Visual paired comparison performance is impaired in a patient with selective hippocampal lesions and relatively intact item recognition

LGF Group, Department of Psychology, The University of Sheffield, Western Bank, Sheffield S10 2TP, UK.
Neuropsychologia (Impact Factor: 3.3). 02/2004; 42(10):1293-300. DOI: 10.1016/j.neuropsychologia.2004.03.005
Source: PubMed
ABSTRACT
In this study, we have examined visual recognition memory in a patient, YR, with discrete hippocampal damage who has shown normal or nearly normal item recognition over a large number of tests. We directly compared her performance as measured using a visual paired comparison task (VPC) with her performance on delayed matching to sample (DMS) tasks. We also investigated the effect of retention interval between familiarisation and test. YR shows good visual recognition with the DMS task up to 10 s after the familiarisation period, but only shows recognition with the VPC task for the shortest retention interval (0 s). Our results are consistent with the view that hippocampal damage disrupts recollection and recall, but not item familiarity memory.

Full-text

Available from: Olivier Pascalis
Neuropsychologia 42 (2004) 1293–1300
Visual paired comparison performance is impaired in a patient with
selective hippocampal lesions and relatively intact item recognition
O. Pascalis
a,
, N. M. Hunkin
a
, J. S. Holdstock
b
, C. L. Isaac
a
, A. R. Mayes
b
a
LGF Group, Department of Psychology, The University of Sheffield, Western Bank, Sheffield S10 2TP, UK
b
Department of Psychology, University of Liverpool, Eleanor Rathbone Building, Bedford Street South, Liverpool L69 3BS, UK
Received 16 July 2003; accepted 16 March 2004
Abstract
In this study, we have examined visual recognition memory in a patient, YR, with discrete hippocampal damage who has shown normal
or nearly normal item recognition over a large number of tests. We directly compared her performance as measured using a visual paired
comparison task (VPC) with her performance on delayed matching to sample (DMS) tasks. We also investigated the effect of retention
intervalbetweenfamiliarisation andtest. YRshowsgood visualrecognition withtheDMStaskupto10 safter thefamiliarisationperiod,but
only shows recognition with the VPC task for the shortest retention interval (0s). Our results are consistent with the view that hippocampal
damage disrupts recollection and recall, but not item familiarity memory.
© 2004 Elsevier Ltd. All rights reserved.
Keywords: Recognition memory; Hippocampus; Amnesia; VPC
1. Introduction
Evidence from lesion studies indicates that damage to the
medial temporal lobes impairs visual recognition memory
and episodic memory (Squire, 1992). There is consistent
evidence from studies in monkeys that a visual item recogni-
tion memory deficit occurs after damage to medial temporal
cortex sites that are adjacent to the hippocampal formation.
Selective lesions of either the entorhinal and perirhinal cor-
tex (Gaffan & Murray, 1992), or the perirhinal cortex and
parahippocampal cortex (Zola-Morgan, Squire, Amaral,
& Suzuki, 1989) or the perirhinal cortex alone (Meunier,
Bachevalier, Mishkin, & Murray, 1993) yielded severe
recognition memory loss. There is, however, conflicting
evidence on the effects of selective hippocampal lesions on
visual item recognition memory and disagreement about the
mnemonic role played by the hippocampus. Whereas, some
animal lesion studies, using the Delayed Non-Matching to
Sample (DNMS) task, have found that selective damage to
the hippocampal formation resulted in impairment at the
longest delays only (Zola-Morgan, Squire, Rempel, Clower,
& Amaral, 1992; Alvarez, Zola-Morgan, & Squire, 1995),
other studies have found no impairment of visual recogni-
tion memory ability (Murray & Mishkin, 1998). The effect
Corresponding author. Tel.: +44-114-222-6548.
E-mail address: o.pascalis@sheffield.ac.uk (O. Pascalis).
of a hippocampal lesion on visual recognition memory as
measured with the DNMS task is then still under debate
(see Bachevalier, Nematic, & Alvarado, 2003; Baxter &
Murray, 2001a for a discussion).
Human lesion studies have found similar conflicting re-
sults. Whereas, some studies have found clear visual item
recognition deficits after relatively selective hippocampal
damage (e.g., Cipolotti et al., 2001; Manns & Squire, 1999;
Manns, Hopkins, Reed, Kitchener, & Squire, 2003; Reed
& Squire, 1997), others have found little or no impairment
in item recognition memory (Mayes, Holdstock, Isaac,
Hunkin, & Roberts, 2002; Vargha-Khadem et al., 1997;
Yonelinas et al., 2002). The explanation of these conflicting
results is unknown, but most probably involves either differ-
ing extents and locations of hippocampal damage, differing
extents of damage or dysfunction in extra-hippocampal sites
critical for visual item recognition, or both (for example,
see Baxter & Murray, 2001a and b; Zola & Squire, 2001).
Aggleton and Brown (1999) have hypothesised that dam-
agerestrictedtothehippocampalformationimpairs only rec-
ollection, leaving familiarity-based item recognition mem-
ory intact. The evidence about whether hippocampal dam-
age leaves familiarity memory intact is conflicting.Whereas,
Yonelinas et al. (2002) found, using several methods of as-
sessment, that familiarity was preserved in patients who
probably suffered hippocampal damage caused by hypoxia
following a cardiac arrest, Manns et al. (2003) found that
0028-3932/$ see front matter © 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuropsychologia.2004.03.005
Page 1
1294 O. Pascalis et al./Neuropsychologia 42 (2004) 1293–1300
familiarity was as impaired as recollection in a group of pa-
tients who had suffered hippocampal damage. Provided, as
many researchers assume, that relatively normal item recog-
nition memorycanbesupportedbyfamiliarityalone,the Ag-
gleton and Brown hypothesis is consistent with hippocam-
pal lesions causing only a mild item recognition deficit that
might not be detectable in single patients. In contrast, if hip-
pocampal lesions disrupt familiarity as well as recollection,
then they should severely disrupt item recognition memory.
Item recognition memory has been assessed not only di-
rectly through the use of tests such as DNMS, but also
indirectly with the Visual Paired Comparison (VPC) task
(McKee & Squire, 1993; Pascalis & Bachevalier, 1999; Zola
et al., 2000). The VPC task, which was developed by Fantz
(1964), is a common way to measure visual recognition in
preverbal and nonverbal individuals (Fagan, 1974; Pascalis
& Bachevalier, 1998). It exploits individuals’ attraction to
novelty in order to assess their recognition memory for pre-
viously seen stimuli. The basic procedure is as follows:
The participant is first presented with a stimulus for a fa-
miliarisation period. Later, the participant is presented with
the same stimulus paired with a novel one. The key mea-
sure is the length of time spent fixating each of the two
stimuli. Longer duration of looking towards one stimulus,
generally the novel one, indicates discrimination and, in-
directly, recognition memory. Long-term recognition mem-
ory has been shown during infancy with this task (Fagan,
1974; Pascalis, de Haan, Nelson, & de Schonen, 1998).
The type of memory assessed by this task in infants has
been controversial, but according to Nelson (1995), it tests a
pre-explicit form of memory. Recently, however, a study of
adults showed that novelty preference correlated with later
recognition of the non-preferred stimulus, which is consis-
tent with the view that it provides an indirect index of the
ability to show aware recognition of studied stimuli (Manns,
Stark, & Squire, 2000).
Performance on the VPC task, in both infant and adult
monkeys with damage to the medial temporal lobe that
included the hippocampal formation, amygdala and sur-
rounding tissue has been found to be abnormal (Bachevalier,
Brickson, & Hagger, 1993). Recently, Pascalis and
Bachevalier (1999) showed that adult monkeys with neona-
tal hippocampal lesions showed preference for novelty at
short delays of 10s, but not at longer delays of 30s to 24h,
whereas normal monkeys showed novelty preference at all
delays. Consistent with this study, other studies of monkeys
with selective lesions within the medial temporal lobe, have
shown that novelty preference depends on the integrity
of the hippocampal formation (Zola et al., 2000) as well
as the perirhinal cortex (Buffalo et al., 1999). In humans,
McKee and Squire (1993), using the VPC task, have shown
that amnesic patients with relatively selective hippocampal
damage also show abnormal novelty preference. The pa-
tients in this study also showed clearly impaired visual item
recognition when this was measured directly, although their
performance was above chance levels.
Although monkeys, who showed an impaired novelty
preference on the VPC task (Pascalis & Bachevalier, 1999),
showed close to normal recognition on a DNMS task
(Bachevalier, Beauregard, & Alvarado, 1999), this has not
been demonstrated in any human patients with relatively se-
lective hippocampal lesions. If a single dissociation can be
shown between VPC performance and a direct performance
measure of visual item recognition, this should help con-
strain hypotheses about the processes that underlie directly
measured recognition and VPC performance. To determine
whether this kind of single dissociation can be found in
humans, we have examined VPC performance in a patient,
YR, with discrete hippocampal damage who has shown
normal or nearly normal item recognition over a large num-
ber of tests (Mayes et al., 2002). We directly compared her
VPC performance with her performance on delayed match-
ing to sample (DMS) tasks. We also investigated the effect
of retention interval between familiarisation and test.
2. Methods
2.1. Participants
The participants were patient YR and five age- and
IQ-matched healthy control participants. YR, a female,
was 62-year-old at the time of testing and had developed a
memory impairment following a possible ischaemic infarct,
arising from the administration of an opiate drug to relieve
severe back pain 14 years previously. YR’s neuropathol-
ogy and neuropsychological profile are reported in detail
by Holdstock et al. (2000) and Mayes et al., in press.For
clarification, an overview of YR’s neuropathological and
neuropsychological details are included here.
Magnetic resonance imaging (MRI) was carried out in
September 1997 using a 1.5T SIGNA whole-body magnetic
imaging system (General Electric, Milwaukee, WI). A 3D
T1-weighted radio-frequency spoiled gradient echo (SPGR)
image revealed a selective lesion affecting the hippocampus
bilaterally along its full anterior–posterior extent. Volumet-
ric analysis indicated that the volumes of the hippocampi
(corrected for intracranial volume) were 2.5 and 3 S.D.s
smaller than the mean volumes in a group of healthy control
participants (matched for sex, age and IQ) on the right and
left, respectively. In contrast, there was no pathology evi-
dent in the parahippocampal gyrus, and the corrected vol-
ume of this region, which included the perirhinal, entorhinal
and parahippocampal cortices, was at least 1 S.D. greater
than that of the control participants. Although the amygdala
appeared small, there was no evidence of pathology. Frontal
lobe structures were intact, and grey to white matter ratios
were normal. There was some evidence of parietal lobe at-
rophy, but this was not atypical for a woman of YR’s age,
and her corrected parietal lobe volume was within the con-
trol range on the right, and only just below the control range
on the left.
Page 2
O. Pascalis et al./Neuropsychologia 42 (2004) 1293–1300 1295
Table 1
Performance of YR on standardised tests of intellectual and memory
function
Tests Subtests YR’s performance
NART FSIQ 115
WAIS-R Full 102
Verbal 108
Performance 97
WMS-R Verbal 62
Visual 102
General 66
Attention 122
Delay 73
WRMT Words 11
a
(50–75 percentile)
Faces 16
a
(>95 percentile)
D&P People (verbal recall) 4
a
(<5 percentile)
Names (verbal recognition) 16
a
(99 percentile)
Shapes (non-verbal recall) 5
a
(1–5 percentile)
Doors (non-verbal recognition) 9
a
(50 percentile)
Key: NART FSIQ: estimated full scale IQ from National Adult Reading
Test (Nelson, 1991); WAIS-R: Wechsler Adult Intelligence Scale, Revised
(Wechsler, 1981); WMS-R: Wechsler Memory Scales, Revised (Wechsler,
1987); WRMT: Warrington Recognition Memory Scale (Warrington,
1984); D&P: Doors and People Test (Baddeley et al., 1994).
a
Denotes age-scaled scores.
YR’s performance on standardised tests of IQ and mem-
ory function given up to the time of scanning are shown in
Table 1. Her current FSIQ (WAIS-R; Wechsler, 1981)was
in the average range and, although she showed a decrement
of 13 points between her estimated pre-morbid IQ (NART;
Nelson, 1991) and her current IQ, this represented a drop
of <1 S.D. in IQ points. On tests of memory function, YR
showed a consistent pattern in that she had a severe im-
pairment on tests of verbal and non-verbal recall, but intact
performance on tests of verbal and non-verbal item recogni-
tion (see Mayes et al., 2002). This pattern of impairment is
shown clearly by her performance on the Doors and People
Test (Baddeley, Emslie, & Nimmo-Smith, 1994) on which
she scored below the fifth percentile on tests of recall, but at
the 50th percentile or above on tests of recognition. Her per-
formance on the verbal and visual subtests of the Warring-
ton Recognition Memory Test (WRMT; Warrington, 1984)
was also above the 50th percentile in both cases. The disso-
ciation between YR’s impaired visual as well as verbal re-
call and her relatively intact performance on tests of visual
and verbal item recognition respectively has been studied in
greater detail, and is reported by Mayes et al. (2002).
In addition to tests of memory, YR’s perceptual func-
tion was assessed with the Visual Object and Space Percep-
tion Battery (VOSP; Warrington & James, 1991). YR scored
within 1 S.D. of the mean of the normative sample for all
subtests, and performed better than the controls on one of
the four object perception subtests (Silhouettes) and three of
the four spatial subtests (Dot Counting, Position Discrimi-
nation, Cube Analysis).
Five control participants were recruited. These were
healthy, female volunteers who were matched to YR in
terms of age (mean age: 63.20 years, S.D.: 2.17) and
WAIS-R FSIQ (mean IQ: 101.60, S.D.: 5.03).
2.2. Stimuli
Different sets of stimuli were used in the VPC and DMS
tasks.
2.2.1. VPC task
One hundred and sixty black and white slides of objects
and faces were used as stimuli. The size and brightness of
the objects were kept uniform on each slide. When projected
onto the screen, the size of the stimuli was 8cm × 10 cm,
and when two stimuli were present, they were separated by
a 5cm gap.
2.2.2. Delayed matching to sample task (DMS)
2.2.2.1. Objects. Seventy-two colour photographs of ev-
eryday objects were selected. The photographs were pre-
sented on a computer screen to eight young healthy vol-
unteers, who were required to rate each photograph on a
scale of familiarity from 1 to 5. The 72 photographs were
then divided into six sets of 12, which were matched on the
familiarity rating. Three of the sets were designated ‘target’
sets and three were designated ‘distractor’ sets; each target
set was paired with a distractor set. Each pair of sets was
allocated to one of the three delay conditions (0, 5, 10 s
delay). The photographs were saved as black and white
pictures, and were presented on a white background. All
pictures occupied a 7cm×8.5cm rectangle on the computer
screen. During study, single pictures were presented at the
centre of the screen. At test, two pictures were presented
side-by-side separated by a 5.5cm gap.
2.2.2.2. Faces. Seventy-two black and white novel faces
were selected from our database. The 72 faces were arranged
into 36 pairs, each of which comprised two faces that were
matched in gender, approximate age and physical appear-
ance. One member of each pair was designated the target
face, and the other was designated the distractor face. The
36 pairs were divided into three sets of 12 pairs. Each set
was allocated to one of the three delay conditions (0, 5, 10 s
delay). All faces were presented on a white background, and
occupied a 12cm× 12cm rectangle on the computer screen.
During study, single pictures were presented at the centre of
the screen. At test, two pictures were presented side-by-side
separated by a 2cm gap.
2.3. Procedure
The two tasks were administered separately over several
sessions. The VPC task was administered first during four
Page 3
1296 O. Pascalis et al./Neuropsychologia 42 (2004) 1293–1300
sessions over a 1 month period. The DMS task was admin-
istered in one session 2 months later.
2.3.1. VPC task
YR and the controls were investigated individually. Ex-
perimental conditions were as follows. First, each partici-
pant was shown a single target stimulus to inspect during
a 5 s familiarisation period. After a delay during which a
blank screen was presented, the participant was shown the
target object or face paired with a new stimulus, for 5s.
The left–right position of the novel stimulus was counter-
balanced across trials. The delays tested were no delay (i.e.,
the time taken to change the slide, which was around 1 s), 5
and 10s. Twelve trials with faces and 12 trials with objects
were used at each delay. The trials for each delay and cat-
egory of stimulus were randomly intermixed and presented
over several testing sessions. Participants were told that they
were part of a vision study, and we explained to YR that she
was the control of a patient with a visual problem. Partici-
pants were instructed that one picture would appear on the
screen for a brief presentation followed by a brief period of
rest, then two pictures would be simultaneously presented.
They were asked only to “look at the screen as if you were
watching TV”. The dependent variable was actual looking
time directed to the new and to the old stimulus.
A video camera with a videotimer was fixed above the
screen and recorded participants’ eye movements onto
videotape. Stimulus fixation was indicated by corneal re-
flection of the stimuli. Inspection of the videotape after the
experiment allowed the time spent inspecting the right and
left images in the 5s recognition phase to be assessed.
2.3.2. DMS task
Each participant was exposed to a stimulus for 5s and in-
structed to remember it. After a brief delay (0, 5 or 10s),
during which a blank screen was presented, the participant
was shown the familiar (target) stimulus together with a
novel (distractor) stimulus. The participant was required to
point to the familiar stimulus. The two stimuli were pre-
sented side-by-side, and the left/right position of the famil-
iar/novel stimuli was counterbalanced across trials. There
were 12 trials at each of the three delays for objects and 12
trials at each of the three delays for faces. Each delay condi-
tion was run separately in the following order: objects 0, 5,
10s; faces 0, 5, 10 s. The order of conditions was the same
for all participants. Within each condition, the stimuli were
presented in random order.
3. Results
To determine whether there was a significant difference
between the novelty preference of YR and that of the control
participants on the VPC task, the difference in time spent
looking at the familiar and novel stimuli was analysed. Nov-
elty preference was defined as the time spent looking at the
Table 2
Mean difference in time spent looking at novel and familiar pictures
Condition Control mean
(S.D.)
YR mean
(S.D.)
t-value Probability
(two-tailed)
Objects
0s 1.030 (1.275) 0.900 (1.258) 0.964 0.349
5s 1.276 (1.552) 0.487 (2.173) 4.456 0.001
10s 1.415 (1.393) 0.200 (1.928) 5.185 0.000
Faces
0s 1.241 (1.216) 1.033 (1.138) 1.873 0.078
5s 1.404 (1.532) 0.325 (1.257) 7.859 0.000
10s 1.669 (1.534) 0.453 (1.357) 7.677 0.000
A positive value indicate a novelty preference.
familiar stimulus subtracted from the time spent looking at
the novel stimulus. This was calculated for each participant
on each of the 12 trials in each condition. YR’s novelty
preference in each condition was then compared with that
of the control participants using a t-test. Since it cannot be
assumed that the variance of YR’s performance was equiv-
alent to that of the group of control participants, Welch’s
procedure, which tests for the significance of the difference
between means when the population variances are unequal
(Ferguson, 1976, p. 168), was adopted. The results of this
analysis are shown in Table 2. YR showed a similar nov-
elty preference to that of the controls for both objects and
faces at a 0 s delay. However, for both types of material she
showed a significant difference in novelty preference from
the controls at delays of 5 and 10s.
Additional t-tests were carried out to determine whether or
not the novelty preference exhibited by YR and the control
participants was significantly above chance. These t-tests
showed that novelty preference in each individual control
participant was significantly above chance in each condition
(all P’s < 0.001). In contrast, YR’s novelty preference was
only significantly above chance in the ‘objects 0 s’ condition
(t(11) = 2.478, P<0.05) and the ‘faces 0 s’ condition (t(11)
= 3.144, P<0.01). In all other conditions, YR’s novelty
preference did not differ from chance (all P’s > 0.2). On
the DMS task, YR and all control participants had perfect
scores.
4. Discussion
On the VPC task, YR showed normal novelty preference
at a 0s delay but was impaired relative to controls, and did
not show novelty preference, at increased delays of 5 and
10s.In contrast YR’s item recognition, as tested by the DMS
task, which used the same types of object and face stimuli
as those used in the VPC task, was unimpaired at delays
of up to 10s. YR correctly recognised all the studied target
stimuli, even though she had not shown a novelty preference
in the VPC task.
YR’s abnormal novelty preference for faces and objects
as shown by the VPC task is consistent with previous re-
Page 4
O. Pascalis et al./Neuropsychologia 42 (2004) 1293–1300 1297
sults from humans and non-human primates. In monkeys,
a deficit has been observed after a 10s delay following a
specific lesion of the hippocampus (Zola et al., 2000), or
after a lesion of the perirhinal cortex (Buffalo et al., 1999),
or after a lesion of the hippocampal formation (Pascalis &
Bachevalier, 1999). A lesion of the area TE led to a deficit
without delay (Buffalo et al., 1999).
A human study involving patients with various aetiologies
showed that a deficit occurred between 2min and a 1h delay
(McKee & Squire, 1993). Some of this evidence suggests
that hippocampal damage, whether in humans or non-human
primates, disrupts VPC performance just as was found in
YR. In YR, an abnormality in novelty preference behaviour
became apparent at short delays of only 5s. The temporal
characteristics of YR’s abnormality are similar to those
observed in studies with monkeys (Pascalis & Bachevalier,
1999; Zola et al., 2000) but appear to differ from those of
the amnesic patients reported by McKee and Squire (1993).
McKee and Squire’s patients showed novelty preference at
a 2 min delay but not after 1h. Although five of McKee and
Squire’s patients had midline diencephalic lesions, six of
them had relatively selective hippocampal lesions so they
should be expected to perform like YR. Unfortunately, Mc-
Kee and Squire did not report the novelty preference scores
for individual patients. Although the mean performance of
the group (53.1%) at the 2min delay was significantly above
chance (50%), the range of performance (50.6–59.0%)
suggests that there were some individual patients whose
novelty preference may not have differed from chance.
Furthermore, the mean performance of the amnesic group
was significantly impaired relative to the controls (64%),
and represented a 16% decrease in novelty preference when
compared with performance at a 0.5 s delay (69.1%). This
can be contrasted with the controls who showed only a 4.1%
decrease in performance from a 0.5 s (68.1%) to a 2min
delay. Thus, the VPC data from YR are not necessarily in-
consistent with those reported by McKee and Squire (1993)
as it is possible that some of McKee and Squire’s patients
may have presented a deficit similar to YR that is masked by
averaging.
There remains the puzzle, however, why YR shows an
impaired novelty preference despite her ceiling level per-
formance on item recognition using DMS when this was
explicitly tested using the same types of materials and task
design. It is unlikely that the solution to this puzzle relates
to differences in the materials used in the two studies. There
is, however, another possible explanation which needs to
be considered briefly. Since running this study in 1999 and
2000, YR’s performance on a range of tests of intelligence,
perception and memory has shown deterioration, and she no
longer exhibits the consistent behavioural pattern described
above, which she had displayed up until early 1999. She
is currently being investigated clinically for a dementing
illness, and it is possible that, if confirmed, the disease may
have been progressing at the time of VPC testing. How-
ever, there was one clear exception to the cognitive decline
that became apparent on reassessment in 2000: YR’s face
recognition appeared to be intact. Moreover, on tests of
face recognition that were given over several years from
1995, she performed at slightly above the mean level of her
controls and showed no deterioration across time (Mayes
et al., 2002). Specifically, her performance on one of these
tests, which was modelled on the WRMT and administered
3 months before VPC testing, was norm