Dissociation between components of spatial memory in rats after

Exp Brain Res (1985) 58:11-28

~ Spriqger-Verlag 1985

Dissociation between components of spatial memory in rats after recovery from the effects of retrohippocampal lesions F. Schenk ~ and R . G . M . Morris 2 t lnstitut de Physiologic, Universit6 dc Lausanne, Lausanne, Vaud, Switzerland 2 MRC Cognitive Neuroscience Group, Psychological Laboratory, University of St. Andrews, St. Andrews, Fife, Scotland

Summary. 1. A series of 4 experiments examined the performance of rats with retrohippocampal lesions on a spatial water-maze task. 2. The animals were trained to find and escape onto a hidden platform after swimming in a large pool of opaque water. The platform was invisible and could not be located using olfactory cues. Successful escape performance required the rats to develop strategies of approaching the correct location with reference solely to distal extramaze cues. 3. The lesions encompassed the entire rostro-caudal extent of the lateral and medial entorhinal cortex, and included parts of the pre- and para-subiculum, angular bundle and subiculum. Groups ECR I and 2 sustained only partial damage of the subiculum, while Group ECR+S sustained extensive damage. These groups were compared with sham-lesion and unoperated control groups. 4. In Expt 1A, a profound deficit in spatial localisation was found in groups ECR 1 and ECR+S, the rats receiving all training postoperatively. In Expt 1B, these two groups showed hyperactivity in an openfield. In Expt 2, extensive preoperative training caused a transitory saving in performance of the spatial task by group ECR 2, but comparisons with the groups of Expt 1A revealed no sustained improvement, except on one measure of performance in a post-training transfer test. All rats were then given (Expt 3) training on a cueing procedure using a visible platform. The spatial deficit disappeared but, on returning to the normal hidden platform procedure, it reappeared. Nevertheless, a final transfer test, during which the platform was removed from the apparatus, revealed a dissociation between two independent measures of performance: the rats with ECR lesions failed to search for the hidden platform but repeatedly crossed its correct location accurately during traverses of the entire Offprint requests to: R.G.M. Morris (address see above)

pool. This partial recovery of performance was not (Expt 4) associated with any ability to discriminate between two locations in the pool. 5. The apparently selective recovery of aspects of spatial memory is discussed in relation to O'Keefe and Nadel's (1978) spatial mapping theory of hippocampal function. We propose a modification of the theory in terms of a dissociation between procedural and declarative subcomponents of spatial memory. The declarative component is a flexible access system in which information is stored in a form independent of action. It is permanently lost after the lesion. The procedural component is "unmasked" by the retrohippocampal lesion giving rise to the partial recovery of spatial localisation performance.

Key words: Spatial memory - Water maze - Hippocampus - Entorhinal cortex - Recovery of function

Introduction The purpose of this paper is to report a series of behavioural experiments examining the effects upon spatial localisation in the rat of lesions of the rctrohippocampal region, including structures of the medial and lateral entorhinal cortex, the subiculum, and pre- and para-subiculum. The reasons for conducting the study were to examine: (1) the role of the major posterior interconnections of the hippocampus in spatial memory; (2) the performance of animals which had received training prior to surgery with others trained only after receiving lesions; and (3) whether extensive postoperative training could ameliorate any aspects of the deficits observed and so cause behavioural recovery or compensation. Extensive reciprocal and topographically organised interconnections are known to exist between the

12 entorhinal cortex (EC) and the hippocampus (Hjorth-Simonsen and Jeune 1972; Steward 1976; Beckstead 1978; Swanson and Cowan 1977; Wyss 1981). Electrical stimulation of the perforant path innervating the hippocampus providcs a powerful excitatory input (Anderson 1975) while lesions of EC disrupt components of hippocampal slow-wave activity (Vanderwolf and Leung 1983). There arc therefore both anatomical and elcctrophysiological grounds to suspect an intimate functional relationship between these structures (sec Van Hoescn 1982). The existence of comparable behavioural deficits after EC and hippocampal lesions is consistent with this analysis. Bilateral E C lesions have been reported to produce hyperactivity in open field tests and other impairments in locomotor and exploratory activity (Kohler and Sundberg 1977; Lasher and Steward 1981; Schenk 1983), just as occur after hippocampal lesions (O'Keefe and Nadel 1978; Gray 1982). Persistent deficits in spontaneous and rewarded alternation, and maze learning, have also been observed (Loesche and Steward 1977; Steward et al. 1977) but there are indications that recovery of performance may be possible over a prescribed time course after unilateral lesions. Two recent studies indicate that bilateral damage to the entorhinal cortex and/or blockade of connections through the hippocampus from the neocortex to the midbrain (e.g. by crossed unilateral perforant path and fimbria-fornix lesions in conjunction with section of the hippocampal commissures) cause severe and lasting impairments in Olton's 8-arm radial maze (Olton et al. 1978; Olton 1983). However, in these and some earlier studies, damage to the entorhinal cortex also extended into the pre- and para-subiculum, and the subiculum proper. In view of the anatomical connections known to exist from the pyramidal cells of area C A l to the subiculum and thence to the diencephalon and mammillary bodies (Swanson and Cowan 1977; Meibach and Siegel 1975), it is not clear whether these 8-arm maze impairments after " E C " lesions are due, in part, to damage to the subiculum. The present studies included groups of animals with both extensive and more restricted retrohippocampal lesions. Several recent studies have shown that both the severity and postoperative duration of the deficit on radial maze tasks after C A I (Jarrard 1978) and CA3 (Handelman and Olton 1981; Jarrard 1983) lesions of the hippocampus is reduced by preoperative training. The present study also included an examination of this issue with respect to EC lesions. The apparatus used in the main experiments was a water-maze describcd elsewhere (Morris 1981, 1984). Rats were trained to escapc from opaque

water onto a small hidden platform that could only be located relative to distal cues around the laboratory room. Normal rats quickly learn to swim directly to the platform from any of several starting positions, while rats with hippocampal (Morris et al. 1982; Sutherland et al. 1983) and frontal cortex lesions (Kolb et al. 1983) are impaired. As the path to the platform is in no way constrained, different learned strategies are possible, each of which may result in successively shorter escape latencies. Measures of path-length, together with measures of performance in post-training transfer tests, provide a behavioural profile from which to draw inferences about the spatial strategies adopted by normal rats and those which have sustained brain lesions. More specifically, the impairment after hippocampal lesions (Morris ct al. 1982) appears to be total providing strong support for O'Keefe and Nadel's "spatial-mapping" theory of hippocampal function. In the prcsent study, the basic procedure of escape training to a hidden platform was supplemented by others (described below) to examinc the nature of the lesion-induccd deficit and the extent of behavioural recovery.

Methods

Subjects ]'he subjects were 64 male hooded rats of the Lister strain bred in the colony at the University of St. Andrews. They were housed 2 rats to a cage with free access to food and water throughout the experiment. Their weights ranged from 300 to 400 g.

Apparatus Spatial learning. The apparatus used for the main series of experiments was an open-field water maze described elsewhere (Morris 1981). It consisted of a large circular pool (1.32 m diameter) of water maintained at 25 + 1~ C, to which 2.1 I of milk had been added. On training trials, thc rats were placed into the water facing the side walls. They could escape by climbing onto an invisible platform (8.3 cm diameter) whose top surface was hiddcn 1 cm below the opaque water surface. Two modifications of the basic task were also used: Thc simplest was a cueing procedure in which thc rats had to cscape onto a black platform protruding 1 cm above the watcr surface. Thc target platform was thus visible from the start area. In a third version of the task, the rats had to discriminate between two visually idcntical targets, namely thc black platform used in thc cucing procedure and a thin disc (8.3 cm diametcr• 1.2 cm) of black foam plastic which floated on the water surface. It was "moored" at any particular place by means of fishing line attached to a movable wcight on the bottom of thc pool. When the rats climbed onto it, it sank bcneath them. In an effort to ease the task of distal localisation, numcrous cues were arranged around the room including a black curtain, a filing cabinet, part of a large monkey cage etc. There was also an entrance door, window and skylight. All experiments were run during the long daylight of a Scottish summer (1982) between the hours of 9:00 a.m. and 8:00 p.m.

13 A vidcocamera was placed directly above the centre of the pool. The associated recorder, monitor and other obscrvational equipmcnt were located behind the black curtain in one corner of the room, where one of the two experimenters (FS or RM) sat controlling the equipment. The second expcrimenter (RM or FS) obtained each rat in turn from his cage in an adjoining room, placed the rat in the pool and then walked quietly to the observation area.

Activity measurements. The apparatus used to monitor locomotor activity was a large cage measuring 125 x 60 x 30 cm. The wooden floor was divided into 18 squares by black painted lines. The walls were of wire netting. A wooden partition divided the cage into two equal parts (63 • 60 cm) and allowed passage through its centre via 20 cm opening. The enclosure was illuminated by dircct dim light. Each rat was placed into this observation area and his behaviour observed by two experimenters (FS or J. Hagan) sitting in the same room. Observations of activity were made between 10 a.m, and 6 p.m.

posterior to brcgma and 3.5 mm lateral. No current was passed through the electrode. After the lesions had been made, the scalp was sutured and the animals kept warm until the effects of anaesthesia had fully worn off. They were then returned to their cages in the colony room.

ltLs'tology. At the end of the series of experiments, the animals were given an overdose of pentobarbital, perfused with 10% formalin and the brains extracted. Hale of the brains were processed for cresyl violet and solochrome cyanine staining, the remainder stained with crcsyl violet and according to the FinkHeimer procedure for degenerating fibres (Hjorth-Simonsen and Jeune 1972). Horizontal sections of either 30 or 40 microns werc taken, with eve R" 4th or 5th section retained. The site and extent of lesions was assessed by the two experimenters together and semi-quantitative ratings made of damage to various structures. Camera lucida drawings were made of the smallest and largest extent of damage to the brains of animals of each group at various horizontal levels. Photomicrographs were prepared of selected animals.

General procedures Assignment of animals to groups. 24 rats were given lesions intended to include the whole entorhinal cortex and the pre- and para-subicular parts of the retrohippocampal area. Of these, 12 rats (Group E C R 1) were given all training postoperativcly, while the others (Group E C R 2) received preoperative training also. A further 12 rats (Group E C R + S ) were given larger lesions intended to encompass the subiculum as well as the whole cntorhinal cortex and the pre- and para-subiculum. All rats in this group were trained postoperatively. Some 16 rats received "sham" lesions in which damage to the brain was restricted to the overlying parts of visual cortex necessarily damaged in obtaining access to the entorhinal cortex. Of these, 12 received all training postoperatively (Group Stl) while the other 4 rats received preopcrative training. Finally, 12 further rats served as unoperated controls; of these, 8 rats were given all training postoperatively, while 4 received preoperative training also. Past data allowed us to plan combined control groups of rats with sham lesions and unopcrated controls. Thus, there were 2 control groups, namely S H + C 1 without preoperative training (N = 20) and S H + C 2, with preoperative training (N = 8). In Experiment 1A below, the 12 sham and 8 unopcrated control rats of Group S H + C 1 are treated separately to demonstrate their comparable levels of performance. Following the histological analysis, 9 rats wcrc discarded from the experiment. The final numbers of animals per group is given later.

Surgery. The animals were administered a tribromoethanol anaesthesia (AVERTIN) at a dose of 10 m~;kg dissolved in saline (1 rag/ ml). They were then placed in a Kopf stercolaxic frame with the incisor bar 2.4 mm below the interaural line such that the surface of the skull was horizontal. Burr holes (2 mm diameter) were made at approximately 8 mm posterior to bregma and 3.5 mm lateral to thc mid-line. The lesions were made using radiofrequency current and for rats given bilateral E C R lesions, the lesioning probe was lowercd along tracks angled 10~ and 15~ to the sagittal plane, at 8 mm posterior to bregma and 3.5 mm lateral to the midline, to depths of 5.9, 6.9 and 7.9 mm (10~ and 6.9 and 7.9 mm (15 ~ measured along the track below a point horizontal to the midline of the skull. At each site, the current was turned on and a tip temperature of 70 ~ C maintained for 30 s. For rats given intended E C R + S lesions, the probe was lowered at the 10~ angle at a position 7.5 mm posterior to bregma. The 15~ penetration aimed at lateral entorhinal cortex continued to be at 8 mm posterior. The "sham" lesion rats had the dura cut and the probe lowered approximately 2 mm into the superficial layers of cortex at 8 mm

Histological results An analysis of the extent of the lesions was conducted at four dorsoventral levels (I, II, III and IV) corresponding to thc scctions +2000 ,tt, +1000 It, -240 Ix and -950 Is of the Simson ct al. (1981) horizontal atlas. Figure 1 illustrates that the entorhinal cortex and the angular bundle were virtually completely destroyed in all subjects. The border of thc lateral entorhinal cortex was occasionally spared at the most ventral levels. In Group E C R + S , EC damage was associated the systematic destruction of the pre- and para-subiculum. The subiculum was completely destroyed at levels II and IIl in all but one rat, which was discarded from the data analysis. Three further rats were discarded because of some unilateral damage to the dentate. Four of the 8 accepted rats showed unilateral damage of the granule cells at the extreme posterior tip of the dentate gyrus. In Group E C R 1, the pre- and para-subiculum were not systematically destroyed. Subicular damage was more extensive than intended (see Fig. I); however, between group comparisons of the subicular damage at the four analysed levels (based on a 4 point scale: 0 = no damage, 1 = probably spared, 2 = partial damage, 3 = probably completely destroyed, and 4 = complete bilateral destruction) showed a significant difference between Groups ECR 1 (Score = 2.l) and E C R + S (Score = 3.0) (MannWhitney test, z = 2.73, p < 0.01). One rat was discarded from Group E C R 1 because of bilateral damage to the dentate gyrus and 2 of tile 11 rats accepted showed some unilateral damage to the tip of the dentate granule cells. Group E C R 2 also showed some sparing of the pre- and parasubiculum. Subicular damage was also less extensive in this group than in the E C R + S group, as indicated by a Mann-Whitney test comparing the scores as calculated above (z = 2.52, p < 0.05; Score = 2.3 for Group E C R 2). Four rats were discarded because of incomplete or asymmetric lesions, or bilateral damagc invading the dentate. None of the 8 accepted rats had any dentate damage. Cortical area 18a was examined at two levels corresponding to +4000 p. and +3290 g of the Sidman et al. atlas (1982). A n estimate of the extent of damage using a similar scoring system as above showed that sham lesions caused as much as damage at the uppermost level as that seen in experimental animals. However, only two sham operated rats showed any damage at the +3290 la level while all lesion rats did. Thus, the final assignment of animals to groups was as follows: ECR 1 (N = I1); E C R + S (N = 8); SH (N = 12); C (N = 8); E C R 2 (N = 8); and S H + C 2 (N = 8).

14

Fig.

1

I

I

~I

IV

15 120 4

. \'x . . .

100

.

" \ .~, . ' . . .

~'~x ,".

'~i " ~!~ ~,~,,'S~

\

~..

.

,': .

"~,.

.-1

I', t,0

1

t

.

~, ~'.

...

..

N:~; N=8

..... e- ....

SH

N='~2

~

C

N=8

9........

Hippocampat 11982)

...

~

.,,. ..,; ,~, ,i ~i

I~ "x]

ECR I ECR*S

.: i.. " ~

,

:..

. ".

9, :

= .a,.--

.

.-

"~. '..

,~

E

.

:

~\

60

.

,

I.

80

.

~:' .'''.

.,' ~ j

"

:". . ~ .' " .'

YI

''

'.

"

.,

i,

Fig. 2. Experiment IA. Latencics to escape (s) from the water onto the hidden platform over trials 1-28 of training. The trial n u m b e r for the first of cach set of daily p o s t o p e r a tive trials is s h o w n on the x-axis. In this and subsequent figures, barmarkers represent 21 S.E.

.

20

,,,',,, 9 '

1

5

".4 .

-,10.

9

.ID-..tD"

/

17

.

25

TTI

TriO.s

Experiment IA. Disruption of place-navigation by retrohippocampal lesions The first experiment had two aims: (1) to establish whether retrohippocampal lesions would cause a deficit in the acquisition of the place-navigation task; and (2) to comparc the effects of the relatively restricted entorhinal lesion with that of the larger lesion encompassing the subiculum.

Procedure. One week after surgery, 4 groups of rats (ECR 1, E C R + S , SH and C) were given a habituation trial, 28 trials of place-navigation training and then a single transfer test in the pool. The experiment was run in 2 replications. The habituation trial conducted on Day 1 consisted of placing the rats into the centre of the pool and then allowing them to swim for a fixed period of 60 s. There was no escape platform in the pool for this trial whose primary purpose was to familiarise the animals with the apparatus. Training trials" began on Day 2 and continued until Day 6, with 4, 4, 8, 8, and 4 trials on each successive day. On a given trial, the rat was placed into the water facing the sidewalls

at one of 4 possible starting positions (N, S, E and W) and allowed to swim until it found the hidden escape platform placed at the centre of either the SW, NW, NE or SE quadrants. The platform location was counterbalanced across groups with 3 rats per group (2 rats for Group C) trained to find the platform at each position. This platform location remained fixed throughout training (Expts 1-4) for each rat. When it found the platform, the rat would climb onto it out of the water; it was left there for 30 s and then removed by hand prior to the start of the next trial which began immediately. The next trial was generally run from a different starting location (quasi-random sequence). If a rat failed to find the platform in 120 s, it was guided there by hand and given an arbitrary score of 120 s; this happened occasionally on Days 1 and 2 of training but rarely thereafter. We recorded the escape latency on each trial, using a stopwatch, and videotaped trials 25-28. The paths taken to escape from the water were transcribed onto charts. Path length was measured using a map-reading device. A transfer test was run on Day 6, immediately after trial 28: The platform was first removed from the pool. Then the rat was placed into the water at a

Fig. 1. A Photomicrographs of cresyl-violet stained sections at reference levels 2, 3 anti 4 in an u n o p e r a t e d control rat. B Schematic drawings of sections at all four reference levels: a', medial entorhinal cortex; a", lateral entorhinal cortex; b, parasubiculum; c, presubiculum; d, subieulum. T h e minimal ( \ \ XX\) and maximal (\'+a\\'~\) extent of the lesion at each level is indicated by g r o u p s

16

Paths SH 1230

-

Trial

28

ECR I

3.z~• 05 metres

ECR+S

6.6 • 2.1

SH

1.2• 0 I

C

I. 0 + _ 0I

ECR+S 1225

between-subjects factor and Trials the within-subjects factor. Significant effects of Group (F = 22.43, df 3/35, p < 0.0001), Trials (F = 15.61, df 27/945, p < 0.0(XIl) and Groups x Trials (F = 1.70, df 81/945, p < 0.0001) were obtained. The data are plotted in Fig. 2, together with the performance of a group of 10 rats given total hippocampal lesions as documented in Morris et al. (1982). Note (1) that all groups improve over the course of training; (2) that the escape latency impairment of the lesion groups is present throughout; (3) that Groups SH and C reach their minimal escape latencies within 8 trials of training, further trials serving only to stabilise performance; and (4) that Groups ECR 1 and ECR+S show faster escape performance than the hippocampal lesion group from a previous study. However, as this last comparison is across separate experiments, it should be treated with caution.

Trials 25-28

Fig. 3. Experiment IA. Typical paths taken by a SH and E C R + S rat and summary of mean path-lengths (m) for each of the four groups

place approximately opposite the correct training quadrant. He was left to swim for 60 s, as in the habituation trial. For this test, we measured both the times spent in the 4 quadrants and the number of annulus crossings - the annuli being marked on the videomonitor to indicate the exact position and surface area of platforms hidden at the centre of each quadrant.

Results. All rats swam effectively using the normal adult swimming posture (Schapiro et al. 1970), staying on the water surface for most of the time, but occasionally diving beneath the surface for brief periods.

Trials 1-28 During the training phase, the rats with lesions were consistently slower to escape and generally took more circuitous routes from the starting position to the platform than those taken by Group SH and C rats. However, all rats showed an improvement in escape behaviour over the course of training. An unequal N analysis of variance of the latencies to escape was conducted in which Groups was the

Over trials 25-28, the groups differed significantly with respect to both latency (p < 0.0001) and pathlength (p < 0.0005). Orthogonal comparisons subsequent to these overall analyses showed, with respect to both measures, that the lesion rats were inferior in performance to the sham and control animals (ps < 0.0005); that Groups SH and C did not differ (Fs < 1); and that Group ECR 1 was faster to escape than Group ECR+S (means 13.2 and 25.3 s; p < 0.005) and took a shorter path length (means 3.4 and 6.6 m; p < 0.025). Dividing the mean path-lengths by escape-latency gave a range of swimming speeds as follows: C = 21.6, SH = 20.2, ECR 1 = 25.7 and ECR+S = 26.1 cm/s. The trend is thercfore, if anything, towards faster swimming by the lesion rats. Figure 3 shows the actual paths taken by 2 typical animals.

Transfer test 1 In the transfer test, both C and SH rats showed the localised "searching" pattern of behaviour described earlier (Morris 1981), while thcre was little indication of any spatial bias to the correct platform quadrant by the lesion rats (Fig. 4). Interestingly, the total number of annuli crossed by the lesion groups was no less than that of the controls; their deficit lay in the way they distributed their movements around the pool. An unequal N analysis of variance of the times spent in the 4 different quadrants of the pool revealed

17

A. Quadrants ECR I

ECR+S

SH

C

\o. ,A,,,~/ ~ ~ d__l~ 1 o . i~ \.___J

o-

B. Annuti /

,..,

.s

6"

0

/

Fig. 4. Experiment IA. Mean times (s) spent in each of the four quadrants of the pool during transfer test 1. For a rat trained to NE, the NW quadrant was adjacent-left (Adj-l), SE was adjacent-right (Adj-r) and SW was opposite (Opp) as shown in insert diagram on the left. Note that sham and control rats distributed their time so as to spend > 30 s (i.e. > 50%) in the training quadrant, while the two lesion groups show little or no spatial bias. B Mean crossingsof the annuli marking the exact surface area of the training quadrant platform and the positions where a platforrn would have been placed in each of the other three quadrants. Again note spatial accuracyof the sham and control groups

/)t\

a highly significant effect of quadrants (F = 78.59, df 2/105, p < 0.0001; numerator degrees of freedom reduced by 1 as the scores for individual animals necessarily add to 60 s); and a significant interaction between Quadrants and Groups (F = 18.40, df 6/105, p < 0.0001; numerator adjustcd) implying (Fig. 4A) that Groups SH and C spent greater time in the correct quadrant than Groups E C R and E C R + S (confirmed in subsequent tests). Analysis of the annulus crossings measure of accuracy showed no overall difference between Groups (F = 1.51, df 3/35, p < 0.25). This implies that the lesion groups did not merely swim helplessly beside the side walls of the pool (Mean total crossings: Group E C R = 7,6; E C R + S = 5.8; SH = 7.8; and C = 8.6). Thc analysis of variance also revealed highly significant effects of annulus position (F = 44.96, df 3/105, p < 0.0001 ; note numerator degrees of freedom does not have to be adjusted for this measure as the four annulus scores are independent), and the critical Groups x Annulus position interaction (F = 10.70, df 9/105, p < 0.0001) which represents the lesion impairment. In relative terms, these findings indicate that the controls spent over 50% of their time searching in the correct training quadrant, crossing the annulus there circa 10 times more often than the other annulus positions; while the lesion groups spent no more than 30% of the transfer test time in the correct quadrant (chance = 25%), crossing the annulus there no more than 1.5 times as often as the other annuli.

Experiment 1B. Disruption of open-field activity by retrohippocampal lesions Procedure. The activity test was conducted 6 days after surgery. The rats were taken individually and placed on the square facing the centre opening of the

'Fable 1

Group

Linecrossings (~289min

27.,-5 min

Rcaring 0-2 89rain

289 min

ECR 1 ECR+S Sham Control

72.4 71.6 39.8 37.4

76.6 90.7 26.1 19.5

2.9 3.0 9.4 9.2

5.1 5.0 9.9 8.0

activity cage. Their behaviour was then observed for 5 min, dividing the observation time into 2 periods of 150 s. The two observers recorded (l) line crossings (both forepaws), and (2) rearing frequency (both forepaws raised, trunk of body vertical).

Results. The rats with lesions were very active throughout the 5 min observation period whereas sham and control rats were not only less active but also, towards the end of the observation period, generally quiet in one corner of the cage (Table 1). An unequal N analysis of variance of the linecrossings revealed significant Groups (F = 21.14, df 3/35, p < 0.0001) and Groups x Time-Periods (F = 16.90, df 3/35, p < 0.0001) effects. A comparable analysis of the rearing data revealed a different pattern with Groups SH and C showing far more rearing than the lesion groups E C R and E C R + S (F = 9.67, df 3/35, p < 0.0001) and no significant change across the two time-periods (F = 1.08, df 3/35, p > 0.25). There was no difference between the two lesion groups in either line-crossings or rearing frequency. Discussion. The main findings of Experiment I A are that rats with rctrohippocampal lesions show a severe postoperative impairment in the initial acquisition of the place-navigation task; this impairment takes the

18 form of longer, more circuitous and inaccurate search paths during training with little spatial bias or discriminative accuracy in the transfer test. Thus, deafferentation from entorhinal input, like large (Morris et al. 1982) and selective (Sutherland et al. 1983) hippocampal lesions, impairs spatial learning. Experiment 1B extends other findings (e.g. Kohler and Sundberg 1977) in demonstrating that hyperactivity in an open field can bc coupled with a pronounced decline in exploratory rearing. However, the impairment was not identical in the two groups with lesions. The rats of Group ECR 1, with relatively restricted entorhinal cortex lesions, were consistently faster to escape than E C R + S rats, and rats with hippocampal lesions (from the earlier Morris et al. 1982) study. Their terminal path lengths were also shorter. As the entorhinal cortex damage in both lesion groups was near complete, these results raise the possibility that the deafferented hippocampus has some spatial learning capacity intact, that depends upon output relays via the subiculum. The status of this residual capacity is the subject of the remaining experiments. Experiment 2 examines the effects of preoperative training while Experiment 3 looks at more extensive postoperative training.

Experiment 2. Can preoperative training reduce the navigation deficit? At a simple level, the purpose of Experiment 2 was to dissociate storage and retrieval. If the integrity of the retrohippocampal region is essential only for storage, preoperative training should ameliorate the lesioninduced deficit. Two additional groups were therefore trained on the hidden platform task prior to being given either E C R lesions, sham surgery or no treatment. These rats were subsequently given postoperative retraining identical to that of Experiment 1A.

Procedure. Two groups of rats were givcn a habituation trial followed by 32 trials of preoperative training on the hidden platform task, with 4, 4, 8, 8, and 8 trials daily. Two days after this training, all rats receivcd their appropriate surgery (if any) according to the procedures outlined for Experiment 1 above. Group ECR 2 (N = 8) received restricted entorhinal cortex lesions; Group S H + C 2 was made up from 4 rats receiving sham surgery and 4 unoperated control rats. After a 7 day recovery period, these groups were retrained on the hidden platform task for 28 trials with the platform location (SW, NW, NE or SE - counterbalanced) unchanged. This retraining was

ECR 2

SH+E 2

Quodronfs 30" t,q aJ

E

I----

20" 10. 0.

Annuti k" r-

3

"

2

9

2 09 0.10), implying that the deficit was sustained throughout postoperative training. Thus, a deficit in escape latency associated with retrohippocampal lesions was not prevented by preoperative training. However, comparison of pretrained rats with groups receiving postoperative training only (Fig. 6) shows that all pretrained rats were much faster to escape over the initial trials of postoperative training (e.g. trials 1-4). We therefore conducted an overall 2 x 2 analysis of variance to compare all four groups and to test whether the effects of preoperative training were transient. This analysis revealed the following effects: Rats with lesions escaped more slowly: F = 64.06, df 1.43, p < 0.0001; rats receiving preoperative training escaped faster: F = 27.67, df 1/43, p < 0.000l; there was a significant interaction (p < 0.05) between Lesion and Training condition; and finally, of importance, a highly significant triple interaction between Lesion, Training-condition and Trials (F = 3.94, df 27/1161, p < 0.0001). Further separate analyses of Trials 1-4 and 25-28 provide evidence that this triple interaction was due to the transient effects of preoperative training. Over trials 1-4, effects of both lesion (p < 0.005) and trainingcondition (p < 0.0001) were obtained; whereas, over trials 25-28, the lesion groups differed from the controls (p < 0.0001) while training condition was not a significant influence (F < 1).

A. Ouctdroni-s 4030-

~2o.

EERI

ECR 2

SH*[I

SH.E 2

[~

vZlO. O.

B. Annuti 8.

~ ~.

~+~

O-

Fig. 7. Experiment 2. A Distribution of searching times (s) in the four quadrants of the water-maze during transfer test 1. Note only weak spatial bias in lesion groups. B Annulus crossing scores. Note Groups ECR 1 and 2 show different distributions, with Group ECR 2 showing almost as many training annulus crossings as the two SH+C control groups ,

Transfer test 1. Transfer test 1 was run on Day 6 immediately after the last trial of postoperative training. The results (Fig. 7) show that preoperative training had little or no effect on the distribution of times between the four quadrants of the pool. Groups ECR 1 and ECR 2 spent 18.5 and 19.4 s in the training quadrant, respectively, while the times for Groups S H + C 1 and S H + C 2 were 32.7 and 34.2 s. An analysis of variance showed a highly significant Lesion • Quadrants interaction

20 (F=38.21, df 3/129, p < 0.0001) but no interaction between Pre- vs. Postop training and Quadrants (F 0.5). The primary difference between the lesion and control groups lay in the lack of occasions for which animals spent more than 5 s in the training quadrant, a point which is emphasised by the shading in Fig. 11. There was apparently no difference between the "recovered" (ECR I and ECR 2) and "non-recovered" ( E C R + S ) lesion groups.

Discussion. The main findings of Experiment 3 were (1) that the cueing procedure promoted a partial recovery of spatial localisation in the two ECR lesion groups, but (2) this recovery had its limits, particularly with respect to (a) the long and indirect paths taken in approaching the platform during training trials, and (b) the animals' reactions to the failure to find the platform during the transfer test. The question raised by these data is: Do the "recovered" rats actually have any "knowledge" of the platform's location'? Or, have they merely learned, during the extensive training trials with both the hidden and the visible platform, to execute a path that results in rapid escape? Experiment 4 provides one lest of these different interpretations.

24

( ~0

ECR I

ECR+S

SH+CI

~- 30 .K

20oJ

~ 10ea

o

01

Fig. 11. Experiment 3. Standardiscd frequency histograms showing distribution of times spent in the training quadrant during transfer test 2. Insert diagram shows how interval-times were derived from tl to t2, t3 to t4 and so on. Successive intervals were collated to gencrate the histogram, which shows percent of occasions spent searching for 0-1 s, 1-2 s etc. Shading from 5 s was chosen arbitrarily to emphasise the greater proportion of longer intervals in Groups SH+C

5678910+

23

secs

ECR2

100 -

SH+C 2

ii ~7

" k n o w l e d g e " of spatial location, they should be able to learn to swim to the rigid p l a t f o r m avoiding the floating one. H o w e v e r , if the animals had only learned the strategy of swimming in such a way as to come close to the platform, they would - in the present task - sometimes swim near to the floating platform also, see it, and then try to climb on. T h e rats would therefore fail to learn the task.

75~J o u

o~

.

.

.

.

50-

25-

9

ECR 1

9 o

ECR+S SH

6

7

O1

2

3

I+

5

8

Brooks

Fig. 12. Experiment 4. Mean percent correct choice in 2 platform simultaneous discrimination task over 8 blocks of 5 trials. Numbers at right hand end of graph lines refer to numbers of animals in each group reaching criterion of 10 successive corrcct responses

Experiment 4. Failure to show effective spatial dis'crimination T h e p u r p o s e of E x p e r i m e n t 4 was to examine simultaneous spatial discrimination. T w o identical visible platforms were used, and the rats task was to learn to a p p r o a c h the rigid platform and avoid the floating one. We r e a s o n e d that if the animals had some

Procedure. Rats (N = 6) which had shown g o o d recovery from G r o u p E C R 1 ( A n n u l u s crossing score = 4.8), rats (N = 4) showing no recovery f r o m G r o u p E C R + S (Crossing score = 2.5) and 8 sham-lesion rats (from G r o u p S H + C 1, Crossing score = 4.8) were given 5 trials per day over 8 days. O n each trial, the rat was placed into the pool at the side walls (at N, S, E or W in q u a s i - r a n d o m order) and allowed to escape onto either a rigid platform placed in that animal's correct platform location, or o n t o a floating platform whose position varied f r o m trial to trial (to SW, N W or N E for rats trained to SE etc.). A correction p r o c e d u r e was used so that after sinking back into the water u p o n first contact with the floating platform, the rats were allowed to swim to the correct one. The intertrial interval (30 s) was spent on the rigid platform, and we r e c o r d e d the n u m b e r o f correct responses from each animal per day.

25

Results. The results presented in Fig. i 2 show that the sham lesion rats transferred immediately to the new task with 7/8 of the animals reaching a criterion of 10 successive correct responses. Neither group of lesion animals learned to avoid the floating platform although both groups were consistently above chance. There was no overlap between any sham and any lesion animal in their overall discrimination performance. Further statistical analysis was deemed unnecessary. Discussion. These findings indicate that both "recovered" and "nonrecovered" lesion rats fail to learn a simple two-choice spatial discrimination task. Animals in both groups failed to inhibit attempts to climb onto the floating platform. The clear implication is that the spatial bias of the annulus crossing scores is not a good predictor of an animal's ability to dis'criminate one spatial location from another. General discussion

This series of experiments has shown (1) a profound impairment in spatial localisation after retrohippocampal lesions, (2) a partial and selective recovery of spatial localisation during postoperative training, (3) that preoperative training accelerates this partial bchavioural recovery while in no way ameliorating the residual deficit in localisation; (4) that the residual deficit includes a persistent impairment in spatial discrimination performance; and (5) that larger lesions encompassing most of the subiculum, in addition to other retrohippocampal structures, may limit thc extent of recovery. Finally (6) retrohippocampal lesions caused hyperactivity in an opcnfield activity test. In this discussion, we outline a hypothesis to account for thesc findings. Thc essential features of this hypothesis are (1) that retrohippocampal lesions cause a "fractionation" of different subsystems of spatial localisation; and (2) that the partial behavioural recovcry is caused by "'unmasking", rather than true functional recovery of a neural subsystem spared by the lesion. We consider, first, a possible artefact that might account for the pattern of results.

Is the apparent dissociation between sub-components of spatial memory an artefact? The major novel finding of these experiments is the apparent dissociation between the accuracy and searching measures of transfer test performance. It is possible that this dissociation is no more than an

artefact of poor learning arising out of differential sensitivity of the two measures. On this view, any procedure which produces poor learning or memory, including limited training or long retention intervals in normal rats, might produce a transfer test profile similar to that of recovered entorhinal rats. This could happen if the accuracy measure (annulus crossings) is more scnsitive to weak memory than the searching measure (quadrant times). We reject this possibility on two grounds: First, Pcarson crosscorrelations of (1) the times spcnt in the training quadrant and (2) the crossings of the training platform position by sham + control animals (N = 28) during both transfer tests were highly significant (TI'I: r = 0.78, df 27, p < 0.001; TT2: r = 0.46, p < 0.02; Range for quadrant times 18 to 46 s; for annulus crossings 1 to 10). This indicates that the two measures were behaving in a similar way across the range of spatial biases of different animals. Second, unpublished studies of retention of the spatial bias in normal rats over training-test intervals of up to 14 days have shown parallel forgetting functions for the accuracy and searching measures of transfcr test performance. Both findings argue against any differential sensitivity of the two measures in normal rats.

Characterising the dissociation Given that the dissociation between accuracy and searching tendency is genuine, how should it be characterised? Late in training, the lesion rats reached the platform quickly from any starting position, though escape latencies and pathways were longer than those of control rats. Moreover, during the 60 s transfer test period, rats of two lesion groups (ECR] and ECR2) were able to generate approximately 4-5 routes crossing the exact spot marking the former platform position. However, the fact that the animals are accurate does not necessarily imply that they "know" that a platform is hidden at place A, or that place A exists as a place to be distinguished from places B, C or D. The failure of recovered entorhinal rats (ECR 1 and 2) to stop and search near where they fail to find the platform during a transfer test, or to stay in the training quadrant for more than 3 to 4 s, argues against these animals having expectations about the platform's position, i.e. an accessible memory of its location serving as a goal of action. Similarly, the failure of these animals to transfer to a simple 2-choice spatial discrimination task points to a general deficit in access to spatial information: the rats can navigate to place A accurately but cannot choose appropriately between places A and B when forced to do so.

26

The hypothesis While it is tempting to suppose that the course of behavioural recovery is subserved by a time- or training-dependent recovery of spatial memory, a simpler interpretation is that bilateral entorhinal cortex lesions cause a fractionation of the animal's spatial localisation system. We propose that this system continues to operate, but because of the EC lesion, it processes only aspects of the information handled in normal animals. More specifically, declarative aspects of spatial memory are lost, while procedural or habitual aspects are spared (Cohen and Squire 1980; Dickinson 1980). By declarative spatial memory, we mean the representation of knowledge in a form which describes the position of the escape platform in relation to other cues in the environment. Following Dickinson's (1980) terminology, a declarative proposition in memory might be of the form: "there is a platform at A near to cues X, Y and Z". This proposition describes the environment in a manner that is logically independent of action. By procedural spatial memory, on the other hand, we mean a representation of stimulus-response habits necessary to guide the animal to the correct location. In addition to learning to swim away from the sidewalls of the pool, and to climb onto platforms, the procedural representation could be of the form "swim along a path passing near to cues X, Y and Z to escape from the water". In this case, spatial knowledge is built into the stimulus-response habits acquired during training but is otherwise inaccessible. On this hypothesis, the dissociation between accuracy and searching is due to EC lesion rats lacking the flexible data base of information that triggers novel searching behaviour in normals when the platform has been removed from its training place. Instead, they only repeat approaches that were reinforced during acquisition. However, it may seem puzzling that our results show a lasting impairment in spatial discrimination learning in the water maze, even in recovered entorhinal cortex lesion rats, as simple discrimination tasks are sometimes classified as procedural (Squire and Zola-Morgan 1983). In a simple plus-maze, the rat is faced with a choice between a left or right turn, and/or approach to one set of extramaze cues or another. That is a much simpler problem and Steward et al. (1977) report that bilateral entorhinal cortex lesions cause no impairment. In the pool, however, there are four possible starting positions and no pathway to the platform is constrained; there is no choice-point where a simple binary rule could help the rat discriminate between the correct and incorrect platforms. Thus, discrimi-

nation failure may occur each time the rat's path happens to take him close to the floating platform. The deficit in place-navigation, the lack of reaction to the absence of the training platform, and the failure of place-discrimination are each predicted by the spatial mapping theory of hippocampal function (O'Keefe and Nadel 1978). However, the residual spatial learning following preoperative experience and/or extensive training raises two questions that were not considered by O'Keefe and Nadel. First, does any type of hippocampal dysfunction (fornix section, total aspiration, retrohippocampal lesion) impair spatial localisation equally? If not, do residual aspects of hippocampal circuitry organise spatial learning? Our results suggest that a partial spatial recovery might be due to the extent of subicular sparing. Second, it is not obvious whether O'Keefe and Nadel's taxon strategies could be responsible for the observed recovery. Stereotyped swimming at a constant distance from the wall was not observed consistently and the analysis of angles of approach to the platform failed to reveal any common strategy of approach. The annulus crossing data of the recovered EC lesion rats indicate that their routes were very precisely controlled by spatial cues in the vicinity of the training position, a pattern of behaviour which is more sophisticated than simply approaching a distal cue. It is for this reason that we argue that some aspects of spatial knowledge are built into stimulus-response habits acquired during training by rats with retrohippocampal lesions. Furthermore, we suspect that our proposed fractionation of spatial memory into two components may help to resolve the current controversy between O'Keefe and Nadel's "spatial mapping" and Olton's (1983) "working-memory" theories of hippocampal function (see Morris 1983). The task used in this series of experiments is a reference memory task according to Olton's classification; and the fact that spatial deficits were observed runs contrary to his theory. However, the strictly procedural recovery that we have observed in the ECR rats may be sufficient to account for successful spatial reference memory performance in the tasks examined by Olton and his associates.

Behavioural recovery does not imply recovery of an underlying neural subsystem A further implication of our hypothesis is that behavioural recovery is not a consequence of any dynamic change in underlying neural mechanisms,

27 i.e. structural reorganisation. Behavioural recovery, defined as the ability to perform a skill at a certain time after surgery which cannot be performed at an earlier post-surgery interval, may be no more than learning (or relearning) by a subsystcm whose behavioural expression is ordinarily masked by the separate, faster but now damaged subsystem. This view has both similarities and differences to the "behavioural compensation" hypothesis due to the LcVere's (LeVere 1975; LeVere and LeVere 1982). The LeVere's argue that lesion-induced behavioural deficits are not just due to the loss of neurological systems, but also the failure of an individual to utilise spared componcnts of the injured systems. This occurs because of attempts to use other neural systems to compensate for brain damage. To the extent that compensation is successful, spared capacity is under-utilised; if unsuccessful, spared capacity is used and apparent recovery can proceed. LeVere and LcVerc (1982) present evidence from a compound-cue discrimination-lcarning paradigm in support of their hypothesis. Our proposal sharcs with the compensation hypothesis the implication that apparent recovery is actually due to spared function; i.e. that the recovery is apparent rather than real. We do not, therefore, seek to explain the recovery of a ncural subsystcm that has not, in fact, been lost. Our hypothesis differs from the LeVere's in being selective, in the sense that the lesion fractionates subsystems of a larger spatial localisation system; and also in relying on a mechanism of unmasking of a subsidiary spatial system, rather than spared capacity of a single system. Unmasking of procedural spatial memory is realised through the loss of declarative spatial memory, rather than the failure of other non-spatial compensatory behaviours. The dclay in postoperative lcarning in the EC lesion rats is probably duc to the fact that procedural learning requires morc training to develop. Moreover, our hypothesis explains why preoperative training causes an acceleration of the recovery of transfer test accuracy - these rats having had twice as many training trials as postoperatively trained rats - but no sparing of efficient search behaviour. One cornerstone of the LcVere's compensation hypothesis has been the belief that the nervous system is incapable of the plasticity necessary for functional reorganisation, a view widely held in clinical neurology. However, there are now several grounds for believing that structural reorganisation of the adult nervous system of mammals is possible (Liu and Chambers 1958; Finger and Stein 1982), and that such reorganisation can mediate behavioural recovery. Studies of the effects of unilateral cntorhi-

nal cortex lesions have shown, for example, that they cause a massive expansion of the crossed projcction of the perforant path together with a correlated timedependent recovery of spontaneous and foodrewarded alternation behaviour. Subsequent lesions of the dorsal psalterium (Locsche and Steward 1977) abolish thc recovered alternation, implicating sprouting of the crossed pathway and its associated reactive synaptogenesis in the underlying process of functional reorganisation. Howcver, Ramirez and Stein (1984) have bccn unable to replicate these findings; and the claim that sprouting of septal, commissural and associated afferents to the molecular layer of the dentate gyrus are responsible for time-dependent changes in open-field activity after bilateral EC lesions is conceded by Steward et al. (1977) to be speculative. Furthermore, Lasher and Steward (1981) have been unable to demonstrate that these time-dependent changes are related to afferent rcorganisation in another species, namely cats. However, while arguing against structural reorganisation being responsible for our results, it should be rccognised that our hypothesis and a functional reorganisation hypothesis arc not, necessarily, in conflict. Structural reorganisation could indeed permit a true recovery of a neural subsystem capable of procedural spatial learning. But the evidence presented here of faster behavioural recovery in the preoperatively trained Group ECR 2 argues against this proposal assuming that reorganisation is a strictly time-dependent process. Handelmann et al.'s (1983) report that different training regimes may induce differential neurochemical sequellae of a brain lesion (in their case damage to area CA3 of the hippocampus) complicates the general problem of distinguishing between time- and training-dependent recovery, but does not affect the interpretation of this study: Groups ECR 1 and ECR 2 received identical post-operative training. In summary, retrohippocampal lesions cause a fractionation of allocentric spatial localisation. We propose that the flexible use of declarative spatial knowledge is permanently lost; and that this unmasks a lower level stimulus-response or procedural subsystem spared by the lesion. Apparent behavioural recovery of aspects of localisation reflects learning (or relearning) by this subordinate system.

Acknowledgements.We are gratefulto Bill Dewar, Chris Barman and Jim Haganfor technicaland other assistance;and to M. Capt for assistance with thc histology.This work was supported by a grant from ETPBBRof the European ScicnccFoundationand bv grants from the UK Medical Rcscarch Council and the Fonds National Suissede la Recherche Scientifique.

28

References Andersen P (1975) Organisation of hippocampal interncurones and their interconnections. In: Isaac.son RL, Pribram KH (eds) The hippocampus, Vol f. Plenum Press, New York, 1975 Beekstead RM (1978) Afferent connections of the entorhinal area in the rat as demonstrated by retrograde cell labelling with horseradish peroxidase. Brain Res 152:249-264 Cohen NJ (1984) Preserved learning capacity in amnesia: evidence for multiple memory systems. In: Butters N, Squire LR (eds) The neuropsychology of memory. Guilford Press, New York Cohen NJ, Squire LR (1980) Preserved learning capacity and retention of pattern-analysing skill in amnesia: dissociation of knowing how and knowing that. Science 210:207-210 Dickinson A (1980) Contemporary animal learning theory. Cambridge University Press, Cambridge Finger S, Stein DG (1982) Brain damage and recovery. Academic Press, New York Gaile GL, Burt JE (1980) Directional statistics. University of East Anglia Press, Norwich Gray JA (1982) The neuropsychology of anxiety. Oxford University Press, Oxford Handelmann GE, Olton DS (1981) Spatial memory following damage to the Hippocampal CA3 pyramidal ceils with kainic acid: impairment and recovery with preoperative training. Brain Rcs 217:41-58 Handelmann GE, Olton DS, O'Donohuc TL, Bcnnfield MC, Jacobowitz DM, Cummins CJ (1983) Effects of time and experience on hippocampal neurochemistry after damage to the CA3 subficld. Pharmacol Biochem Behav 18:551-561 Hjorth-Simonsen A, Jeunc B (1972) Origin and termination of the hippocampal perforant path in the rat studied by silver impregnation. J Comp Ncurol 144:215-232 Jarrard LE (1978) Selective hippocampal lesions: differential effects on performance by rats of a spatial task with preoperative vs. postoperative training. J Comp Physiol Psychol 92: 1119-1127 Jarrard LE (1983) Selective hippocampal lesions and behaviour: effects of kainic acid lesions on performance of place and cue tasks. Behav Neurosci 97:873-889 Kolb B, Sutherland RJ, Whishaw IQ (1983) A comparison of the contributions of frontal and parietal association cortex to spatial Iocalisation in rats. Behav Ncurosci 97:13-27 Kohler C, Sundberg H (1977) Locomotor activity and exloratory behaviour after dorsal entorhinal cortex lesions in the albino rat. Behav Biol 20:419-432 Lasher SS, Steward O (1981) The time-course of changes in openfield activity following bilateral entorhinal lesions in rats and cats. Behav Neurol Biol 32:1-20 LeVerc TE (1975) Neural stability, sparing and bchavioural recovery following brain damage. Psychol Rev 82:344-358 LeVere ND, LcVere TE (1982) Recovery of function after brain damage: support for the compensation theory of the behavioural deficit. Physiol Psychol 10:165-174 Liu CN, Chambers WW (1958) Intraspinal sprouting of dorsal root axons. Arch Neurol Psychiat 79:46-61 Locsche J, Steward O (1977) Behavioural correlates of denervation and reinnervation of the hippocampal formation of the rat: recovery of alternation performance following unilateral cntorhinal cortcx lesions. Brain Res Bull 2:31-39 Mcibach RC, Sicgel A (1975) The origin of fornix fibers which project m the mammillary bodies in the rat: a horseradish peroxidase study. Brain Res 88:508-512 Morris RGM (1981) Spatial localisation does not depend upon the presence of local cucs. Learn Motiv 12:239-260

Morris RGM (1983) An attempt to dissociate "spatial mapping" and "working-memory" theories of hippocampal function. In: Sicfert W (ed) The neurobiology of the hippocampus. Academic Press, London Morris RGM (1984) Developments of a water-maze procedure for studying spatial learning in the rat. J Neurosci Meth 11:47-60 Morris RGM, Garrud P, Rawlins JNP, O'Keefe J (1982) Placenavigation impaired in rats with hippocampal lesions. Nature 297:681-683 O'Keefe J, Nadel L (1978) The hippocampus as a cognitive map. Oxford University Press, Oxford Olton DS (1983) Mcmory functions of thc hippocampus. In: Siefert W (ed) The neurobiology of the hippocampus. Academic Press, London Olton DS, Walker JA, Gage FH (1978) Hippocampal connections and spatial discrimination. Brain Rcs 139:295-308 Ramirez JJ, Stein DG (1984) Sparing and recovery of spatial alternation performance after entorhinal cortex lesions in rats. Bchav Brain Res 13:53-61 Schapiro S, Salas N, Vukovich K (1970) Hormonal effects on ontogeny of swimming ability in the rat: assessment of central nervous system dcvelopment. Science 168:147-151 Schenk F (1983) Activity and exploratory behaviour after lesions of the medial entorhinal cortex in the woodmouse (Apodemus sylvaticus). Behav Neural Biol 37:89-107 Simson EL, Jones AP, Gold RM (1981) Horizontal stereotaxic atlas of the "albino rat brain. Brain Res Bull 6:297-326 Squire LR, Zola-Morgan S (1983) The neurology of memory: the case for correspondence between the findings for man and non-human primates. In: Dcutsch JA (ed) The physiological basis of memory (2nd edn). Academic Press, New York Steward O (1976) Topographic organisation of the projections from the cntorhinal area to the hippocampal formation of the rat. J Comp Neurol 167:285-314 Steward O, Loeschc .1, Horton WC (1977) Behavioural correlates of denervation and reinncrvation of the hippocampal formation of the rat: open field activity and cue utilisation following bilateral antorhinal cortex lesions. Brain Res Bull 2:41-48 Sutherland RJ, Whishaw IQ, Kolb B (1983) A bchavioural analysis of spatial Iocalisation following electrolytic, kainateor colchicine-induced damage to the hippocampal formation in the rat. Behav Brain Res 7:133-153 Suzuki S, Augerinos G, Black All (1980) Stimulus control of spatial behaviour on the eight-arm maze in rats. Learn Motiv 11:1-18 Swanson LW, Cowan WM (1977) An autoradiographic study of the organisation of the efferent connections of the hippocampal formation in the rat. J Comp Neurol 172:49-84 Van-Hoesen GW (1982) The para-hippocampal gyrus - new observations regarding its cortical connections in the monkey. Trends Ncurosci 5:345-350 Vanderwolf CII, Leung LWS (1983) Hippocampal slow activity: a brief history and the effects of cntorhinal lesions and phcncyclidine. In: Sicfert W (ed) The neurobiology of the hippocampus. Academic Press, London Weiskrantz L, Warrington EK (1975) The problem of the amnesic syndrome in man and animals. In: lsaacson RL, Pribram KH (eds) The hippocampus, Vol 2. Plenum Press, New York Winocur G, Breckenridge CB (1973) Cue-dependent behaviour of hippocampally damaged rats in a complex maze. J Comp Physiol Psychol 82:512-522 Wyss JM (1981) Autoradiographic study of the cfferent connections of entorhinal cortex in the rat. J Comp Neurol 199: 495-512

Received April 2, 1984 / Accepted October 17, 1984