on the perirhinal data showed a significant main effect of task (F ⫽ 4.705, df ⫽ 4, P ⬍ 0.01) and post
hoc comparisons confirmed that LCGU in the DMS, DA, and
DOA groups was significantly higher than the SMC group
(Fisher’s LSD, P ⬍ 0.01, 0.01, 0.02, respectively).
Although a significant task effect did not emerge in the EC
(F ⫽ 1.433, df ⫽ 4, P ⫽ 0.249), the pattern of LCGU was
similar to that observed in the PC except that the DMS group
now displayed a conspicuously low LCGU in the EC compared
with the PC, a finding confirmed statistically, i.e., a one-way
ANCOVA revealed a significant area effect: F ⫽ 5.449, df ⫽
1, P ⬍ 0.04. In addition, the DOA group was the only other
group to exhibit lower LCGU in the EC over that in the PC
although at a marginal level of significance (F ⫽ 4.41, df ⫽ 1,
P ⫽ 0.06). Interestingly, there was no significant difference in
LCGU between perirhinal and EC in the animals performing
the DA and DR spatial memory tasks.
Given the heterogeneity of activation within EC (see preceding text), with lateral regions exhibiting stronger activation
than medial regions and posterolateral regions exhibiting more
activation than anterolateral regions, separate analyses were
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
2597
3D). The local maxima observed in the lateral PC were significantly stronger in the animals performing the DMS task relative to controls and the DR and DOA working memory groups
(main effect: F ⫽ 4.05, df ⫽ 4, P ⫽ 0.03: Fisher’s LSD post
hoc comparisons revealed DMS ⬎ SMC, P ⬍ 0.01; DMS ⬎
DR, P ⫽ 0.05, DMS ⬎ DOA, P ⬍ 0.02; see Fig. 8). In
contrast, group differences were not evident for local maxima
in the fundus of the rhinal sulcus (main effect: F ⫽ 1.054, df ⫽
4, P ⫽ 0.428; see Fig. 8).2
DISCUSSION
FIG. 5. Mean LCGU in the rhinal cortex. Graph shows the adjusted mean
LCGU (␮mol 䡠 100 g⫺1 䡠 min⫺1) for each task for the entire rhinal cortex
(includes entorhinal and PC). SMC, sensory-motor control task; DR, delayed
response task; DMS, delayed match-to-sample task; DA, delayed alternation
task; DOA, delayed object alternation task. * DMS, DA, and DOA show
increased LCGU over SMC, P ⬍ 0.02.
conducted to determine if task differences were concealed by
collapsing across EC regions. These analyses confirmed that
there were no significant differences between groups in these
regions (medial, lateral, anterolateral and posterolateral EC).
However, there was a trend for the animals performing the DA
task to show increased activation in the lateral EC compared
with controls (F ⫽ 3.318, df ⫽ 2, P ⫽ 0.069) when collapsing
across both anterior and posterior regions of the lateral EC.
Implications for this trend are discussed in the following text.
Local maxima analysis
Figure 7 displays examples of local maxima in four animals
[DMS (A1–A4), DOA (B1–B4), and 2 DA (C1–C4 and D1–
D4)] at similar rostrocaudal levels throughout the extent of the
rhinal cortex. All animals displayed peak activations in the
fundus of the rhinal sulcus and in the lateral PC. These activations were evident along a considerable portion of the anterior-posterior axis of the rhinal cortex in all animals (see Fig.
FIG. 6. Mean LCGU in the perirhinal and entorhinal cortices. Graph shows
the adjusted mean LCGU (␮mol 䡠 100 g⫺1 䡠 min⫺1) for each task for the
perirhinal and entorhinal cortices separately. * DA, DOA, and DMS show
increased LCGU over SMC, P ⬍ 0.05.
Our results confirm that the PC is activated during the
performance of both working and recognition memory tasks,
whether spatial or object based. LCGU was significantly elevated in comparison to controls in the PC in animals performing the DA, DOA and DMS tasks. Our data also demonstrate
that animals performing the object tasks (DMS and DOA)
activated the PC significantly more than the EC. Furthermore,
within the lateral PC, animals performing the DMS task exhibited a distinct focus of enhanced LCGU that was greater
than that seen in controls and other memory tasks. In contrast,
the only group difference in the EC was a trend for the animals
performing a spatial working memory task to show enhanced
LCGU in the lateral EC (but not the medial EC). Finally, the
fundus of the rhinal sulcus was elevated in all animals including controls, indicating that areas within the rhinal cortex may
be activated by nonmnemonic factors common to all behavioral tasks, i.e., visual processing, attention or motivation.
These findings reveal that the rhinal cortex is functionally
heterogeneous and add to the increasing number of studies
implicating the rhinal cortex not only in visual recognition
memory but also in a broad range of other tasks including
visual–visual associations, tactile delayed nonmatch to sample,
spatial scene learning and configural learning (Murray et al.
1993, 1998; Suzuki et al. 1993). The present study adds explicit spatial working memory tasks to this list.
Correspondence with lesion studies
Although studies of postoperative performance on delayed
matching tasks following lesions to the rhinal cortex report
deficits at delays longer than those used in the present study,
impaired acquisition of these tasks with a shorter delay (8 s)
has been reported (Gaffan and Murray 1992; Zola-Morgan and
Squire 1985), and acquisition of delayed matching tasks with
short delay intervals correlates strongly with performance at
2
However, because the fundus region lies within the perirhinal cortex in
anterior sections and within the lateral entorhinal region in posterior sections,
we proceeded to reanalyze the ROI data using precise cytoarchitectonic borders described by Suzuki and Amaral (1994) (Fig. 7). The motivation behind
this analysis was that a reviewer suggested that if the fundus, a region of high
activation, was always partly included in the lateral entorhinal as well as
perirhinal cortex, then it is possible that inaccurate boundaries could be
masking a more striking pattern underlying the fundus activations. Specifically, we looked at both anterior and posterior EC in the DA, DMS, and SMC
groups. This analysis revealed a significant main effect of region (F ⫽ 7.252,
df ⫽ 1, P ⬍ 0.02) with posterior EC exhibiting greater LCGU than anterior
regions. Furthermore, there was a marginally significant effect of group (F ⫽
3.318, df ⫽ 2, P ⫽ 0.069) with a trend for the DA group to show higher LCGU
than the SMC group. This is an intriguing trend and warrants further investigation in future studies. The same type of analysis performed on anterior
versus posterior regions of medial PC did not reveal any differences.
2598
L. DAVACHI AND P. S. GOLDMAN-RAKIC
FIG. 7. Example of local maxima in the rhinal cortex at 4 similar rostrocaudal levels in four animals performing memory tasks.
A1–A4: DMS animal; B1–B4: DOA animal; C1–C4: DA animal; D1–D4: DA animal. sts, superior temporal sulcus; amyg,
amygdala.
longer delay intervals (Murray and Mishkin 1984). The present
data support these findings in showing that the rhinal cortex is
activated in animals performing the DMS with short delay
intervals (12 s). Our study further reveals that LCGU is relatively more enhanced in the perirhinal region than the entorhinal region for this task. These findings are also consistent with
lesion studies showing that lesions restricted to the PC cause as
severe a deficit in memory performance as lesions following
combined entorhinal and perirhinal lesions (Meunier et al.
1993).
Increases in LCGU above control levels were observed in
the PC in the animals performing the DA task but not the
animals performing the DR task. This result may in part
explain why a number of previous studies involving medial
temporal lobe ablations have failed to show deficits on DR
tasks (Mahut 1971; Murray and Mishkin 1986; Zola-Morgan
and Squire 1985), whereas such lesions produce impairments
on spatial DA tasks (Correll and Scoville 1967; Mahut 1971;
Mahut and Cordeau 1963; Zola-Morgan and Squire 1985). The
differential activations observed between the DA and DR tasks
suggest that the two tasks may differ in ways beyond reliance
on spatial working memory processes. In the present study,
many more trials were required to reach criterion on the DA
task than on DR (see Table 1). In contrast to DR, performance
on both DA and DOA requires continuous updating of the
location or identity of the previous stimulus similar to n-back
tasks used in humans (Smith and Jonides 1998). Their continuous nature introduces a higher level of proactive interference
than is present in DR, where each trial is independent of the
previous trial. Another significant parameter that may differentiate DA, DOA, and DMS from DR is the requirement in the
former tasks to suppress movement toward the location or
object where the food reward was last seen. In DMS, monkeys
have to overcome their strong predisposition to approach novel
objects (Mishkin and Delacour 1975). Alternatively, activation
of the PC by the working memory tasks may be modulated by
increased processing demands while activation of the PC by
the DMS task may be related to the process of object recognition as indicated by recent lesion and electrophysiological
studies (Fahy et al. 1993; Gaffan and Murray 1992; Meunier et
al. 1993; Xiang and Brown 1998; Zhu et al. 1995; ZolaMorgan et al. 1989). The importance of task parameters for
adjusting processing demands is evident in lesion studies,
where, for example, deficits following perirhinal ablations in
monkeys can be revealed by increasing processing demands,
either by lengthening delays, using foils, increasing the number
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
2599
served in the present study of memory-guided behavior, further
heterogeneity in its functional architecture can be expected and
deserves further investigation.
Local maxima within the EC and PC
FIG. 8. Local maxima in the fundus of the rhinal sulcus and the lateral PC.
Graph shows the adjusted mean LCGU (␮mol 䡠 100 g⫺1 䡠 min⫺1) for all groups
studied for the lateral PC peak activation and the fundus peak activation. Mean
LCGU in the DMS group was significantly higher than that seen in the SMC,
DR, and DOA groups. No group differences were seen in the fundus of the
rhinal sulcus. * DMS ⬎ SMC; P ⬍ 0.05.
of stimuli, or changing the orientation of stimuli (Buckley and
Gaffan 1997, 1998a). In this respect, the lesion data and the
present 2-DG results correspond in showing that the PC is
more likely to be recruited when the level of processing required for successful task performance is sufficiently high.
Functional heterogeneity within EC
One of the most striking results emerging from this study is
the heterogeneity of glucose uptake within the EC. Specifically, medial regions of EC show strikingly lower LCGU
values than lateral regions. In fact, the lateral regions of EC
appear more similar to perirhinal regions than to medial EC
regions in terms of the general levels and patterns of glucose
uptake seen across all animals. This is interesting in light of
previous anatomical investigations that have demarcated the
region medial to the rhinal sulcus [referred to here as lateral EC
based on divisions by Suzuki and Amaral (1994)] as the
“prorhinal” cortex based on connectional and cytoarchitectonic
criteria. For example, Insauti et al. (1987) have shown that the
lateral regions of EC receive a much stronger cortical innervation than the medial EC. Nevertheless, they still include this
region as part of the EC (see Insausti et al. 1987, p. 391). The
present data, however, may renew the debate as to whether this
region is to be considered part of the EC or a transitional zone
between EC and PC based on physiological grounds of differential glucose uptake. Further evidence for a heterogeneity
within the EC comes from the present trend (P ⫽ 0.069) for the
animals performing the DA task, a spatial working memory
task, to show greater activation in lateral regions of EC. Specifically, it may imply that the lateral EC, which receives a
stronger input from area TF of the parahippocampal cortex
(Insausti et al. 1987), as well as PC may be activated by spatial
memoranda while the lateral perirhinal region, which receives
a stronger innervation from inferotemporal cortex, may be
preferentially activated by object information (Suzuki et al.
1997). In light of the heterogeneity of the rhinal cortex ob-
The local maxima analysis revealed that all animals exhibited local maxima within the fundus of the rhinal sulcus and the
lateral PC. However, although the local maxima in the fundus
was prominent in all animals, there were no group differences
in the quantitative estimates of LCGU within this region. The
area in the fundus of the rhinal sulcus is often included in
lesions of either the entorhinal or perirhinal cortices and it is
possible that damage to this region may cause nonspecific
deficits since we have shown that it is equally active in all
animals, including controls.
In contrast to the activations observed in the fundus, the
local maxima in the lateral perirhinal region revealed that the
animals performing the DMS task exhibited the highest levels
of LCGU in this region. As the PC receives substantial cortical
input from inferotemporal cortex (Suzuki and Amaral 1994), it
is possible that this increase reflects neuronal activation resulting from the greater number of novel visual objects seen by the
DMS animals compared with animals in the other groups.
However, it seems unlikely that this activation is purely perceptual since damage to PC has been shown to leave perception
intact while altering memory abilities (Buffalo et al. 1998;
Gutnikov et al. 1997). Thus given the wealth of data supporting
the crucial role of the PC in memory, these focal activations in
lateral PC may reflect the involvement of the neural tissue most
responsible for supporting stimulus memory within this region,
and, arguably, within the medial temporal lobe.
While an overall significant increase in LCGU in the PC was
observed in the monkeys performing the DA and DOA tasks,
no evidence of focal activity comparable to that observed in the
DMS monkeys was observed. It is possible that such a focus
would be found outside the regions of interest analyzed in the
present study. Indeed the dorsolateral prefrontal cortex, which
has been shown to support performance of the DA and DOA
tasks, has stronger connections with the hippocampal formation through the parahippocampal cortex and the caudomedial
lobule than through the perirhinal cortices (Goldman-Rakic et
al. 1984). Future studies could be directed toward examining
the role of these posterior structures in the battery of tasks
employed here. It also remains to be determined whether the
focal activation exhibited by the DMS group is directly or
multisynaptically connected to the ventral/orbital regions of the
prefrontal cortex that have been implicated in the performance
of DMS tasks (Bachevalier and Mishkin 1986; Kowalska et al.
1991; Meunier et al. 1997; Parker and Gaffan 1998; Passingham 1975) distinct from dorsal lateral regions engaged in the
performance of working memory tasks (Goldman and Rosvold
1970; Wilson et al. 1993).
We thank Dr. Harriet Friedman for valuable contributions to the earlier
stages of this investigation. We also thank J. Coburn, M. Pappy, and H. Findlay
for help in cryostat sectioning of the large primate brains; J. Hickey for help
in training animals; T. Beattie for help in all phases of the experiment; and L.
Romanski, S. O’Scalaidhe, and N. Vnek for helpful comments on the manuscript.
2600
L. DAVACHI AND P. S. GOLDMAN-RAKIC
This work was supported by National Institute of Mental Health Grant
MH-38546 to P. S. Goldman-Rakic.
REFERENCES
ALVAREZ P, ZOLA-MORGAN S, AND SQUIRE LR. The animal model of human
amnesia: long-term memory impaired and short-term memory intact. Proc
Natl Acad Sci USA 91: 5637–5641, 1994.
BACHEVALIER J AND MISHKIN M. Visual recognition impairment follows ventromedial but not dorsolateral prefrontal lesions in monkeys. Behav Brain
Res 20: 249 –261, 1986.
BUCKLEY MJ AND GAFFAN D. Impairment of visual object-discrimination
learning after perirhinal cortex ablation. Behav Neurosci 111: 467– 475,
1997.
BUCKLEY MJ AND GAFFAN D. Learning and transfer of object-reward associations and the role of the perirhinal cortex. Behav Neurosci 112: 15–23,
1998a.
BUCKLEY MJ AND GAFFAN D. Perirhinal cortex ablation impairs visual object
identification. J Neurosci 18: 2268 –2275, 1998b.
BUFFALO EA, REBER PJ, AND SQUIRE LR. The human perirhinal cortex and
recognition memory. Hippocampus 8: 330 –339, 1998.
CORRELL RE AND SCOVILLE WB. Significance of delay in the performance of
monkeys with medial temporal lobe resections. Exp Brain Res 4: 85–96,
1967.
FAHY FL, RICHES IP, AND BROWN MW. Neuronal signals of importance to the
performance of visual recognition memory tasks: evidence from recordings
of single neurones in the medial thalamus of primates. Prog Brain Res 95:
401– 416, 1993.
FRIEDMAN HR, BRUCE CJ, AND GOLDMAN-RAKIC PS. Resolution of metabolic columns by a double-label 2-DG technique: interdigitation and
coincidence in visual cortical areas of the same monkey. J Neurosci 9:
4111– 4121, 1989.
FRIEDMAN HR AND GOLDMAN-RAKIC PS. Activation of the hippocampus and
dentate gyrus by working memory: a 2-deoxyglucose study of behaving
rhesus monkeys. J Neurosci 8: 4693– 4706, 1988.
FRIEDMAN HR AND GOLDMAN-RAKIC PS. Coactivation of prefrontal cortex
and inferior parietal cortex in working memory tasks revealed by 2DG
functional mapping in the rhesus monkey. J Neurosci 14: 2775–2788,
1994.
FRIEDMAN HR, JANAS JD, AND GOLDMAN-RAKIC PS. Enhancement of metabolic activity in the diencephelon of monkeys performing working memory
tasks: a 2-deoxyglucose study in behaving rhesus monkeys. J Cognit Neurosci 2: 18 –31, 1990.
GAFFAN D AND MURRAY EA. Monkeys (Macaca fascicularis) with rhinal cortex
ablations succeed in object discrimination learning despite 24-hr intertrial
intervals and fail at matching to sample despite double sample presentations.
Behav Neurosci 106: 30 –38, 1992.
GELLERMAN LW. Chance orders of alternating stimuli in visual discrimination
experiments. J Gen Psychol 42: 207–208, 1933.
GOLDMAN PS AND ROSVOLD HE. Localization of function within the dorsolateral prefrontal cortex of the rhesus monkey. Exp Neurol 27: 291–
304, 1970.
GOLDMAN-RAKIC PS. Circuitry of primate prefrontal cortex and regulation of
behavior by representational memory. In: Handbook of Physiology. The
Nervous System. Higher Functions of the Brain. Bethesda, MD: Am.
Physiol. Soc., 1987, sect. 1, vol. V, p. 373– 417.
GOLDMAN-RAKIC PS, SELEMON LD, AND SCHWARTZ ML. Dual pathways connecting the dorsolateral prefrontal cortex with the hippocampal formation
and parahippocampal cortex in the rhesus monkey. Neuroscience 12: 719 –
743, 1984.
GUTNIKOV SA, MA YY, BUCKLEY MJ, AND GAFFAN D. Monkeys can associate
visual stimuli with reward delayed by 1 s even after perirhinal cortex
ablation, uncinate fascicle section or amygdalectomy. Behav Brain Res 87:
85–96, 1997.
INSAUSTI R, AMARAL DG, AND COWAN WM. The entorhinal cortex of the
monkey. II. Cortical afferents. J Comp Neurol 264: 356 –395, 1987.
KAPUR S, CRAIK FIM, TULVING E, WILSON AA, HOULE S, AND BROWN GM.
Neuroranatomical correlates of encoding in episodic memory: levels of
processing effects. Proc Natl Acad Sci USA 91: 2008 –2011, 1994.
KENNEDY C, SAKURADA O, SHINOHARA M, JEHLE J, AND SOKOLOFF L. Local
cerebral glucose utilization in the normal conscious macaque monkey. Ann
Neurol 4: 293–301, 1978.
KOWALSKA DM, BACHEVALIER J, AND MISHKIN M. The role of the inferior
prefrontal convexity in performance of delayed nonmatching-to-sample.
Neuropsychologia 29: 583– 600, 1991.
LEONARD BW, AMARAL DG, SQUIRE LR, AND ZOLA-MORGAN S. Transient
memory impairment in monkeys with bilateral lesions of the entorhinal
cortex. J Neurosci 15: 5637–5659, 1995.
LYNCH G, MATTHEWS DA, MOSKO S, PARKS T, AND COTMAN C. Induced
acetylcholinesterase-rich layer in rat dentate gyrus following entorhinal
lesions. Brain Res 42: 311–318, 1972.
MAHUT H. Spatial and object reversal learning in monkeys with partial temporal lobe ablations. Neuropsychologia 9: 409 – 424, 1971.
MAHUT H AND CORDEAU J-P. Spatial reversal deficit in monkeys with amygdalohippocampal ablations. Exp Neurol 7: 426 – 434, 1963.
MEUNIER M, BACHEVALIER J, AND MISHKIN M. Effects of orbital frontal and
anterior cingulate lesions on object and spatial memory in rhesus monkeys.
Neuropsychologia 35: 999 –1015, 1997.
MEUNIER M, BACHEVALIER J, MISHKIN M, AND MURRAY EA. Effects on visual
recognition of combined and separate ablations of the entorhinal and perirhinal cortex in rhesus monkeys. J Neurosci 13: 5418 –5432, 1993.
MISHKIN M. Memory in monkeys severely impaired by combined but not by
separate removal of amygdala and hippocampus. Nature 273: 297–298,
1978.
MISHKIN M AND DELACOUR J. An analysis of short-term visual memory in the
monkey. J Exp Psychol Anim Behav Process 1: 326 –334, 1975.
MURRAY EA, BAXTER MG, AND GAFFAN D. Monkeys with rhinal cortex
damage or neurotoxic hippocampal lesions are impaired on spatial scene
learning and object reversals. Behav Neurosci 112: 1291–1303, 1998.
MURRAY EA, GAFFAN D, AND MISHKIN M. Neural substrates of visual
stimulus-stimulus association in rhesus monkeys. J Neurosci 13: 4549 –
4561, 1993.
MURRAY EA AND MISHKIN M. Severe tactual as well as visual memory deficits
follow combined removal of the amygdala and hippocampus in monkeys.
J Neurosci 4: 2565–2680, 1984.
MURRAY EA AND MISHKIN M. Visual recognition in monkeys following rhinal
cortical ablations combined with either amygdalectomy or hippocampectomy. J Neurosci 6: 1991–2003, 1986.
PARKER A AND GAFFAN D. Interaction of frontal and perirhinal cortices in
visual object recognition memory in monkeys. Eur J Neurosci 10: 3044 –
3057, 1998.
PASSINGHAM R. Delayed matching after selective prefrontal lesions in monkeys
(Macaca mulatta). Brain Res 92: 89 –102, 1975.
PRIBRAM KH AND MISHKIN M. Analysis of the effects of frontal lesions in
monkey. III. Object alternation. J Comp Physiol Psychol 49: 41– 45,
1956.
SALEEM KS AND TANAKA K. Divergent projections from the anterior inferotemporal area TE to the perirhinal and entorhinal cortices in the macaque
monkey. J Neurosci 16: 4757– 4775, 1996.
SMITH EE AND JONIDES J. Neuroimaging analyses of human working memory.
Proc Natl Acad Sci USA 95: 12061–12068, 1998.
SOKOLOFF L, REIVICH M, KENNEDY C, DES ROSIERS MH, PATLAK CS, PETTIGREW KD, SAKURADA O, AND SHINOHARA M. The [14C]deoxyglucose
method for the measurement of local cerebral glucose utilization: theory,
procedure, and normal values in the conscious and anesthetized albino rat.
J Neurochem 28: 897–916, 1977.
SQUIRE LR AND ZOLA SM. Structure and function of declarative and nondeclarative memory systems. Proc Natl Acad Sci USA 93: 13515–13522,
1996.
SQUIRE LR AND ZOLA-MORGAN S. The neuropsychology of memory: new links
between humans and experimental animals. Ann NY Acad Sci 444: 137–149,
1985.
SQUIRE LR AND ZOLA-MORGAN S. The medial temporal lobe memory system.
Science 253: 1380 –1386, 1991.
SUZUKI WA AND AMARAL DG. Perirhinal and parahippocampal cortices of
the macaque monkey: cortical afferents. J Comp Neurol 350: 497–533,
1994.
SUZUKI WA, ZOLA-MORGAN S, SQUIRE LR, AND AMARAL DG. Lesions of the
perirhinal and parahippocampal cortices in the monkey produce long-lasting
memory impairment in the visual and tactual modalities. J Neurosci 13:
2430 –2451, 1993.
SUZUKI WA, MILLER EK, AND DESIMONE R. Object and place memory in the
macaque entorhinal cortex. J Neurophysiol 78: 1062–1081, 1997.
GLUCOSE UTILIZATION IN THE RHINAL CORTEX
THORNTON JA, ROTHBLAT LA, AND MURRAY EA. Rhinal cortex removal
produces amnesia for preoperatively learned discrimination problems but
fails to disrupt postoperative acquisition and retention in rhesus monkeys.
J Neurosci 17: 8536 – 8549, 1997.
TULVING E, KAPUR S, MARKOWITSCH HJ, CRAIK FIM, HABIB R, AND
HOULE S. Neuroanatomical correlates of retrieval in episodic memory:
auditory sentence recognition. Proc Natl Acad Sci USA 91: 2012–2015,
1994.
WILSON FA, SCALAIDHE SP, AND GOLDMAN-RAKIC PS. Dissociation of object
and spatial processing domains in primate prefrontal cortex [see comments].
Science 260: 1955–1958, 1993.
XIANG JZ AND BROWN MW. Differential neuronal encoding of novelty, famil-
2601
iarity and recency in regions of the anterior temporal lobe. Neuropharmacology 37: 657– 676, 1998.
ZHU XO, BROWN MW, AND AGGLETON JP. Neuronal signalling of information
important to visual recognition memory in rat rhinal and neighbouring
cortices. Eur J Neurosci 7: 753–765, 1995.
ZOLA-MORGAN S AND SQUIRE LR. Medial temporal lesions in monkeys impair
memory on a variety of tasks sensitive to human amnesia. Behav Neurosci
99: 22–34, 1985.
ZOLA-MORGAN S, SQUIRE LR, AMARAL DG, AND SUZUKI WA. Lesions of
perirhinal and parahippocampal cortex that spare the amygdala and hippocampal formation produce severe memory impairment. J Neurosci 9:
4355– 4370, 1989.