- Email: [email protected]
Lack of effect of dorsomedial thalamic lesions on automated tests of spatial memory in the rat
Behavioural Brain Research, 55 (1993) 39-49 © 1993 Elsevier Science Publishers B.V. All rights reserved. 0166-4328/93/$06.00
39
BBR 01433
Lack of effect of dorsomedial thalamic lesions on automated tests of spatial memory in the rat N . N e a v e a, A. S a h g a l b a n d J . P . A g g l e t o n a aDepartment of Psychology, University of Durham, Durham (UK) and hMR C Neuroehemical Pathology Unit, Newcastle General Hospital Ne wcastle-upon-Tyne (UK) (Received 28 August 1992) (Revised version received 13 January 1993) (Accepted 14 January 1993)
Key words: Nucleus medialis dorsalis; Spatial memory; Nonmatching-to-position; Amnesia; Thalamus
The effects of cytotoxic lesions of nucleus medialis dorsalis on tests of spatial memory were examined in the rat. Extensive lesions of the nucleus did not impair either the acquisition or the subsequent performance of an automated test of working memory, delayed nonmatchingto-position. Detailed analysis of the animals' performance over varying retention delays failed to reveal any evidence of a deficit. The same animals performed normally in a spatial discrimination task and its subsequent reversals. The present results can be contrasted with those from animals with hippocampal system damage.
INTRODUCTION
Interest in the contribution of the thalamic nucleus medialis dorsalis (MD) to learning and memory derives from two main sources. First, neuropathological studies have linked damage in M D with diencephalic amnesias of varied aetiology 22'34'41. Unfortunately, the lack of cases with damage confined to M D coupled with the close proximity of M D to other nuclei or fibre tracts also implicated in amnesia 9'23, has left uncertain the contribution of this nucleus to memory processes. Second, M D has very dense, reciprocal connections with the prefrontal cortex. As a consequence it is assumed that an understanding of the prefrontal cortex will, in part, depend on an understanding of M D and vice versa. Evidence that parts of the prefrontal cortex have mnemonic functions s'1°'26 therefore, indirectly, implicates M D in similar functions. The present study examined the performance of rats with intrathalamic injections of N-methyl-D-aspartic acid ( N M D A ) on an automated test of working memory, delayed nonmatching-to-position. The rationale for this experiment arose from the opportunity to produce selective M D lesions with a neurotoxin that limCorrespondence: J.P. Aggleton, Department of Psychology, University of Durham, South Road, Durham DH1 3LE, UK.
its damage to fibres of passage 27. The particular memory task, delayed nonmatching-to-position (DNMP), was selected for several reasons. It has recently been confirmed that automated tests of spatial working memory are sensitive to diencephalic amnesia in humans iv. Furthermore, the demands of the D N M P task share many features with delayed response and delayed alternation. As these two tasks are the benchmark tests for prefrontal cortex function in animals, the D N M P task is particularly appropriate for assessing the effects of M D damage. Finally, the D N M P task may help resolve the seemingly inconsistent effects of M D damage on spatial memory in rats 8'14'20'25'26'36. It is for this final reason that the rats were also tested on a spatial discrimination task after completing the D N M P task. For this discrimination task the animals were reinforced for selecting the correct lever in an operant chamber, the position of the correct choice remaining constant throughout a session but changing in a random manner across sessions. In the D N M P task the animal is trained to press a sample lever, and, after a variable delay, is presented with a choice of two levers, one of which was the sample. The animal is rewarded for selecting the lever that was not the previous sample i.e. nonmatching. Using this task it is not only possible to derive forgetting curves for individual animals but it is also possible to
40 assess performance using a variety of accuracy and bias indices 1~,3t,32 derived from the theory of signal detection (TSD). A number of previous studies have revealed that this task is sensitive to the effects of hippocampal system damage and to prefrontal cortex damage in the rat 3'~°, although the effects of MD damage are unknown. In the light of recent proposals concerning the contribution of M D to learning and memory 12'35, the present study examined the effects of M D lesions both on the acquisition of the D N M P task (MDpre) and on the performance of rats that had received surgery after D N M P acquisition (MDpost).
MATERIALS AND M E T H O D S
Subjects The study involved 37 naive, male rats of the pigmented DA strain (Bantin and Kingman, Hull). Throughout the period of the experiment the animals were housed individually under diurnal conditions (14 h light/10 h dark), all testing occurring at a regular time during the light period. The animals were fed approximately 15 g of laboratory diet (Labsure ERM) daily so that they did not drop below 8 5 ~ of normal body weight. At the start of testing the animals were aged approximately 4 months and weighed between 200 and 300 g. All animals had free access to water.
Apparatus Testing was carried out in 7 operant chambers (Campden Instruments Ltd., Loughborough) under the online control of 2 Spider microprocessors (Paul Fray Ltd., Cambridge). Each chamber was fitted with two retractable levers situated 7.5 cm either side of a central food tray. This food dispenser, which delivered 45 mg pellets (Campden Instruments), had a hinged perspex lid at which nose pokes could be recorded. A light was located inside the food dispenser and a house light was located in the centre of the roof. A further light was situated above each lever.
Delayed nonmatching-to-position Approximately 2 weeks after 10 animals had received M D surgery, all subjects (10 MDpre and 27 Normal Controls) were given the same, standardized training protocol. (I) Magazine training. The animals were allowed to eat ten reinforcement pellets that had been placed in the food tray with all of the house lights on. Training stopped when the animal ate all pellets during a period of 10 min on three consecutive daily sessions.
(2) Autoshaping. A randomly selected lever was inserted into the chamber and the light above it lit. The lever remained 'out' until the animal responded by pressing, at which point the lever was retracted, the food tray light came on and a single reward was delivered 3~. If no response was made after 10 s the lever was retracted. Autoshaping finished when the animal had responded on at least 50 out of 60 lever presentations over three consecutive sessions. (3) Delayed nonmatching-to-position. The testing procedure has been described in full in a number of recent publications 2"3"31. During the first stage of testing ('0' s delay) a trial would consist of the following sequence of events. After a 10 s inter-trial interval, either the left or the right hand lever emerged and the corresponding lever light illuminated. The rat then had to respond to the lever within 10 s, upon which the lever would retract, a pellet would be dispensed and the magazine tray illuminated. The subject then had to approach the tray and, within 10 s, operate the magazine flap. Both levers then emerged, and the stimulus lights above them were illuminated. The rat now had to respond, for a food pellet, to the lever which had not appeared as the sample; these choice responses also had to be made within 10 s. Incorrect responses, or a failure to respond, resulted in a 'time-out' of 10 s, during which all lights in the chamber were switched off, and both levers retracted; these were the only occasions when the chamber was unlit. A session consisted of 60 trials. A formal procedure was used to encourage an animal to respond on the opposite lever if it had developed a pronounced side bias. Delays between the sample and choice lever presentations were introduced once rats had reached a criterion of 900/0 correct on three consecutive sessions. For all subsequent trials magazine responses following the sample presentation were ineffective (but were recorded) until the appropriate delay interval had lapsed; the first response after this resulted in the choice levers being presented, providing this response occurred within 10 s of the end of the delay. The rats were first given a retention delay of 2 s with the same performance criterion as 0 s. Following this the number of trials was increased to 96 and each session contained an equal mixture of 0, 2 and 4 s (four sessions), 0, 2, 4 and 8 s (four sessions), and 0, 2, 4, 8, 16, and 32 s (six sessions) delay intervals. (4) Postoperative performance. Following completion of this protocol, 15 of the Normal Controls (now 7 months old) received surgeries (7 MDpost, 8 Controls). Approximately one month after surgery, these 15 rats, and 9 of the rats operated prior to acquisition (MDpre), were retested on the D N M P task. All ani-
41 mals received 15 sessions with a balanced, mixture of 0-32 s (as in the final stage of acquisition).
Spatial discriminations and reversals One day after completing the D N M P task, the animals were trained on a spatial discrimination task in the same operant chamber. For each session the animal was presented with both levers, one of which had been arbitrarily designated as S +, the other as S - . The pair of levers were extended until one was pressed (there were no 'miss' trials) at which time both were retracted. Responding on the correct (S + ) lever was followed by the delivery of one reward pellet. Responding on the incorrect lever (S - ) was recorded as an error and no reward given. The intertrial interval was set at 7 s. Each session consisted of 30 trials, the correct lever always being on the same side. The animals received 16 sessions, the position of the correct lever following a balanced, pseudorandom sequence.
Performance measures and analyses The data from the 0 to 32 s delay condition on the D N M P task were analysed to provide three indices of accuracy. These consisted of percent correct and two measures of sensitivity (A' and SI) derived from nonparametric signal detection models al. Three measures of bias (B", RI, and Iy) were also calculated aa'32. The first of these (B") reflects perceptual bias, the second (RI) reflects response bias, while the third (Iy) contrasts accuracy between the two levers 1~'32. These various measures are of interest as they may help distinguish between different types of impairments. Finally, estimates of general responsivity were also recorded. These included (a) latency to respond to the sample lever pooled across delays, since delay was unpredictable at the sampling stage, (b) latency to make the first magazine response ('nose-poke'), (c)rate of responding to the magazine flap, or nose-pokes/s, excluding the 0 delay condition, (d) choice (and average choice) latencies and (e) the number of missed trials. The justification and calculation of these signal detection measures and bias indices has been described in detail elsewhere 3'32. It should be noted that only the absolute values of B" and RI were analysed, since only the magnitude of the bias is of interest, rather than position or direction (for example, left vs. right position bias). The data were transformed as appropriate (arcsin: all accuracy and bias indices; logarithmic: latencies; square-root: misses) and analysed by parametric analysis of variance (ANOVA). For the sake of brevity, analyses failing to reach at least a P < 0 . 1 level are not presented in detail.
Surgical and histological procedures For the M D surgeries the animals were anaesthetised by intraperitoneal injection (4 ml/kg) of a solution containing 42 mg/ml of chloral hydrate and 9.7 mg/ml pentobarbitone sodium (Equithesin). Each animal was then placed in a stereotaxic frame (David Kopf Instruments, Tujunga). The scalp was cut and reflected to expose the skull. Following craniotomy, 0.36 #1 of 0.12 M N M D A (Sigma Chemical Company Ltd., Poole) dissolved in phosphate buffer (pH 7.2) was injected through a 1 #1 Hamilton Syringe in each hemisphere. The stereotaxic coordinates relative to ear-bar zero, with the incisorbar set at + 5.0 relative to the horizontal plane, were AP + 3.7, HT + 4.6, and LAT + 0.7. Each injection was made gradually over a 5-min period and the needle allowed to remain in situ for a further 5 min before being withdrawn. After completion of the lesion sulphanilamide powder was applied and the skin sutured. The animals acting as non surgical controls (Control group) were just injected with the same dose of Equithesin and allowed to recover. On completion of the experiments the rats were killed and perfused intracardially with 5 ~o formol-saline. The brains were then rapidly removed and placed in 5~o formol-saline. Subsequently, the brains were blocked, embedded in wax (Paraplast), and cut in 10-/~m coronal sections. Every tenth section was mounted and stained with Cresyl violet, a Nissl stain.
RESULTS
Histological analysis Every animal receiving an M D surgery displayed extensive cell loss within the nucleus and no animals were excluded on histological grounds. As there was no discernable difference between the location and extent of the MDpre and MDpost surgeries these two groups are described together. Fig. 1 shows the largest and smallest of the MDpre and MDpost lesions. All 17 animals showed extensive damage throughout nucleus medialis dorsalis, the N M D A causing the loss of virtually every neuronal cell in the affected areas (Fig. 2). In all cases there was shrinkage of tissue associated with the lesion. Estimates of the extent of the lesions, based on planimetric measurements of four standard sections, indicated that they varied from 72 ~o to 100Yo (median 88~o) of the total area of MD. The only other regions to show consistent cell loss were the midline paraventricular nucleus, those parts of the parataenial nucleus lying immediately in front of MD, and the medial edges of nucleus lateralis dorsalis
•
/
-
!
"--~.
~
"\\
--
tul~-I "
L
I
6
I
~
,
!
)
\\
ir
if r
fl l
43
Fig. 2. Photomicrograph of coronal section (Nissl stain) showing typical appearance of MD lesion following injection of NMDA. H, hippocampus; LD, nucleus lateralis dorsalis, MD, nucleus medialis dorsalis.
(Fig. 1). In nine cases the lesion extended dorsally to involve the lateral habenula. Bilateral cell loss was seen in the caudal half of the anterior dorsal nucleus in six cases while unilateral damage was observed in a further seven cases. Slight, unilateral cell loss was found in the most caudal part of the anterior ventral nucleus in five cases. The anterior medial nucleus was always unaffected. In four cases there was a very restricted region of cell loss in the dentate gyrus associated with the injection tract.
Delayed nonmatching-to-position The effects of MD lesions prior to acquisition (MDpre vs. Normal Controls) Acquisition. There was no evidence that the MDpre (n -- 10) group were slower to acquire the D N M P task than the Normal Controls (n = 27). Indeed, the M D p r e group were significantly faster at reaching the magazine
training criterion, and performed more accurately than the Normal Control group on the 0 - 4 s condition of the D N M P task. The median number of sessions to reach the magazine training criterion for the MDpre and Normal Control groups was 4.5 and 7 respectively (Mann-Whitney, U=66.5, P = 0 . 0 3 6 , two-tailed). The corresponding scores for the auto shaping procedure were 8.5 (MDpre) and 10 (Normal Controls; U-- 110, P > 0 . 1 ) . Two animals, one from each group, which failed to learn the autoshaping procedure were excluded from further testing. This left group sizes of 9 (MDpre) and 26 (Normal Controls). Further comparisons found no evidence of a difference in the number of sessions required to reach the 0 s and 2 s performance criterion (both P > 0 . 1 ) . While the MDpre group performed more accurately than the Normal Controls on the 0 - 4 s condition (overall mean percent correct, MDpre = 96.2, Normal Controls = 94.0; t33 = 2.37, P<0.025), there was no group
Fig, 1. Coronal sections showing the MDpre and MDpost cases with the largest (diagonal lines) and smallest (black) lesions. The numbers refer to the approximate corresponding levels in the atlas of Pellegrino and Cushman29. F, fornix; H, hippocampus.
44 difference for the subsequent 0-8 s condition (overall mean MDpre = 94.5, Normal Controls = 95.2; t < 1). Delay performance. The 9 MDpre and the 26 Normal Control rats were compared on their performance over the six D N M P sessions with retention delays of 0-32 s (Fig. 3). Comparisons using a variety of measures of
accuracy, bias, and general responsivity failed to reveal any difference between the two groups. The only exception was provided by evidence that the mean latency to first nose-poke was significantly slower in the MDpre group (P < 0.05). Accuracy measures. Although percent correct pro-
Normal Controls vs MDpre
N o r m a l Controls
I(X) '
.,,.
ii1) ¸ ~
fit)'
i
J,O
'
2()' ,.,
L)
2
4
X
16
32
()
"~
S
4
,
Delay Iseconds)
,%
32
Ih
Delay (seconds)
I(X)
I(X)
"
80
80
"
6O
,~
60 40
40
0
20
]
,
0
"~
4
8
i
r
16
32
0
T
i
,
o 2 4
i
i
i
8
16
32
Delay (seconds)
Delay (seconds)
10o
100 80
80
6O
"
v
60 40 40
20
oT , o 2 4
0 8
16
Delay (seconds)
32
,
|
i
0 2 a
~
8
t
16
''
'1
32
Delay (seconds)
Fig. 3. A c c u r a c y (left column) and bias (right c o l u m n ) indices as a function of lesion ( N o r m a l C ont rol s (n = 26) vs. M D p r e (n = 9) and delay. See text for e x p l a n a t i o n o f indices. All scores have been c o n v e r t e d to read from 0 to 100. In the c a s e of the bias scores, 100 r e p r e s e n t s a c o m p l e t e bias an d 0 neutral, or no bias.
45 vided the expected effect of delay (F5,165 =244.66, P<0.001), there was no evidence of a group effect ( F < 1) or of an effect of day ( F < 1). No other comparisons were significant at the P = 0.05 level (Fig. 3). Similar results were found for the two T S D measures, A' and SI. In both cases there was a highly significant effect of delay (both P<0.001), but no evidence of a group effect (maximum F1,33 = 1.57) or of a group × delay interaction (maximum F5,165 = 1.90). Bias indices. None of the three bias indices (Iy, B", RI) provided evidence of a group effect at the P = 0.05 level (Fig. 3). There were, however, delay effects for all three indices (all P < 0.001). The only significant interaction concerned the group × delay effect for index RI (F5,]65=2.43; P=0.04). This indicated a greater increase in response bias among the Normal Controls with delays of over 4 s (Fig. 3). General responsivity. The various measures of latency all indicated that the M D p r e group were slightly slower to respond than the Normal Controls. This difference was, however, only significant for the Latency to First Magazine Response (F1,33 = 11.53, P < 0 . 0 0 5 ) as the other two measures (Latency to Sample, Choice Latency) failed to reach significance (both 0.1 > P > 0.05). Although the MDpre group made relatively more 'misses' with the longer delays than the control group, as reflected in the group × delay interaction (F5,165 = 2.52, P < 0.05), there was no clear overall group difference (0.1 > P > 0.05). Similarly, there was no evidence that the two groups differed in their overall magazine response rate ( F < 1).
The effects of MD lesions after acquisition (Mdpost vs. Controls versus MDpre) Following the second series of surgeries, a total of 7 M D p o s t and 8 Control animals, as well as the 9 M D pre animals, were retested on the D N M P task. After a single practice session ('0' delays), all animals received 15 sessions on the final condition (variable retention intervals of 0-32 s). The mean percent correct score for each of the 15 postoperative sessions is shown in Fig. 4. An analysis of variance using this data indicated that while there was a marked effect of session ( F 1 4 , 2 9 4 = 6 . 1 7 , P < 0.001), there was no clear group effect (F2,21 = 3.30, 0 . 1 > P < 0 . 5 ) and no group×session interaction ( F < 1). As can be seen from Fig. 4, the suggestion of a group effect came from the high performance of the M D p r e group and not from any impairment associated with the M D p o s t group. More detailed comparisons using the full range of response measures focussed on the final five sessions where the animals maintained a stable level of perform-
100
I--
U
90
I.I.I
tv
O
u
8o
I---
Z
7o
CONTROLS
u ~
•
MOpre MDpost
6O
O-
i
1
2
i
i
i
i
i
i
i
,
,
J
,
,
,
3
4
5
6
7
S
9
I0
11
12
13
14
15
SESSIONS Fig. 4. M e a n percent correct over each of the 15 postoperative sessions using all six delays, for the three experimental groups (Controls (n = 8) vs. M D pre (n = 9) vs. M D p o s t (n = 7)).
ance (Fig. 4). Initial comparisons between just the MDpre and M D p o s t groups for these five sessions found no evidence of a group effect for the three accuracy measures (percent correct, A', SI; F~,14 = 1.38, F < 1, F < 1, respectively). In view of this lack of difference between the MDpre and M D p o s t animals, the two groups were combined prior to subsequent comparisons with the Control group. This provided the largest group of animals with M D lesions and, hence, a more powerful test of the effects of this surgery. Accuracy measures. There was a highly significant effect of delay for percent c o r r e c t (F5,1t0 = 106.8, P<0.001), but no group ( F < I ) or g r o u p × d e l a y (Fs,~0 = 1.51) effect. The results for the two T S D measures (A' and SI) were very similar to each other. As expected they both revealed a highly significant effect of delay ( P < 0.001), but there was also an indication of a possible group effect (both 0.1 > P > 0.05). In addition, there was a significant group × delay interaction for both measures (A', F5,110 = 2.96, P < 0 . 0 5 ; SI, F5,11 o = 3.40, P<0.01). It can, however, be seen that for both indices (A' and SI) these interactions reflected poorer performance by the Control animals at the longest delay intervals (Fig. 5). Bias indices. None of the three bias indices (Iy, B", RI) provided evidence of a group effect (all F < 1). There were, however, delay effects for all three indices (all P<0.001). There were no significant interactions between group and delay for any of these three measures (Fig. 5). General responsivity. There were no significant group differences or any significant group × delay interactions for any of the latency measures, for the number of 'misses' or the overall magazine response rate.
46
M D combined vs Controls
I00"
Ioo. 8O
i
80' g
60" M D comb0ned
20 o
0 2 4
8
16
32
0 2 4
8
16
32
Delay (seconds)
Delay (seconds) 100"
100 80
80"
60
60
h 40
4O
20
T
!
i
~
0 2 4
i
i
i
8
16
32
0
i
!
'1
0 2 4
t
8
Delay (seconds)
16
i2
3
Delay (seconds) 100"
100"
80" 80" 60"
r,
60
40" 40
0
20"
' 0
' 2
' 4
P' 8
' 16
3'2
Delay (seconds)
0
i
0 2 z
8
16
32
Delay (~econds)
Fig. 5.
Accuracy (left column) and bias (right column) indices as a function of lesion (Controls vs. M D p r e and M D p o s t combined) for all of scores from the last five sessions taken together. See text for explanation of indices. All scores have been converted to read from 0 to 100. In the case of the bias scores, 100 represents a complete bias and 0 neutral, or no bias.
Lever reversal task
The last part of the study used a spatial (lever) discrimination task with reversals between specified ses-
sions. During the running o f this t a s k one o f the M D post animals b e c a m e ill and its results were excluded f r o m the final analyses. T h e m e a n percent correct responses o f the three groups over the 16 sessions were:
47 I00 F=U r,1
0 U I--
9O 8o 7o
60 50
7' uu
40 30
IX
20
R
I
MOpre/MDpos t (combined)
R
REVERSAL
I0 0
I
I
I
I
I
I
I
2
3
4
5
6
I
I
I
I
I
I
I
I
I
J
7
8
9
10
II
12
13
14
15
16
SESSIONS Fig. 6. Mean percent correct scores for the Control and combined MD groups (MDpre plus MDpost) on the lever discrimination and reversal task. Sessions where the correct lever changed from the preceding session are labelled as R.
MDpre = 69.8, MDpost = 62.9, and Controls = 70.9. Excluding the first session, these scores could be divided between those six sessions where the correct lever was in the same ('consistent') position as in the previous session (mean MDPre 75.6~o, MDpost 72.3~o, Controls 78.1~o), and those nine sessions where the correct lever reversed side (mean MDpre 66.7 ~o, MDpost 57.8~o, Controls 68.4~o). In view of their lack of difference on the D N M P task the percent correct scores of two M D groups were first compared with each other. As no significant difference was found for these two groups for either the consistent ( t < 1) or the reversal (tl3 = 2 . 0 1 , 0.1 > P > 0.05) sessions they were combined prior to comparison with the Control group. Subsequent comparisons between the overall scores of the Control and the combined MD groups (Fig. 6) showed that the animals performed considerably worse on the reversal sessions (F1,2~ -- 53.91, P < 0.001). There was, however, no group difference (F1,21 = 1.98), nor was there evidence of an interaction ( F < 1) between lesion and type of session (same or reversal). Further comparisons using the data from just the consistent sessions or just the reversal sessions showed that there was an effect of session (both conditions, P < 0.001) but no lesion or lesion x session interaction (P>0.1). Finally, there was no evidence of a group difference on the very first acquisition session (t< 1).
DISCUSSION
A number of previous studies have examined the effects of M D lesions on spatial memory in rats. The findings have, however, been inconsistent. Marked impairments were reported for tests of both spontaneous and reinforced alternation 4°'42, and for radial arm maze
performance 36-38. Less clear-cut impairments have been described for other tests of spatial alternation 8, as well as spatial reversal tasks 25, and for a modified radial arm maze 18. Finally, a number of studies have found little or no evidence that M D damage disrupts spatial tasks, including spatial alternation 13'14, spatial reversal 39, the Morris water maze and the radial arm maze6,20,28. The present findings were consistent with the latter group of studies. That is, there was no evidence that complete or near-complete neurotoxic lesions of MD produced any impairment on the acquisition or performance of a spatial test of working memory, delayed nonmatching-to-position. Indeed, following a detailed analysis of task performance the only group differences on the D N M P task reflected slightly superior performances by animals with M D lesions (during initial acquisition and for indices A' and SI over the longer delays), along with somewhat slower response latencies during initial acquisition. The lack of effect on D N M P performance was found whether the lesions were made before or after acquisition of the task. A similar lack of lesion effect was found for a second spatial learning task involving spatial discriminations and their subsequent reversals. One possible explanation for the inconsistencies in the literature on the effects of M D lesions on spatial memory concerns differential damage to fibre tracts. Unfortunately, a comparison between those studies using conventional lesion techniques and those using cytotoxins suggests that this is not the explanation. Thus deficits have been reported in some studies of spatial memory using neurotoxins 18,37,38 but not in others 6'7'14, including the present one. It should also be added that these inconsistencies do not simply reflect the use of different tests of spatial memory. A related explanation concerns differences in the extent of damage to regions adjacent to MD. Of particular interest are the anterior thalamic nuclei in the light of their dense hippocampal afferents. With this possibility in mind Hunt and Aggleton 14 reviewed the effects of M D lesions in rats on spatial working memory. Evidence was found of a link between anterior thalamic damage and the presence of an 'MD' impairment 14, although this analysis was limited by a lack of histological detail in some of the papers reviewed. The present findings can be seen as consistent with this proposal i.e. that anterior thalamic damage is of critical importance. Although the present MD lesions often damaged the medial portions of nucleus lateralis dorsalis they consistently spared all of the anterior medial nucleus and the large majority of the anterior ventral nucleus.
48 The failure of the MD lesions to affect either the acquisition or subsequent performance of the D N M P task should be considered in the light of the association between this region and the prefrontal cortex. As has been pointed out, the D N M P task shares many demands with tests of delayed alternation and delayed response, tasks known to be highly sensitive to prefrontal cortex damage in monkeys. Furthermore, some studies have shown that M D damage can disrupt delayed response 33 or delayed alternation 15 performance in the monkey, although this has not always been observed 4'3°. In those studies that have found deficits it has been assumed that this reflects the close relationship between M D and the prefrontal cortex. With this relationship in mind it is clearly of interest that rats with medial prefrontal lesions are also often poor at delayed alternation and cued delayed response tasks 19. Of particular note is the finding ~° that rats with aspiration lesions of the medial prefrontal cortex are impaired on an automated delayed matching-toposition task (DMP), very similar to that used in the present study. That study found that rats with medial prefrontal lesions could be subdivided into two groups; one group showing delay dependent deficits, the other group being severely impaired at all delays m. Neither of these patterns of performance were found in the present study. Evidence that damage to the prelimbic area within the prefrontal cortex was closely linked to the extent of the D M P deficit ~° may, however, help explain why M D damage does not mirror the effects of prefrontal damage on such tasks. The prelimbic area, along with the adjacent medial orbital area, is the only prefrontal region to receive direct, heavy projections from the hippocampus as well as M D 16'21. Given that the hippocampus is vital for normal D N M P and D M P performance 2'3'9 it is quite likely that the prefrontal lesion effect reflects the importance of the hippocampal, but not the MD, inputs 8'2°. Finally, the discovery that lesions of nucleus medialis dorsalis need not affect D N M P performance, may be compared with the effects of other limbic lesions on this same task. In particular, it has repeatedly been found that lesions in either the hippocampus or the fornix produce marked performance deficits 2'3'9. The opposite dissociation has, however, been found for a nonspatial delayed nonmatching-to-sample task m4. That is, while hippocampal lesions have no apparent effect, rats with ibotenate lesions of medialis dorsalis were impaired. These data, that provide evidence of a double dissociation, indicate that the contributions of these limbic regions to normal memory are likely to prove very different.
ACKNOWLEDGEMENTS
This research was supported by the M.R.C. The authors wish to thank the assistance of A. Ennaceur, P. Hunt, S. Lloyd, C. Mullaney, M. Rolling, and S. Whiteley.
REFERENCES 1 Aggleton, J.P., Hunt, P.R. and Rawlins, J.N.P., The effects of hippocampal lesions upon spatial and non-spatial tests of working memory, Behav. Brain Res., 19 (1986) 133-146. 2 Aggleton, J.P., Keith, A.B., Rawlins, J.N.P., Hunt, P.R. and Sahgal, A., Removal of the hippoeampus and transection of the fornix produce comparable deficits on delayed non-matching to position by rats, Behav. Brain Res., 52 (1992) 61-71. 3 Aggleton, J.P., Keith, A.B. and Sahgal, A., Both fornix and anterior thalamic, but not mammillary, lesions disrupt delayed nonmatching-to-position memory in rats, Behav. Brain Res., 44 (1991) 151-161. 4 Aggleton, J.P. and Mishkin, M,, Visual recognition impairment following medial thalamic lesions in monkeys, Neuropsyehologia, 21 (1983) 189-197. 5 Bachevalier, J. and Mishkin, M., Visual recognition imprairment follows ventromedial but not dorsolateral prefrontal lesions in monkeys, Behav. Brain Res., 20 (1986) 249-261. 6 Beracochea, D.J., Jaffard, R. and Jarrard, L.E., Effects of anterior or dorsomedial thalamic ibotenic lesions on learning and memory in rats, Behav. Neural Biol., 51 (1989) 364-375. 7 Beracochea, D.J. and Krazem, A., Effects of mammillary body and mediodorsal thalamic lesions on elevated plus maze expoloration, Neuroreport, 2 (1991) 793-796. 8 Brito, G.N.O., Thomas, G.J., Davis, B.J. and Gingold, S.I., Prelimbic cortex, mediodorsal thalamus, septum, and delayed alternation in rats, Exp. Brain Res., 46 (1982) 52-58. 9 Cramon, von D.Y., Hebel, N. and Schuri, U., A contribution to the anatomical basis of thalamic amnesia, Brain, 108 (1985) 9931008. 10 Dunnett, S.B., Role of prefrontal cortex and striatal output systems in short-term memory deficits associated with ageing, basal forebrain lesions, and cholinergic-rich grafts, Canad. J. Psvchol., 44 (1990) 210-232. 11 Frey, P.W. and Colliver, J.A., Sensitivity and responsiveness measures for discrimination learning, Learn. Motiv.. 4 (1973) 327-342. 12 Gaffan, D. and Watkins, S., Mediodorsal thalamic lesions impair long-term visual associative memory in macaques, Eur. Y. Neurosci., 3 (1991) 615-620. 13 Greene, E. and Naranjo, J.N., Thalamic role in spatial memory, Behav. Brain Res., 19 (1986) 123-131. 14 Hunt, P.R. and Aggleton, J.P., Medial dorsal thalamic lesions and working memory in the rat, Behav. Neural Biol., 55 (1991) 227-246. 15 Isseroff, A., Rosvold, H.E., Galkin, T.W. and Goldman-Rakic, P.S., Spatial memory impairments following damage to the mediodorsal nucleus of the thalamus in rhesus monkeys, Brain Res., 232 (1982) 97-113. 16 Jay, T.M. and Witter, M.P., Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of the anterograde transport of Phaseolus vulgarisLeucoagglutinin, J. Comp. Neurol., 313 (1991) 574-586.
49 17 Joyce, E.M. and Robbins, T.W., Frontal lobe function in Korsakoff and non-Korsakoff alcoholics: planning and spatial working memory, Neuropsychologia, 29 (1991) 709-723. 18 Kessler, J., Markowitsch, H.J. and Otto, B., Subtle but distinct impairments of rats with chemical lesions in the thalamic mediodorsal nucleus, tested in a radial arm maze, J. Comp. Physiol. Psychol., 96 (1982) 712-720. 19 Kolb, B., Functions of the frontal cortex of the rat: a comparative review, Brain Res. Rev., 8 (1984) 65-98. 20 Kolb, B., Pittman, K., Sutherland, R.J. and Whishaw, I.Q., Dissociation of the contributions of the prefrontal cortex and dorsomedial thalamic nucleus to spatially guided behavior in the rat, Behav. Brain Res., 6 (1982) 365-378. 21 Krettek, J.E. and Price, J.L., The cortical projections of the mediodorsal nucleus and adjacent thalamic nuclei in the rat, J. Comp. Neurol., 171 (1977) 157-192. 22 Markowitsch, H.J., Thalamic mediodorsal nucleus and memory: a critical evaluation of studies in animals and man, Neurosci. Biobehav. Rev., 6 (1982) 351-380. 23 Markowitsch, H.J., Diencephalic amnesia: a reorientation toward tracts? Brain Res. Rev., 13 (1988) 351-370. 24 Means, L.W., Harrell, T.H., Mayo, E.S. and Alexander, G.B., Effects of dorsomedial thalamic lesions on spontaneous alternation, maze activity and runway performance in the rat, Physiol. Behav., 12 (1974) 973-979. 25 Means, L.W., Hershey, A.E., Waterhouse, G.J. and Lane, C.J., Effects of dorsomedial thalamic lesions on spatial discrimination reversal in the rat, Physiol. Behav., 14 (1975) 725-729. 26 Milner, B., Some cognitive effects of frontal-lobe lesions in man, Phil. Trans. R. Soc. Lond. B., 298 (1982) 211-226. 27 Olney, J.W., Excitotoxins: an overview. In K. Fuxe, P. Roberts and R. Schwarcz (Eds.), Excitotoxins, Vol. 39, Wenner-Gren International Symposium Series, Plenum, New York, 1984, pp. 8296. 28 Olton, D.S., The function of the septo-hippocampal connections in spatially organized behavior. In Functions of the septohippocampal system (CIBA foundation symposium No. 58), Elsevier, Amsterdam, 1978, pp. 327-349. 29 Pellegrino, L.J. and Cushman, A.J., Stereotaxic Atlas of the Rat Brain, Appleton-Century-Crofts, New York, 1967.
30 Peters, R.H., Rosvold, H.E. and Mirsky, A.F., Effect ofthalamic lesions upon delayed response-type tests in rhesus monkey, J. Comp. Physiol. PsychoL, 49 (1956) 111-116. 31 Sahgal, A., Vasopressin retards the acquisition of positively reinforced lever pressing in homozygous Brattelboro rats, Regul. Pept., 5 (1983) 317-326. 32 Sahgal, A., Some limitations of indices derived from signal detection theory: evaluation of an alternative index for measuring bias in memory tasks, Psychopharmacology, 91 (1987) 517520. 33 Schulman, S., Impaired delay response from thalamic lesions, Arch. Neurol., 11 (1964)477-499. 34 Speedie, L.J. and Heilman, K.M., Amnestic disturbance following infarction of the left dorsomedial nucleus of the thalamus, Neuropsychologia, 20 (1982) 597-604. 35 Staubli, U., Schottler, F. and Nejat-Bina, D., Role of the dorsomedial thalamic nucleus and piriform cortex in processing olfactory information, Behav. Brain Res., 25 (1987) 117-129. 36 Stokes, K.A. and Best, P.J., Mediodorsal thalamic lesions impair radial maze performance in the rat, Behav. Neurosci., 102 (1988) 294-300. 37 Stokes, K.A. and Best, P.J., Mediodorsal thalamus lesions in rats impair radial-arm maze performance in a cued environment, Psychobiology, 18 (1990) 63-67. 38 Stokes, K.A. and Best, P.J., Response biases do not underlie the radial maze deficit in rats with mediodorsal thalamus lesions, Behav. Neural Biol., 53 (1990) 334-345. 39 Tigner, J.C., The effects of dorsomedial thalamic lesions on learning, reversal, and alternation behavior in the rat, Physiol. Behav., 12 (1974) 13-17. 40 Vicedomini, J.P., Corwin, J.V. and Nonneman, A.J., Behavioral effects of lesions to the caudate nucleus or mediodorsal thalamus in neonatal, juvenile, and adult rats, Physiol. Psychol., 10 (1982) 246-250. 41 Victor, M., Adams, R.D. and Collins, G.H., The WernickeKorsakoff Syndrome, Blackwell, Oxford, 1971. 42 Weiss, B.J. and Means, L.W., A comparison of the effects of medial frontal, dorsomedial thalamic, and combination lesions on discrimination and spontaneous alternation in the rat, Physiol. Psychol., 8 (1980) 325-329.
39
BBR 01433
Lack of effect of dorsomedial thalamic lesions on automated tests of spatial memory in the rat N . N e a v e a, A. S a h g a l b a n d J . P . A g g l e t o n a aDepartment of Psychology, University of Durham, Durham (UK) and hMR C Neuroehemical Pathology Unit, Newcastle General Hospital Ne wcastle-upon-Tyne (UK) (Received 28 August 1992) (Revised version received 13 January 1993) (Accepted 14 January 1993)
Key words: Nucleus medialis dorsalis; Spatial memory; Nonmatching-to-position; Amnesia; Thalamus
The effects of cytotoxic lesions of nucleus medialis dorsalis on tests of spatial memory were examined in the rat. Extensive lesions of the nucleus did not impair either the acquisition or the subsequent performance of an automated test of working memory, delayed nonmatchingto-position. Detailed analysis of the animals' performance over varying retention delays failed to reveal any evidence of a deficit. The same animals performed normally in a spatial discrimination task and its subsequent reversals. The present results can be contrasted with those from animals with hippocampal system damage.
INTRODUCTION
Interest in the contribution of the thalamic nucleus medialis dorsalis (MD) to learning and memory derives from two main sources. First, neuropathological studies have linked damage in M D with diencephalic amnesias of varied aetiology 22'34'41. Unfortunately, the lack of cases with damage confined to M D coupled with the close proximity of M D to other nuclei or fibre tracts also implicated in amnesia 9'23, has left uncertain the contribution of this nucleus to memory processes. Second, M D has very dense, reciprocal connections with the prefrontal cortex. As a consequence it is assumed that an understanding of the prefrontal cortex will, in part, depend on an understanding of M D and vice versa. Evidence that parts of the prefrontal cortex have mnemonic functions s'1°'26 therefore, indirectly, implicates M D in similar functions. The present study examined the performance of rats with intrathalamic injections of N-methyl-D-aspartic acid ( N M D A ) on an automated test of working memory, delayed nonmatching-to-position. The rationale for this experiment arose from the opportunity to produce selective M D lesions with a neurotoxin that limCorrespondence: J.P. Aggleton, Department of Psychology, University of Durham, South Road, Durham DH1 3LE, UK.
its damage to fibres of passage 27. The particular memory task, delayed nonmatching-to-position (DNMP), was selected for several reasons. It has recently been confirmed that automated tests of spatial working memory are sensitive to diencephalic amnesia in humans iv. Furthermore, the demands of the D N M P task share many features with delayed response and delayed alternation. As these two tasks are the benchmark tests for prefrontal cortex function in animals, the D N M P task is particularly appropriate for assessing the effects of M D damage. Finally, the D N M P task may help resolve the seemingly inconsistent effects of M D damage on spatial memory in rats 8'14'20'25'26'36. It is for this final reason that the rats were also tested on a spatial discrimination task after completing the D N M P task. For this discrimination task the animals were reinforced for selecting the correct lever in an operant chamber, the position of the correct choice remaining constant throughout a session but changing in a random manner across sessions. In the D N M P task the animal is trained to press a sample lever, and, after a variable delay, is presented with a choice of two levers, one of which was the sample. The animal is rewarded for selecting the lever that was not the previous sample i.e. nonmatching. Using this task it is not only possible to derive forgetting curves for individual animals but it is also possible to
40 assess performance using a variety of accuracy and bias indices 1~,3t,32 derived from the theory of signal detection (TSD). A number of previous studies have revealed that this task is sensitive to the effects of hippocampal system damage and to prefrontal cortex damage in the rat 3'~°, although the effects of MD damage are unknown. In the light of recent proposals concerning the contribution of M D to learning and memory 12'35, the present study examined the effects of M D lesions both on the acquisition of the D N M P task (MDpre) and on the performance of rats that had received surgery after D N M P acquisition (MDpost).
MATERIALS AND M E T H O D S
Subjects The study involved 37 naive, male rats of the pigmented DA strain (Bantin and Kingman, Hull). Throughout the period of the experiment the animals were housed individually under diurnal conditions (14 h light/10 h dark), all testing occurring at a regular time during the light period. The animals were fed approximately 15 g of laboratory diet (Labsure ERM) daily so that they did not drop below 8 5 ~ of normal body weight. At the start of testing the animals were aged approximately 4 months and weighed between 200 and 300 g. All animals had free access to water.
Apparatus Testing was carried out in 7 operant chambers (Campden Instruments Ltd., Loughborough) under the online control of 2 Spider microprocessors (Paul Fray Ltd., Cambridge). Each chamber was fitted with two retractable levers situated 7.5 cm either side of a central food tray. This food dispenser, which delivered 45 mg pellets (Campden Instruments), had a hinged perspex lid at which nose pokes could be recorded. A light was located inside the food dispenser and a house light was located in the centre of the roof. A further light was situated above each lever.
Delayed nonmatching-to-position Approximately 2 weeks after 10 animals had received M D surgery, all subjects (10 MDpre and 27 Normal Controls) were given the same, standardized training protocol. (I) Magazine training. The animals were allowed to eat ten reinforcement pellets that had been placed in the food tray with all of the house lights on. Training stopped when the animal ate all pellets during a period of 10 min on three consecutive daily sessions.
(2) Autoshaping. A randomly selected lever was inserted into the chamber and the light above it lit. The lever remained 'out' until the animal responded by pressing, at which point the lever was retracted, the food tray light came on and a single reward was delivered 3~. If no response was made after 10 s the lever was retracted. Autoshaping finished when the animal had responded on at least 50 out of 60 lever presentations over three consecutive sessions. (3) Delayed nonmatching-to-position. The testing procedure has been described in full in a number of recent publications 2"3"31. During the first stage of testing ('0' s delay) a trial would consist of the following sequence of events. After a 10 s inter-trial interval, either the left or the right hand lever emerged and the corresponding lever light illuminated. The rat then had to respond to the lever within 10 s, upon which the lever would retract, a pellet would be dispensed and the magazine tray illuminated. The subject then had to approach the tray and, within 10 s, operate the magazine flap. Both levers then emerged, and the stimulus lights above them were illuminated. The rat now had to respond, for a food pellet, to the lever which had not appeared as the sample; these choice responses also had to be made within 10 s. Incorrect responses, or a failure to respond, resulted in a 'time-out' of 10 s, during which all lights in the chamber were switched off, and both levers retracted; these were the only occasions when the chamber was unlit. A session consisted of 60 trials. A formal procedure was used to encourage an animal to respond on the opposite lever if it had developed a pronounced side bias. Delays between the sample and choice lever presentations were introduced once rats had reached a criterion of 900/0 correct on three consecutive sessions. For all subsequent trials magazine responses following the sample presentation were ineffective (but were recorded) until the appropriate delay interval had lapsed; the first response after this resulted in the choice levers being presented, providing this response occurred within 10 s of the end of the delay. The rats were first given a retention delay of 2 s with the same performance criterion as 0 s. Following this the number of trials was increased to 96 and each session contained an equal mixture of 0, 2 and 4 s (four sessions), 0, 2, 4 and 8 s (four sessions), and 0, 2, 4, 8, 16, and 32 s (six sessions) delay intervals. (4) Postoperative performance. Following completion of this protocol, 15 of the Normal Controls (now 7 months old) received surgeries (7 MDpost, 8 Controls). Approximately one month after surgery, these 15 rats, and 9 of the rats operated prior to acquisition (MDpre), were retested on the D N M P task. All ani-
41 mals received 15 sessions with a balanced, mixture of 0-32 s (as in the final stage of acquisition).
Spatial discriminations and reversals One day after completing the D N M P task, the animals were trained on a spatial discrimination task in the same operant chamber. For each session the animal was presented with both levers, one of which had been arbitrarily designated as S +, the other as S - . The pair of levers were extended until one was pressed (there were no 'miss' trials) at which time both were retracted. Responding on the correct (S + ) lever was followed by the delivery of one reward pellet. Responding on the incorrect lever (S - ) was recorded as an error and no reward given. The intertrial interval was set at 7 s. Each session consisted of 30 trials, the correct lever always being on the same side. The animals received 16 sessions, the position of the correct lever following a balanced, pseudorandom sequence.
Performance measures and analyses The data from the 0 to 32 s delay condition on the D N M P task were analysed to provide three indices of accuracy. These consisted of percent correct and two measures of sensitivity (A' and SI) derived from nonparametric signal detection models al. Three measures of bias (B", RI, and Iy) were also calculated aa'32. The first of these (B") reflects perceptual bias, the second (RI) reflects response bias, while the third (Iy) contrasts accuracy between the two levers 1~'32. These various measures are of interest as they may help distinguish between different types of impairments. Finally, estimates of general responsivity were also recorded. These included (a) latency to respond to the sample lever pooled across delays, since delay was unpredictable at the sampling stage, (b) latency to make the first magazine response ('nose-poke'), (c)rate of responding to the magazine flap, or nose-pokes/s, excluding the 0 delay condition, (d) choice (and average choice) latencies and (e) the number of missed trials. The justification and calculation of these signal detection measures and bias indices has been described in detail elsewhere 3'32. It should be noted that only the absolute values of B" and RI were analysed, since only the magnitude of the bias is of interest, rather than position or direction (for example, left vs. right position bias). The data were transformed as appropriate (arcsin: all accuracy and bias indices; logarithmic: latencies; square-root: misses) and analysed by parametric analysis of variance (ANOVA). For the sake of brevity, analyses failing to reach at least a P < 0 . 1 level are not presented in detail.
Surgical and histological procedures For the M D surgeries the animals were anaesthetised by intraperitoneal injection (4 ml/kg) of a solution containing 42 mg/ml of chloral hydrate and 9.7 mg/ml pentobarbitone sodium (Equithesin). Each animal was then placed in a stereotaxic frame (David Kopf Instruments, Tujunga). The scalp was cut and reflected to expose the skull. Following craniotomy, 0.36 #1 of 0.12 M N M D A (Sigma Chemical Company Ltd., Poole) dissolved in phosphate buffer (pH 7.2) was injected through a 1 #1 Hamilton Syringe in each hemisphere. The stereotaxic coordinates relative to ear-bar zero, with the incisorbar set at + 5.0 relative to the horizontal plane, were AP + 3.7, HT + 4.6, and LAT + 0.7. Each injection was made gradually over a 5-min period and the needle allowed to remain in situ for a further 5 min before being withdrawn. After completion of the lesion sulphanilamide powder was applied and the skin sutured. The animals acting as non surgical controls (Control group) were just injected with the same dose of Equithesin and allowed to recover. On completion of the experiments the rats were killed and perfused intracardially with 5 ~o formol-saline. The brains were then rapidly removed and placed in 5~o formol-saline. Subsequently, the brains were blocked, embedded in wax (Paraplast), and cut in 10-/~m coronal sections. Every tenth section was mounted and stained with Cresyl violet, a Nissl stain.
RESULTS
Histological analysis Every animal receiving an M D surgery displayed extensive cell loss within the nucleus and no animals were excluded on histological grounds. As there was no discernable difference between the location and extent of the MDpre and MDpost surgeries these two groups are described together. Fig. 1 shows the largest and smallest of the MDpre and MDpost lesions. All 17 animals showed extensive damage throughout nucleus medialis dorsalis, the N M D A causing the loss of virtually every neuronal cell in the affected areas (Fig. 2). In all cases there was shrinkage of tissue associated with the lesion. Estimates of the extent of the lesions, based on planimetric measurements of four standard sections, indicated that they varied from 72 ~o to 100Yo (median 88~o) of the total area of MD. The only other regions to show consistent cell loss were the midline paraventricular nucleus, those parts of the parataenial nucleus lying immediately in front of MD, and the medial edges of nucleus lateralis dorsalis
•
/
-
!
"--~.
~
"\\
--
tul~-I "
L
I
6
I
~
,
!
)
\\
ir
if r
fl l
43
Fig. 2. Photomicrograph of coronal section (Nissl stain) showing typical appearance of MD lesion following injection of NMDA. H, hippocampus; LD, nucleus lateralis dorsalis, MD, nucleus medialis dorsalis.
(Fig. 1). In nine cases the lesion extended dorsally to involve the lateral habenula. Bilateral cell loss was seen in the caudal half of the anterior dorsal nucleus in six cases while unilateral damage was observed in a further seven cases. Slight, unilateral cell loss was found in the most caudal part of the anterior ventral nucleus in five cases. The anterior medial nucleus was always unaffected. In four cases there was a very restricted region of cell loss in the dentate gyrus associated with the injection tract.
Delayed nonmatching-to-position The effects of MD lesions prior to acquisition (MDpre vs. Normal Controls) Acquisition. There was no evidence that the MDpre (n -- 10) group were slower to acquire the D N M P task than the Normal Controls (n = 27). Indeed, the M D p r e group were significantly faster at reaching the magazine
training criterion, and performed more accurately than the Normal Control group on the 0 - 4 s condition of the D N M P task. The median number of sessions to reach the magazine training criterion for the MDpre and Normal Control groups was 4.5 and 7 respectively (Mann-Whitney, U=66.5, P = 0 . 0 3 6 , two-tailed). The corresponding scores for the auto shaping procedure were 8.5 (MDpre) and 10 (Normal Controls; U-- 110, P > 0 . 1 ) . Two animals, one from each group, which failed to learn the autoshaping procedure were excluded from further testing. This left group sizes of 9 (MDpre) and 26 (Normal Controls). Further comparisons found no evidence of a difference in the number of sessions required to reach the 0 s and 2 s performance criterion (both P > 0 . 1 ) . While the MDpre group performed more accurately than the Normal Controls on the 0 - 4 s condition (overall mean percent correct, MDpre = 96.2, Normal Controls = 94.0; t33 = 2.37, P<0.025), there was no group
Fig, 1. Coronal sections showing the MDpre and MDpost cases with the largest (diagonal lines) and smallest (black) lesions. The numbers refer to the approximate corresponding levels in the atlas of Pellegrino and Cushman29. F, fornix; H, hippocampus.
44 difference for the subsequent 0-8 s condition (overall mean MDpre = 94.5, Normal Controls = 95.2; t < 1). Delay performance. The 9 MDpre and the 26 Normal Control rats were compared on their performance over the six D N M P sessions with retention delays of 0-32 s (Fig. 3). Comparisons using a variety of measures of
accuracy, bias, and general responsivity failed to reveal any difference between the two groups. The only exception was provided by evidence that the mean latency to first nose-poke was significantly slower in the MDpre group (P < 0.05). Accuracy measures. Although percent correct pro-
Normal Controls vs MDpre
N o r m a l Controls
I(X) '
.,,.
ii1) ¸ ~
fit)'
i
J,O
'
2()' ,.,
L)
2
4
X
16
32
()
"~
S
4
,
Delay Iseconds)
,%
32
Ih
Delay (seconds)
I(X)
I(X)
"
80
80
"
6O
,~
60 40
40
0
20
]
,
0
"~
4
8
i
r
16
32
0
T
i
,
o 2 4
i
i
i
8
16
32
Delay (seconds)
Delay (seconds)
10o
100 80
80
6O
"
v
60 40 40
20
oT , o 2 4
0 8
16
Delay (seconds)
32
,
|
i
0 2 a
~
8
t
16
''
'1
32
Delay (seconds)
Fig. 3. A c c u r a c y (left column) and bias (right c o l u m n ) indices as a function of lesion ( N o r m a l C ont rol s (n = 26) vs. M D p r e (n = 9) and delay. See text for e x p l a n a t i o n o f indices. All scores have been c o n v e r t e d to read from 0 to 100. In the c a s e of the bias scores, 100 r e p r e s e n t s a c o m p l e t e bias an d 0 neutral, or no bias.
45 vided the expected effect of delay (F5,165 =244.66, P<0.001), there was no evidence of a group effect ( F < 1) or of an effect of day ( F < 1). No other comparisons were significant at the P = 0.05 level (Fig. 3). Similar results were found for the two T S D measures, A' and SI. In both cases there was a highly significant effect of delay (both P<0.001), but no evidence of a group effect (maximum F1,33 = 1.57) or of a group × delay interaction (maximum F5,165 = 1.90). Bias indices. None of the three bias indices (Iy, B", RI) provided evidence of a group effect at the P = 0.05 level (Fig. 3). There were, however, delay effects for all three indices (all P < 0.001). The only significant interaction concerned the group × delay effect for index RI (F5,]65=2.43; P=0.04). This indicated a greater increase in response bias among the Normal Controls with delays of over 4 s (Fig. 3). General responsivity. The various measures of latency all indicated that the M D p r e group were slightly slower to respond than the Normal Controls. This difference was, however, only significant for the Latency to First Magazine Response (F1,33 = 11.53, P < 0 . 0 0 5 ) as the other two measures (Latency to Sample, Choice Latency) failed to reach significance (both 0.1 > P > 0.05). Although the MDpre group made relatively more 'misses' with the longer delays than the control group, as reflected in the group × delay interaction (F5,165 = 2.52, P < 0.05), there was no clear overall group difference (0.1 > P > 0.05). Similarly, there was no evidence that the two groups differed in their overall magazine response rate ( F < 1).
The effects of MD lesions after acquisition (Mdpost vs. Controls versus MDpre) Following the second series of surgeries, a total of 7 M D p o s t and 8 Control animals, as well as the 9 M D pre animals, were retested on the D N M P task. After a single practice session ('0' delays), all animals received 15 sessions on the final condition (variable retention intervals of 0-32 s). The mean percent correct score for each of the 15 postoperative sessions is shown in Fig. 4. An analysis of variance using this data indicated that while there was a marked effect of session ( F 1 4 , 2 9 4 = 6 . 1 7 , P < 0.001), there was no clear group effect (F2,21 = 3.30, 0 . 1 > P < 0 . 5 ) and no group×session interaction ( F < 1). As can be seen from Fig. 4, the suggestion of a group effect came from the high performance of the M D p r e group and not from any impairment associated with the M D p o s t group. More detailed comparisons using the full range of response measures focussed on the final five sessions where the animals maintained a stable level of perform-
100
I--
U
90
I.I.I
tv
O
u
8o
I---
Z
7o
CONTROLS
u ~
•
MOpre MDpost
6O
O-
i
1
2
i
i
i
i
i
i
i
,
,
J
,
,
,
3
4
5
6
7
S
9
I0
11
12
13
14
15
SESSIONS Fig. 4. M e a n percent correct over each of the 15 postoperative sessions using all six delays, for the three experimental groups (Controls (n = 8) vs. M D pre (n = 9) vs. M D p o s t (n = 7)).
ance (Fig. 4). Initial comparisons between just the MDpre and M D p o s t groups for these five sessions found no evidence of a group effect for the three accuracy measures (percent correct, A', SI; F~,14 = 1.38, F < 1, F < 1, respectively). In view of this lack of difference between the MDpre and M D p o s t animals, the two groups were combined prior to subsequent comparisons with the Control group. This provided the largest group of animals with M D lesions and, hence, a more powerful test of the effects of this surgery. Accuracy measures. There was a highly significant effect of delay for percent c o r r e c t (F5,1t0 = 106.8, P<0.001), but no group ( F < I ) or g r o u p × d e l a y (Fs,~0 = 1.51) effect. The results for the two T S D measures (A' and SI) were very similar to each other. As expected they both revealed a highly significant effect of delay ( P < 0.001), but there was also an indication of a possible group effect (both 0.1 > P > 0.05). In addition, there was a significant group × delay interaction for both measures (A', F5,110 = 2.96, P < 0 . 0 5 ; SI, F5,11 o = 3.40, P<0.01). It can, however, be seen that for both indices (A' and SI) these interactions reflected poorer performance by the Control animals at the longest delay intervals (Fig. 5). Bias indices. None of the three bias indices (Iy, B", RI) provided evidence of a group effect (all F < 1). There were, however, delay effects for all three indices (all P<0.001). There were no significant interactions between group and delay for any of these three measures (Fig. 5). General responsivity. There were no significant group differences or any significant group × delay interactions for any of the latency measures, for the number of 'misses' or the overall magazine response rate.
46
M D combined vs Controls
I00"
Ioo. 8O
i
80' g
60" M D comb0ned
20 o
0 2 4
8
16
32
0 2 4
8
16
32
Delay (seconds)
Delay (seconds) 100"
100 80
80"
60
60
h 40
4O
20
T
!
i
~
0 2 4
i
i
i
8
16
32
0
i
!
'1
0 2 4
t
8
Delay (seconds)
16
i2
3
Delay (seconds) 100"
100"
80" 80" 60"
r,
60
40" 40
0
20"
' 0
' 2
' 4
P' 8
' 16
3'2
Delay (seconds)
0
i
0 2 z
8
16
32
Delay (~econds)
Fig. 5.
Accuracy (left column) and bias (right column) indices as a function of lesion (Controls vs. M D p r e and M D p o s t combined) for all of scores from the last five sessions taken together. See text for explanation of indices. All scores have been converted to read from 0 to 100. In the case of the bias scores, 100 represents a complete bias and 0 neutral, or no bias.
Lever reversal task
The last part of the study used a spatial (lever) discrimination task with reversals between specified ses-
sions. During the running o f this t a s k one o f the M D post animals b e c a m e ill and its results were excluded f r o m the final analyses. T h e m e a n percent correct responses o f the three groups over the 16 sessions were:
47 I00 F=U r,1
0 U I--
9O 8o 7o
60 50
7' uu
40 30
IX
20
R
I
MOpre/MDpos t (combined)
R
REVERSAL
I0 0
I
I
I
I
I
I
I
2
3
4
5
6
I
I
I
I
I
I
I
I
I
J
7
8
9
10
II
12
13
14
15
16
SESSIONS Fig. 6. Mean percent correct scores for the Control and combined MD groups (MDpre plus MDpost) on the lever discrimination and reversal task. Sessions where the correct lever changed from the preceding session are labelled as R.
MDpre = 69.8, MDpost = 62.9, and Controls = 70.9. Excluding the first session, these scores could be divided between those six sessions where the correct lever was in the same ('consistent') position as in the previous session (mean MDPre 75.6~o, MDpost 72.3~o, Controls 78.1~o), and those nine sessions where the correct lever reversed side (mean MDpre 66.7 ~o, MDpost 57.8~o, Controls 68.4~o). In view of their lack of difference on the D N M P task the percent correct scores of two M D groups were first compared with each other. As no significant difference was found for these two groups for either the consistent ( t < 1) or the reversal (tl3 = 2 . 0 1 , 0.1 > P > 0.05) sessions they were combined prior to comparison with the Control group. Subsequent comparisons between the overall scores of the Control and the combined MD groups (Fig. 6) showed that the animals performed considerably worse on the reversal sessions (F1,2~ -- 53.91, P < 0.001). There was, however, no group difference (F1,21 = 1.98), nor was there evidence of an interaction ( F < 1) between lesion and type of session (same or reversal). Further comparisons using the data from just the consistent sessions or just the reversal sessions showed that there was an effect of session (both conditions, P < 0.001) but no lesion or lesion x session interaction (P>0.1). Finally, there was no evidence of a group difference on the very first acquisition session (t< 1).
DISCUSSION
A number of previous studies have examined the effects of M D lesions on spatial memory in rats. The findings have, however, been inconsistent. Marked impairments were reported for tests of both spontaneous and reinforced alternation 4°'42, and for radial arm maze
performance 36-38. Less clear-cut impairments have been described for other tests of spatial alternation 8, as well as spatial reversal tasks 25, and for a modified radial arm maze 18. Finally, a number of studies have found little or no evidence that M D damage disrupts spatial tasks, including spatial alternation 13'14, spatial reversal 39, the Morris water maze and the radial arm maze6,20,28. The present findings were consistent with the latter group of studies. That is, there was no evidence that complete or near-complete neurotoxic lesions of MD produced any impairment on the acquisition or performance of a spatial test of working memory, delayed nonmatching-to-position. Indeed, following a detailed analysis of task performance the only group differences on the D N M P task reflected slightly superior performances by animals with M D lesions (during initial acquisition and for indices A' and SI over the longer delays), along with somewhat slower response latencies during initial acquisition. The lack of effect on D N M P performance was found whether the lesions were made before or after acquisition of the task. A similar lack of lesion effect was found for a second spatial learning task involving spatial discriminations and their subsequent reversals. One possible explanation for the inconsistencies in the literature on the effects of M D lesions on spatial memory concerns differential damage to fibre tracts. Unfortunately, a comparison between those studies using conventional lesion techniques and those using cytotoxins suggests that this is not the explanation. Thus deficits have been reported in some studies of spatial memory using neurotoxins 18,37,38 but not in others 6'7'14, including the present one. It should also be added that these inconsistencies do not simply reflect the use of different tests of spatial memory. A related explanation concerns differences in the extent of damage to regions adjacent to MD. Of particular interest are the anterior thalamic nuclei in the light of their dense hippocampal afferents. With this possibility in mind Hunt and Aggleton 14 reviewed the effects of M D lesions in rats on spatial working memory. Evidence was found of a link between anterior thalamic damage and the presence of an 'MD' impairment 14, although this analysis was limited by a lack of histological detail in some of the papers reviewed. The present findings can be seen as consistent with this proposal i.e. that anterior thalamic damage is of critical importance. Although the present MD lesions often damaged the medial portions of nucleus lateralis dorsalis they consistently spared all of the anterior medial nucleus and the large majority of the anterior ventral nucleus.
48 The failure of the MD lesions to affect either the acquisition or subsequent performance of the D N M P task should be considered in the light of the association between this region and the prefrontal cortex. As has been pointed out, the D N M P task shares many demands with tests of delayed alternation and delayed response, tasks known to be highly sensitive to prefrontal cortex damage in monkeys. Furthermore, some studies have shown that M D damage can disrupt delayed response 33 or delayed alternation 15 performance in the monkey, although this has not always been observed 4'3°. In those studies that have found deficits it has been assumed that this reflects the close relationship between M D and the prefrontal cortex. With this relationship in mind it is clearly of interest that rats with medial prefrontal lesions are also often poor at delayed alternation and cued delayed response tasks 19. Of particular note is the finding ~° that rats with aspiration lesions of the medial prefrontal cortex are impaired on an automated delayed matching-toposition task (DMP), very similar to that used in the present study. That study found that rats with medial prefrontal lesions could be subdivided into two groups; one group showing delay dependent deficits, the other group being severely impaired at all delays m. Neither of these patterns of performance were found in the present study. Evidence that damage to the prelimbic area within the prefrontal cortex was closely linked to the extent of the D M P deficit ~° may, however, help explain why M D damage does not mirror the effects of prefrontal damage on such tasks. The prelimbic area, along with the adjacent medial orbital area, is the only prefrontal region to receive direct, heavy projections from the hippocampus as well as M D 16'21. Given that the hippocampus is vital for normal D N M P and D M P performance 2'3'9 it is quite likely that the prefrontal lesion effect reflects the importance of the hippocampal, but not the MD, inputs 8'2°. Finally, the discovery that lesions of nucleus medialis dorsalis need not affect D N M P performance, may be compared with the effects of other limbic lesions on this same task. In particular, it has repeatedly been found that lesions in either the hippocampus or the fornix produce marked performance deficits 2'3'9. The opposite dissociation has, however, been found for a nonspatial delayed nonmatching-to-sample task m4. That is, while hippocampal lesions have no apparent effect, rats with ibotenate lesions of medialis dorsalis were impaired. These data, that provide evidence of a double dissociation, indicate that the contributions of these limbic regions to normal memory are likely to prove very different.
ACKNOWLEDGEMENTS
This research was supported by the M.R.C. The authors wish to thank the assistance of A. Ennaceur, P. Hunt, S. Lloyd, C. Mullaney, M. Rolling, and S. Whiteley.
REFERENCES 1 Aggleton, J.P., Hunt, P.R. and Rawlins, J.N.P., The effects of hippocampal lesions upon spatial and non-spatial tests of working memory, Behav. Brain Res., 19 (1986) 133-146. 2 Aggleton, J.P., Keith, A.B., Rawlins, J.N.P., Hunt, P.R. and Sahgal, A., Removal of the hippoeampus and transection of the fornix produce comparable deficits on delayed non-matching to position by rats, Behav. Brain Res., 52 (1992) 61-71. 3 Aggleton, J.P., Keith, A.B. and Sahgal, A., Both fornix and anterior thalamic, but not mammillary, lesions disrupt delayed nonmatching-to-position memory in rats, Behav. Brain Res., 44 (1991) 151-161. 4 Aggleton, J.P. and Mishkin, M,, Visual recognition impairment following medial thalamic lesions in monkeys, Neuropsyehologia, 21 (1983) 189-197. 5 Bachevalier, J. and Mishkin, M., Visual recognition imprairment follows ventromedial but not dorsolateral prefrontal lesions in monkeys, Behav. Brain Res., 20 (1986) 249-261. 6 Beracochea, D.J., Jaffard, R. and Jarrard, L.E., Effects of anterior or dorsomedial thalamic ibotenic lesions on learning and memory in rats, Behav. Neural Biol., 51 (1989) 364-375. 7 Beracochea, D.J. and Krazem, A., Effects of mammillary body and mediodorsal thalamic lesions on elevated plus maze expoloration, Neuroreport, 2 (1991) 793-796. 8 Brito, G.N.O., Thomas, G.J., Davis, B.J. and Gingold, S.I., Prelimbic cortex, mediodorsal thalamus, septum, and delayed alternation in rats, Exp. Brain Res., 46 (1982) 52-58. 9 Cramon, von D.Y., Hebel, N. and Schuri, U., A contribution to the anatomical basis of thalamic amnesia, Brain, 108 (1985) 9931008. 10 Dunnett, S.B., Role of prefrontal cortex and striatal output systems in short-term memory deficits associated with ageing, basal forebrain lesions, and cholinergic-rich grafts, Canad. J. Psvchol., 44 (1990) 210-232. 11 Frey, P.W. and Colliver, J.A., Sensitivity and responsiveness measures for discrimination learning, Learn. Motiv.. 4 (1973) 327-342. 12 Gaffan, D. and Watkins, S., Mediodorsal thalamic lesions impair long-term visual associative memory in macaques, Eur. Y. Neurosci., 3 (1991) 615-620. 13 Greene, E. and Naranjo, J.N., Thalamic role in spatial memory, Behav. Brain Res., 19 (1986) 123-131. 14 Hunt, P.R. and Aggleton, J.P., Medial dorsal thalamic lesions and working memory in the rat, Behav. Neural Biol., 55 (1991) 227-246. 15 Isseroff, A., Rosvold, H.E., Galkin, T.W. and Goldman-Rakic, P.S., Spatial memory impairments following damage to the mediodorsal nucleus of the thalamus in rhesus monkeys, Brain Res., 232 (1982) 97-113. 16 Jay, T.M. and Witter, M.P., Distribution of hippocampal CA1 and subicular efferents in the prefrontal cortex of the rat studied by means of the anterograde transport of Phaseolus vulgarisLeucoagglutinin, J. Comp. Neurol., 313 (1991) 574-586.
49 17 Joyce, E.M. and Robbins, T.W., Frontal lobe function in Korsakoff and non-Korsakoff alcoholics: planning and spatial working memory, Neuropsychologia, 29 (1991) 709-723. 18 Kessler, J., Markowitsch, H.J. and Otto, B., Subtle but distinct impairments of rats with chemical lesions in the thalamic mediodorsal nucleus, tested in a radial arm maze, J. Comp. Physiol. Psychol., 96 (1982) 712-720. 19 Kolb, B., Functions of the frontal cortex of the rat: a comparative review, Brain Res. Rev., 8 (1984) 65-98. 20 Kolb, B., Pittman, K., Sutherland, R.J. and Whishaw, I.Q., Dissociation of the contributions of the prefrontal cortex and dorsomedial thalamic nucleus to spatially guided behavior in the rat, Behav. Brain Res., 6 (1982) 365-378. 21 Krettek, J.E. and Price, J.L., The cortical projections of the mediodorsal nucleus and adjacent thalamic nuclei in the rat, J. Comp. Neurol., 171 (1977) 157-192. 22 Markowitsch, H.J., Thalamic mediodorsal nucleus and memory: a critical evaluation of studies in animals and man, Neurosci. Biobehav. Rev., 6 (1982) 351-380. 23 Markowitsch, H.J., Diencephalic amnesia: a reorientation toward tracts? Brain Res. Rev., 13 (1988) 351-370. 24 Means, L.W., Harrell, T.H., Mayo, E.S. and Alexander, G.B., Effects of dorsomedial thalamic lesions on spontaneous alternation, maze activity and runway performance in the rat, Physiol. Behav., 12 (1974) 973-979. 25 Means, L.W., Hershey, A.E., Waterhouse, G.J. and Lane, C.J., Effects of dorsomedial thalamic lesions on spatial discrimination reversal in the rat, Physiol. Behav., 14 (1975) 725-729. 26 Milner, B., Some cognitive effects of frontal-lobe lesions in man, Phil. Trans. R. Soc. Lond. B., 298 (1982) 211-226. 27 Olney, J.W., Excitotoxins: an overview. In K. Fuxe, P. Roberts and R. Schwarcz (Eds.), Excitotoxins, Vol. 39, Wenner-Gren International Symposium Series, Plenum, New York, 1984, pp. 8296. 28 Olton, D.S., The function of the septo-hippocampal connections in spatially organized behavior. In Functions of the septohippocampal system (CIBA foundation symposium No. 58), Elsevier, Amsterdam, 1978, pp. 327-349. 29 Pellegrino, L.J. and Cushman, A.J., Stereotaxic Atlas of the Rat Brain, Appleton-Century-Crofts, New York, 1967.
30 Peters, R.H., Rosvold, H.E. and Mirsky, A.F., Effect ofthalamic lesions upon delayed response-type tests in rhesus monkey, J. Comp. Physiol. PsychoL, 49 (1956) 111-116. 31 Sahgal, A., Vasopressin retards the acquisition of positively reinforced lever pressing in homozygous Brattelboro rats, Regul. Pept., 5 (1983) 317-326. 32 Sahgal, A., Some limitations of indices derived from signal detection theory: evaluation of an alternative index for measuring bias in memory tasks, Psychopharmacology, 91 (1987) 517520. 33 Schulman, S., Impaired delay response from thalamic lesions, Arch. Neurol., 11 (1964)477-499. 34 Speedie, L.J. and Heilman, K.M., Amnestic disturbance following infarction of the left dorsomedial nucleus of the thalamus, Neuropsychologia, 20 (1982) 597-604. 35 Staubli, U., Schottler, F. and Nejat-Bina, D., Role of the dorsomedial thalamic nucleus and piriform cortex in processing olfactory information, Behav. Brain Res., 25 (1987) 117-129. 36 Stokes, K.A. and Best, P.J., Mediodorsal thalamic lesions impair radial maze performance in the rat, Behav. Neurosci., 102 (1988) 294-300. 37 Stokes, K.A. and Best, P.J., Mediodorsal thalamus lesions in rats impair radial-arm maze performance in a cued environment, Psychobiology, 18 (1990) 63-67. 38 Stokes, K.A. and Best, P.J., Response biases do not underlie the radial maze deficit in rats with mediodorsal thalamus lesions, Behav. Neural Biol., 53 (1990) 334-345. 39 Tigner, J.C., The effects of dorsomedial thalamic lesions on learning, reversal, and alternation behavior in the rat, Physiol. Behav., 12 (1974) 13-17. 40 Vicedomini, J.P., Corwin, J.V. and Nonneman, A.J., Behavioral effects of lesions to the caudate nucleus or mediodorsal thalamus in neonatal, juvenile, and adult rats, Physiol. Psychol., 10 (1982) 246-250. 41 Victor, M., Adams, R.D. and Collins, G.H., The WernickeKorsakoff Syndrome, Blackwell, Oxford, 1971. 42 Weiss, B.J. and Means, L.W., A comparison of the effects of medial frontal, dorsomedial thalamic, and combination lesions on discrimination and spontaneous alternation in the rat, Physiol. Psychol., 8 (1980) 325-329.