- Email: [email protected]
Place and Response Learning of Rats in a Morris Water Maze: Differential Effects of Fimbria Fornix and Medial Prefrontal Cortex Lesions
Neurobiology of Learning and Memory 75, 164–178 (2001) doi:10.1006/nlme.2000.3962, available online at http://www.idealibrary.com on
Place and Response Learning of Rats in a Morris Water Maze: Differential Effects of Fimbria Fornix and Medial Prefrontal Cortex Lesions Jan P. C. de Bruin, Marta P. Moita,1 Hannie M. de Brabander, and Ruud N. J. M. A. Joosten Graduate School of Neurosciences Amsterdam, Netherlands Institute for Brain Research, Amsterdam, The Netherlands
The question examined in this study is concerned with a possible functional dissociation between the hippocampal formation and the prefrontal cortex in spatial navigation. Wistar rats with hippocampal damage (inflicted by a bilateral lesion of the fimbria fornix), rats with damage to the medial prefrontal cortex, and controloperated rats were examined for their performance in either one of two different spatial tasks in a Morris water maze, a place learning task (requiring a locale system), or a response learning task (requiring a taxon system). Performance of the classical place learning (allocentric) task was found to be impaired in rats with lesions of the fimbria fornix, but not in rats with damage of the medial prefrontal cortex, while the opposite effect was found in the response learning (egocentric) task. These findings are indicative of a double functional dissociation of these two brain regions with respect to the two different forms of spatial navigation. When the place learning task was modified by relocating the platform, the impairment in animals with fimbria fornix lesions was even more pronounced than before, while the performance of animals with medial prefrontal cortex lesions was similar to that of their controls. When the task was again modified by changing the hidden platform for a clearly visible one (visual cue task), the animals with fimbria fornix lesions had, at least initially, shorter latencies than their controls. By contrast, in the animals with medial prefrontal cortex damage this change led to a slight increase in escape latency. 䉷 2001 Academic Press Key Words: prefrontal cortex; fimbria fornix; place learning; response learning; Morris water maze; rat.
This research was supported by a Leonardo da Vinci training placement grant from FORBITEC (Associac¸a˜o para a formac¸a˜o te´cnica em biotecnologia) to Marta P. Moita. The authors thank Henk Stoffels for preparing the lesion reconstruction figures. Address correspondence and reprint requests to Jan P. C. de Bruin, The Netherlands Institute for Brain Research, Meibergdreef 33, 1105 AZ Amsterdam, The Netherlands. Fax: ⫹31 20 6961006. E-mail: [email protected]. 1 Current address: Instituto Gulbenkian Cieˆencia, 2781 Oeiras Codex, Portugal. 1074-7427/01 $35.00 Copyright 䉷 2001 by Academic Press All rights of reproduction in any form reserved.
164
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
165
INTRODUCTION Animals use two different orientation systems to locate a target in space (e.g., an invisible platform in a water maze), a place (or locale) system and a taxon system (O’Keefe and Nadel, 1978). A place system implies the development of a spatial cognitive map, an internal representation of the relations among distal (extramaze) cues. Since the spatial relations between places are independent of the animal’s location, it has also been termed an allocentric cognitive map (e.g., Kesner, Farnsworth, & DiMattia, 1989). The learning process involved is called place learning or allocentric learning. A taxon system refers to animals following a route and, in the absence of distal cues, orientation may take place by a repeated use of relatively fixed motor movements to locate the target (e.g., turn left, turn right, move at a particular angle, etc.). Dimattia and Kesner (1988) have termed this a praxic rule, while others refer to it as (position) response learning, or egocentric (spatial) learning (Kesner et al., 1989; Packard & McGaugh, 1996). Place learning and response learning are the subject of the present study. The concept of a spatial cognitive map together with its cerebral localization has drawn the attention of many researchers to place learning, while the neural correlates of response learning have been the subject of fewer studies. Searching for the neural mechanisms underlying place learning, several lines of evidence led O’Keefe and Nadel (1978) to propose the hippocampus as the main neural structure involved. Strong support came from electrophysiological recordings of hippocampal cells (place cells) which are activated when the animal is in a particular location in an open field (O’Keefe & Nadel, 1978). Additionally, data from hippocampal lesions, both in humans and in various other mammalian species (Morris, Garrud, Rawlins, & O’Keefe, 1982; O’Keefe & Nadel, 1978), have indicated that the hippocampus is involved in the processing of spatial information. This model has received wide support, and research on spatial navigation has focused on the hippocampus as the main neural structure involved in spatial learning and memory. More recently, evidence has accumulated pointing to a multiple memory system where different cortical regions (such as the posterior parietal cortex, frontal cortex) would also be involved in the processing of spatial information (Packard & McGaugh, 1992; Winocur, 1991). Posterior parietal cortical lesions have been shown to induce severe impairments in the classical place learning version of both the Morris water maze task (DiMattia & Kesner, 1988; Kolb, Buhrmann, McDonald, & Sutherland, 1994) and in the cheese board task, a “dry” version of the former (Kesner et al., 1989). Primates with parietal cortical damage are impaired in the learning of mazes and in making visual–spatial associations (Petrides & Iversen, 1979). Frontal brain damage has also been reported to disrupt the performance of tasks with a spatial component. In monkeys an impairment in right/left discriminations was observed (Pohl, 1973). In rats frontal cortical damage was shown to lead to deficits in spatial delayed alternation in a T-maze (De Brabander, de Bruin, & Van Eden, 1991) and to impaired performance in the (egocentric) adjacent arm task in a radial maze (Kesner et al., 1989; Kolb et al., 1994). With this new line of thought (a multiple memory system with different cortical regions involved) Poucet (1993) has proposed a model for spatial orientation which emphasises place navigation. The hippocampus and the posterior parietal cortex would function in a cooperative manner to form an allocentric cognitive map. According to this model the
166
DE BRUIN ET AL.
hippocampus would be essential during the learning period, but not for the retention of spatial information, i.e., after hippocampal damage an animal would still be able to orient itself in a well-know environment. Thus, this area would merely be responsible for the establishment of topological relations between places. The posterior parietal cortex would be involved in both acquisition and retention, playing a crucial role in processing the geometric structure of the environment. Data from a study on the effects of hippocampal and posterior parietal cortical damage in the Morris water maze task strongly support Poucet’s model (DiMatttia & Kesner, 1988). The frontal cortex (less prominent in this model) would be involved in solving problems with a spatial component. There has been some contradicting evidence concerning the involvement of the frontal cortical region in spatial learning. Some studies, using the Morris water maze task, have shown a disruption of place learning in rats with frontal lesions (Kolb, Pittman, Sutherland, & Whishaw, 1982; Kolb, Sutherland, & Whishaw, 1983; Kolb et al., 1994). Thus, it was proposed that the frontal cortex would form part of a system with some sort of supramodal function in the control of spatially guided behavior (Kolb et al., 1983). On the other hand data became available which pointed to the involvement of the (pre-)frontal cortex in spatial behavior guided by response learning rather than place learning. Lesions of the prefrontal cortex have been found to be associated with impairments in egocentric spatial learning, both in the adjacent radial arm maze task and in an egocentric version of the Morris water maze task, while failing to show impairments in spatial tasks requiring place learning, e.g., in the cheese board task and the allocentric Morris water maze task (De Bruin, Sanchez-Santed, Heinsbroek, Donker, & Postmes, 1994; De Bruin, Swinkels, & De Brabander, 1997; Kesner et al., 1989; Maaswinkel, Gispen, & Spruijt, 1996). Recently, the involvement of the prefrontal cortex was shown in a response working memory task (Kesner, Hunt, Williams, & Long, 1996). Taken together, these latter studies are indicative of a dissociation between neural systems required for spatial navigation based on either allocentric spatial learning (place learning) or egocentric spatial learning (response learning). The hippocampus and the posterior parietal cortex would be responsible for allocentric spatial learning, while the prefrontal cortex, possibly together with the caudate nucleus (Packard & McGaugh, 1996), would be responsible for egocentric spatial learning. We have conducted the present study to further investigate this concept of a double functional dissociation between the prefrontal cortex and the hippocampus. We have used two versions of the Morris water maze paradigm, its classical “allocentric” (place learning) version (De Bruin et al., 1994; Morris, 1981) and the “egocentric” (response learning) version (De Bruin et al., 1997; Whishaw, 1985). This has the advantage that the two systems (place and taxon) are investigated in the same testing apparatus, that food restriction is not required for either one of the two spatial tasks, and that for both versions the same “reinforcer” (hidden platform) is used, thus facilitating comparisons. Anatomically, bilateral lesions were inflicted in either the medial prefrontal cortex or the fimbria fornix. Rather than the hippocampus proper, this latter area was chosen because of its practical advantages and because its damage, which interferes with afferent and efferent connections of the hippocampal formation, has been shown to result in behavioral effects which are, in general, quite comparable with those of hippocampal damage (Aggleton, Keith, Rawlins, Hunt, & Sahgal, 1992; Cassel, Cassel, Galani, Kelche, Will, & Jarrard, 1998; Maren & Fanselow, 1997; Packard & McGaugh, 1992; Shaw & Aggleton, 1993; Whishaw & Jarrard, 1995).
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
167
MATERIALS AND METHODS Subjects A total of 72 male Wistar rats (Harlan CPB-WU) were used for this study. They were socially housed in groups of four per macrolon cage (54*33*20 cm) with a normal light–dark cycle (light on from 7.00 AM to 7.00 PM ). Food (standard rodent pellets, Hope Farms B. V., Woerden, The Netherlands) and water were freely available. Operations took place approximately 1 week after the arrival of the animals in our laboratory and their weights were then between 250 and 300 g (mean weight, 271.4 ⫾ 12.6). Following a 2week postoperative recovery period the animals were behaviorally tested in a Morris water maze either in a place (allocentric) task or in a response (egocentric) task. In the place task the animals with fimbria fornix (FF) lesions and those with medial prefrontal cortex (mPFC) lesions each had their own control group. In the response task one control group was used for both animals with FF lesions and those with mPFC lesions. Prior to and following surgery all animals were regularly transported to the water maze room, where they were handled and weighed and could acclimate to the testing room. Behavioral Apparatus The animals were either subjected to a place task or a response task in the same water maze. The water maze was a black, circular pool with an inner diameter of 140 cm and walls 34 cm high. It was filled with normal tap water to a depth of 30 cm. The rim of the pool was raised by a 27 cm high transparent screen preventing the animals from climbing out. The water was at room temperature (22⬚C). The pool was divided in four quadrants of equal size by two (imaginary) diagonal lines running through the center of the pool. The quadrants were designated NE, NW, SW, and SE. A removable circular escape platform (diameter, 10 cm) could be positioned in the middle of each quadrant with the center 30 cm away from the wall of the pool. Two types of platforms were used: an invisible one, painted black and always 1.5 cm below the water surface, and a visible one, painted white and always 1.5 cm above the water surface. Both platforms had a rough (metal grid) surface providing sufficient grip for the animal to climb on top of them. Release sites were marked on the outside of the pool, each one directly opposite to either one of the four possible platform positions. The pool was placed in a room dimly lit by white light. The walls of the room were equipped with a variety of spatial cues which remained unchanged over the whole experiment. A radio broadcasting popular music was 1.5 m away from the pool. A video camera was mounted above the pool to record the movements of the animal. They were analyzed with a computerized detection system, “Hawktrack,” a tracking system used with an Acom Archimedes Computer, originally developed and described by Morris (1984). The data were analyzed with the accompanying program “Water Maze,” which provides illustrations of the swimming path along with a number of measures, such as escape latency, path length, average swimming speed, and directionality. Place Task The animals were trained and tested using the following procedures. Spatial training (eight sessions). For half of the animals the invisible black platform
168
DE BRUIN ET AL.
was placed in quadrant NE; for the other half in quadrant SW. Training was conducted in four-trial sessions, with each animal being released into the pool from one of the four release sites. The sequence of the four release sites varied from session to session, but was identical for all animals within one session. Each animal was released into the pool with its head facing the wall. If the platform was not found within the maximal trial duration of 90 s, the animal was guided to the platform. In either case the animal was left on top of the platform for a 30-s period. In between the successive trials of one session the animal was put in a black plastic bucket for a 30-s intertrial interval. During this period fecal bolusses (if present) were removed from the pool, and the transparent wall was wiped clean. Following the last trial of a session the animal was dried with a cloth towel and placed in a clean cage. There were two sessions a day with an interval of approximately 3 h. Reversal training (three sessions). Following 2 days without behavioral testing, reversal training began. The platform was now placed in the quadrant opposite to the one used during spatial training; otherwise all training procedures were identical to the ones described for spatial training. A total of three reversal sessions (each consisting of four trials) was conducted. Visual cue task (three sessions). On the day after the completion of the reversal training the animals were subjected to the visual cue task. Instead of a black invisible platform, a white visible one, extending 1.5 cm above the surface, was used. While the release site of the animal remained the same (opposite to the training quadrant of the spatial training), the platform position was varied over the four quadrants from trial to trial in a random order. Response Task For the response task a total of 12 four-trial sessions was conducted. Both the release site and the platform position varied from trial to trial. The release site varied in the same way as for spatial training and the platform was positioned in a fixed spatial relation to the release site. The animal, facing the wall, was released into the pool. For half of the animals from each group the platform was placed at a fixed distance to the right of the release site; for the other half the platform was placed at the same distance to the left of the release site. Otherwise the procedures were the same as those for the place task, except for the intertrial interval, which was 60 s (instead of 30 s) to allow sufficient time to change the platform position. A pretraining session was conducted on the first testing day to adapt the animals to the demands of this task. Each rat was given some assistance to locate the platform by guiding the animal to the platform at the end of the trial and progressively reducing the maximum swimming time from 90 s (first trial) to 15 s (fourth trial). Surgery One week after arrival in the Institute’s animal facilities the animals were operated under Hypnorm anesthesia (0.1 ml/100 g, im); they either received a bilateral lesion of the fimbria fornix or the medial PFC or they were sham-operated. Following placement
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
169
in a David Kopf stereotaxic apparatus a longitudinal skin incision was made, exposing the skull, and two holes were drilled (2 mm in diameter) in the skull at either side of the midsagittal suture line. At this point in the sham-operated animals, bone pieces were replaced and skin and periosteum were sutured, and the animals were taken out of the stereotaxic apparatus. In the other animals bilateral lesions were made with a highfrequency lesion generator (Radionics Inc., Model RFG-4) with thermistor-type electrodes, with an uninsulated tip 1.5 mm in length and 0.7 mm in diameter. Following incision of the dura mater the electrode was positioned at four different sites in each hemisphere for mPFC lesions and one site in each hemisphere for FF lesions. The coordinates were based on the Paxinos and Watson atlas (1986); anterior posterior (AP) coordinates referred to bregma, lateral (L) to the midsagittal suture line, and ventral (V) to the dura mater. After placement of the electrode at its predetermined site, current was passed and a stable tip temperature of 60⬚C was maintained for a 60-s period, followed by an additional 60-s period with the electrode in situ. For the two types of lesions coordinates were as follows: mPFC (angle ⫹ 16⬚), L ⫽ 1.8, AP ⫽ 2.7, V ⫽ 4.5 and ⫺2.0; L ⫽ 1.8, AP ⫽ 4.2, V ⫽ ⫺3.6 and ⫺2.0; FF (angle 0⬚), L ⫽ 0.9, AP ⫽ ⫺1.3, and V ⫽ ⫺3.5. Following surgery the wound was sutured and each animal was given a postoperative injection of Temgesic (0.03 ml/100 g, sc). Histology At the end of the experiment animals were deeply anesthesized with Nembutal (60 mg/ ml, 0.1 ml/100 g, ip), perfused with saline, followed by 4% paraformaldehyde, and decapitated. Brains were removed from the skull and were fixed in a 4% paraformaldehyde/ 15% sucrose solution. Coronal 25-m cryostat sections, stained with thionin, were prepared for visual inspection, and reconstructions of the lesions were made on templates of the rat brain atlas of Swanson (1992). Data Analysis Mean values (⫾ the standard errors of the mean) of escape latencies (time between release into the pool and climbing on top of the platform) were calculated for each fourtrial session. These data were analyzed with analysis of variance (ANOVA), with GROUP and SESSION as main factors, and with repeated measures on the last factor. When appropriate, posthoc comparisons were made with the Newman–Keuls test (Winer, 1971). The p ⬍ .05 level was accepted as significant. Data on the path lengths (between release site and platform) were also recorded and analyzed. Since the outcome of these analyses was quite similar to those of the escape latencies, they are not further dealt with under Results. RESULTS Histology The largest and smallest mPFC lesions which were included for further analysis are portrayed in Fig. 1. The largest lesion begins at the frontal pole and extends posteriorly for just over 3 mm, slightly more than the smallest lesion. In all lesions damage was
170
DE BRUIN ET AL.
FIG. 1. Reconstructions of the largest (solid line) and the smallest (dotted line) lesions of the medial prefrontal cortex (mPFC) of the animals tested in the place task (left) and the response task (right). Drawings are based on templates of the rat brain atlas of Swanson (1992).
inflicted to areas Fr2 (frontal area 2), ACd (dorsal part of the anterior cingulate cortex), and PL (prelimbic cortex), with variable damage of IL (infralimbic cortex). In a few animals, there was some damage to the most dorsal part of the olfactory structures in the frontal pole region; at the most posterior level marginal damage to the forceps of the corpus callosum was occasionally seen. Figure 2 illustrates the degree of damage to the FF. The lesion had affected the major part of this fiber tract, with some variation in its most (ventro)lateral segment which was spared in some animals. The lesion tract also caused some damage to the cingular cortex. Based on the analysis of the reconstructed lesions 12 animals were discarded from further analysis, which (together with 2 animals that died postoperatively) resulted in the following group sizes: Place learning: FF, n ⫽ 10, and C, n ⫽ 7; mPFC, n ⫽ 9, and C, n ⫽ 8. Response learning: FF, n ⫽ 8, mPFC,
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
171
FIG. 2. Reconstructions of the largest (solid line) and the smallest (dotted line) fimbria fornix (FF) lesions of the animals examined in the place task (top) and the response task (bottom). Drawings are based on templates of the rat brain atlas of Swanson (1992).
n ⫽ 9, and C, n ⫽ 7. The figures illustrate that mPFC lesion size (Fig. 1) and degree of damage to the FF fiber tract (Fig. 2) were quite similar in the animals subjected to the place learning task and those subjected to the response learning task. Behavior Place task, reversal task, and visual task (experiment 1). The time the animals took to find the hidden platform (escape latency) during the training sessions is depicted in Fig. 3 (FF and C) and Fig. 4 (mPFC and C). All animals reduced their escape latencies, which is evidenced by the highly significant session effects for the analyses involving the FF and C animals (F(7, 105) ⫽ 23.83, p ⬍ .001) and the mPFC and C animals (F(7, 98) ⫽ 28.43, p ⬍ .001). However, although learning takes place during the place task, the FF animals were clearly slower than their controls in locating the platform (F(1, 15) ⫽ 16.17, p ⫽ .001). Posthoc analysis shows that during sessions 4 to 8 the escape latencies of the FF animals were significantly higher than those of their controls (Newman– Keuls test, p ⬍ .05). By contrast, there was no significant difference between the mPFC animals and their controls (F(1, 14) ⫽ 1.26, p ⫽ .281). After the change in platform location (first session of reversal task) the mean escape latencies significantly increased, both in the animals with lesions and in their respective controls. This increase was especially evident in the animals with FF lesions (Fig. 3), but
172
DE BRUIN ET AL.
FIG. 3. Escape latencies during the place task, reversal task, and visual task of animals with fimbria fornix lesions (FF, n ⫽ 10) and control-operated animals (control, n ⫽ 7). Group differences are present during the place task ( p ⫽ .001), reversal task ( p ⬍ .001), and visual task ( p ⫽ .038).
perhaps the most striking difference was that in control animals and animals with mPFC lesions there was a rapid drop in the values of the escape latencies: as early as the second reversal session the values were similar to those of the last spatial training session (session 8). In the FF animals this decrease was slower and more gradual, and they differed from
FIG. 4. Escape latencies during the place task, reversal task, and visual task of animals with medial prefrontal cortex lesions (mPFC, n ⫽ 8) and control-operated animals (control, n ⫽ 8). There are no significant group differences.
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
173
their controls (F(1, 15) ⫽ 28.98, p ⬍ .001). This group difference was present during all three reversal sessions (Newman–Keuls test, p ⬍ .05). By contrast, escape latencies during the three reversal sessions were not significantly different between animals with mPFC lesions and their controls (F(1, 14) ⫽ 0.33, p ⫽ .576). During the visual task the FF animals were at least initially, faster in locating the visible platform than their controls, and there was a significant group different (F(1, 15) ⫽ 5.12, p ⫽ .039). Posthoc analysis showed significant group differences during the first and second sessions of the visual task (Newman–Keuls test, p ⬍ .05), a difference which had disappeared by the third session. Although animals with mPFC lesions tended to be somewhat slower when the visual platform was present, the difference with their controls failed to be significant (F(1, 14) ⫽ 3.85, p ⫽ .07). Response task (experiment 2). The response task was clearly more difficult to learn than the place task, and acquisition was a slower process. Animals with FF lesions did not differ significantly from the controls (F(1, 13) ⫽ 0.37, p ⫽ .552). Although the training data of Fig. 5 suggest that the acquisition curves of the two groups were not the same (i.e., the FF animals being initially faster, but slower at the end) this impression is not corroborated by the stastistical analysis, since the group x session interaction was not significant (F(11, 143) ⫽ 1.81, p ⫽ .057). By contrast there was a clear group difference (Fig. 6) between the animals with mPFC lesion and the controls (F(1,14) ⫽ 7.55, p ⫽ .016). The group x session interaction was not significant (F(11, 154) ⫽ 0.57, p ⫽ .854). These statistical data indicate that the learning process in the mPFC animals was slower than but not different from that of the controls. DISCUSSION Both the place and response learning tasks require the retention of information in the form of reference memory. In the first, the platform remains in the same position throughout
FIG. 5. Escape latencies during the response task of animals with fimbria fornix lesions (FF, n ⫽ 8) and control-operated animals (control, n ⫽ 7). The groups are not significantly different.
174
DE BRUIN ET AL.
FIG. 6. Escape latencies during the response task of animals with medial prefrontal cortex lesions (mPFC, n ⫽ 9) and control-operated animals (control, n ⫽ 7). There is a significant group difference ( p ⫽ .016).
the training period. Therefore, in order to learn this task the animal has to remember its location (place or allocentric learning). In the second, the site of the platform varies, but the spatial relation between platform and release site remains unaltered (response or egocentric learning). This study has shown that place learning is affected by FF lesions (which interfere with hippocampal connectivity), but not by lesions of the mPFC. By contrast, response learning is impaired by mPFC lesions, but not by FF lesions. The findings on impaired place learning following FF lesions are in agreement with those of a number of other findings reported in litterature (Cassel et al., 1998; Hannesson & Skelton, 1998; Packard & McGaugh, 1992; Sutherland & Rodriguez, 1989; Whishaw, Cassel, & Jarrard, 1995; Whishaw & Jarrard, 1995). Our observations have shown that animals with FF lesions do show the ability to swim toward the area where the platform is situated, but are less acccurate in finding it. They swim around in its vicinity, before actually climbing on top of it, resulting in longer escape latency times. In fact, their swim patterns resemble the ones mentioned by Wishaw and Jarrard (1995, Fig. 6). By contrast, both control-operated animals and animals with mPFC lesions soon improve their latency times; after approximately three to four sessions they swim directly to the platform. The findings for animals with mPFC lesions are similar to previously reported ones with comparable mPFC lesions (De Bruin et al., 1994) and with more restricted mPFC lesions (Maaswinkel et al., 1996), but differ from those reported by Kolb et al. (1983). We have previously argued (De Bruin et al., 1994) that these differences are probably caused by slight differences in experimental procedures. Several studies with the cheese board task, considered a dry version of the Morris water maze task, showed that an intact medial PFC is not essential for a good performance (Kesner et al., 1989; King & Corwin, 1992). In a recent study, Kesner et al. (1996) have demonstrated that the mPFC mediates working memory for spatial response information but not for spatial location information. Although the Morris water maze task is a test for reference memory, the outcome from Kesner et al.’s experiments further supports the idea that the mPFC is not necessary for place
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
175
navigation (spatial location information processing). In a study on the responses to spatial changes, Poucet (1989) has demonstrated that rats with damage of the medial PFC are able to detect small changes in topographical relations between objects, which suggests an accurate knowledge of the surrounding space. In the visual cue task the animals of the various groups eventually performed at a similar level, which precludes that differences might have been due to other factors, such as impaired visual functions or differences in swimming speed. However, during the first session the animals with lesions were different from the controls, the animals with FF lesions were faster, while the animals with mPFC damage were somewhat slower in finding the visible platform. This latter observation is in agreement with previous results (De Bruin et al., 1994), where it was argued that the change in strategy (from allocentric to visually cued) resulted in this transient impairment. Animals with mPFC lesions are perhaps less flexible than control animals, therefore needing more time to adapt their strategy. It was remarkable that the opposite was seen in the animals with FF lesions. Their initial shorter escape latencies in the visual task could be due to an impairment of their allocentric place learning system, making them more susceptible to responding to other cues, such as a clearly visible platform. This finding is in agreement with results reported by White, Matthews, and Best (1995). In the response learning task, the animals with mPFC lesions were impaired, i.e., slower in reaching the platform than their controls throughout the whole training period, Thus, these findings indicate that the integrity of the mPFC is required for successful response (egocentric) navigation. This result confirms the outcome of previous studies using this task in the Morris water maze (De Bruin et al., 1997) and other egocentric tasks in a radial maze (Kesner et al., 1989; Kolb et al., 1994). Taken together, this evidence supports the role of the medial PFC in processing information based on a response learning based strategy. The animals with FF lesions did not differ significantly from the controls, nor was there a significant group x session interaction. Nevertheless, the data of Fig. 5 suggest that acquisition of the response task is different from that in control animals. The animals with FF lesions were initially faster than controls in reaching the hidden platform, while after a few training sessions the scores of the control animals were similar. By the end of the training control animals appear to be faster than the animals with FF lesions. Two mechanisms, acting either separately or together, could account for this effect. (i) At the beginning of training, the control animals try to locate the platform in the pool using a place strategy. Failing to succeed, they change their strategy and eventually learn the route to the platform by using a taxon strategy (response learning). The predominance of a place system would render the control animals slower than the ones with FF lesions, because place learning is impaired in the latter animals. They start off by searching for the platform using a taxon system, the one which brings the fastest solution. Such an explanation is in line with the mechanisms proposed by Mogenson, Pedersen, Holm, and Bang (1995). Their test paradigm was a modified version of the water maze, where distal cues were restricted in number, form, and dimensions. In this setting parietal cortical lesions even improved performance. This was thought to be due to the same type of mechanism which we offered for the performance of animals with FF lesions in our experiment, i.e., animals with an impaired place learning system (which is not adaptive in this version of the task) are “faster” than normal animals, which rely first of all on their place learning system. The predominance of the place system has previously been
176
DE BRUIN ET AL.
suggested in a study where normal rats used a place strategy to find a hidden platform in a water maze, while atropine-treated rats preferred a taxon strategy, and when taken off atropine, switched to a place strategy (Whishaw, 1985). Other support for this explanation comes from experiments using a cross maze: place and response learning occurred concurrently, with normal rats initially displaying place learning and subsequently shifting to (more succesfull) response learning (Packard & McGaugh, 1992). (ii) Normal rats may use place learning simultaneously with response learning to improve the search for the platform. This could explain why rats with FF lesions (which begin the response task with low latency scores) improve less with progressive training, since they cannot use a proper place strategy to ameliorate their accuracy in reaching the hidden platform. In a study using a Morris water maze Whishaw (1985) has shown that the simultaneous use of various strategies may result in a good performance (hippocampal rats used cued and response navigation quite succesfully to solve a place problem). This is no direct evidence for the use of the place system to improve performance in a response task (although visual observations, based on the inspection of swim paths, are in support, De Bruin et al., 1997), but it suggests a common mechanism whereby animals use a combination of strategies to solve a problem. In summary, a double functional dissociation between prefrontal cortex and fimbria fornix has been established, with the first one mediating response learning and the second one place learning. These findings were mainly compared with related results obtained in studies with rats as experimental subjects. In addition, they show similarity with data from human and (nonhuman) primate studies. In brain-damaged human patients Semmes, Weinstein, Ghent, and Teuber (1963) distinguished between two types of spatial disorientation, involving either “the body” or “the space external to the body.” The first type of deficiency was especially seen in patients with anterior (frontal) cortical damage and the second type in patients with posterior (parietal) cortical damage. Monkeys with parietal cortical lesions were impaired in extrapersonal spatial tasks, whereas frontal cortical damage did not affect performance of such tasks (Ungerleider & Brody, 1977). Using a different experimental approach, recording single neuron activity, Feigenbaum and Rolls (1991) have reported that in the primate hippocampus there is a preponderance of allocentric encoding cells, while those encoding egocentric information are scarce. Finally, although the present experiments were designed to study the two spatial navigation systems separately, it should be emphasised that they are likely to be used concurrently. REFERENCES Aggleton, J. P., Keith, A. B., Rawlins, J. N. P., Hunt, P. R., & Sahgal, A. (1992). Removal of the hippocampus and transection of the fornix produce comparable deficits on delayed non-matching to position by rats. Behavioural Brain Research, 52, 61–71. Cassel, J.-C., Cassel, S., Galani, R., Kelche, C., Will, B., & Jarrard, L. (1998). Fimbria-fornix vs selective hippocampal lesions in rats: Effects on locomotor activity and spatial learning and memory. Neurobiology of Learning and Memory, 69, 22–45. De Brabander, J. M., De Bruin, J. P. C., & Van Eden, C. G. (1991). Comparison of the effects of neonatal and adult medial prefrontal cortex lesions on food hoarding and spatial delayed alternation. Behavioural Brain Research, 42, 67–75. De Bruin, J. P. C., Sanchez-Santed, F., Heinsbroek, R. P. W., Donker, A. & Postmes, P. (1994). A behavioral analysis of rats with damage to the medial prefrontal cortex using the Morris water maze: Evidence for behavioral flexibility, but not for impaired spatial navigation, Brain Research, 652, 323–333.
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
177
De Bruin, J. P. C., Swinkels, W. A. M., & De Brabander, J. M. (1997). Response learning of rats in a Morris water maze: Involvement of the medial prefrontal cortex. Behavioural Brain Research, 85, 47–55. DiMattia, B. D., & Kesner, R. P. (1988). Spatial cognitive maps: Differential role of parietal cortex and hippocampal formation. Behavioral Neuroscience, 102, 471–480. Feigenbaum, J. D., & Rolls, E. T. (1991). Allocentric and egocentric information processing in the hippocampal formation of the behaving primate. Psychobiology, 19, 21–40. Hannesson, D. K., & Skelton, R. W. (1998). Recovery of spatial performance in the Morris water maze following bilateral transection of the fimbria/fornix in rats. Behavioural Brain Research, 90, 35–56. Kesner, R. P., Farnsworth, G., & DiMattia, B. (1989). Double dissociation of egocentric and allocentric space following medial prefrontal and parietal cortex lesions in the rat. Behavioral Neuroscience, 103, 956–961. Kesner, R. P., Hunt, M. E., Williams, J. M., & Long, J. M. (1996). Prefrontal cortex and working memory for spatial response, spatial location, and visual object information in the rat. Cerebral Cortex, 6, 311–318. King, V., & Corwin, J. V. (1992). Spatial deficits and hemispheric asymmetries in the rat following unilateral and bilateral lesions of posterior parietal or medial agranular cortex. Behavioural Brain Research, 50, 53–68. Kolb, B., Pittman, K., Sutherland, R. J., & Whishaw, I. Q. (1982). Dissociation of the contributions of the prefrontal cortex and dorsomedial thalamic nucleus to spatially guided behavior in the rat. Behavioural Brain Research, 6, 365–378. Kolb, B., Buhrmann, K., McDonald, R., & Sutherland, R. J. (1994). Dissociation of the medial prefrontal, posterior parietal, and posterior temporal cortex for spatial navigation and recognition memory in the rat. Cerebral Cortex, 4, 664–680. Kolb, B., Sutherland, R. J., & Whishaw, I. Q. (1983). A Comparison of the contributions of the frontal and parietal association cortex to spatial localization in rats. Behavioral Neuroscience, 97, 13–27. Maaswinkel, H., Gispen, W.-H., & Spruijt, B. M. (1996). Effects of an electrolytic lesion of the prelimbic area on anxiety-related and cognitive tasks in the rat. Behavioural Brain Research, 79, 51–59. Maren, S., & Fanselow, M. S. (1997). Electrolytic lesions of the fimbria/fornix, dorsal hippocampus, or entorhinal cortex produce anterograde deficits in contextual fear conditioning in rats. Neurobiology of Learning and Memory, 67, 142–149. Mogenson, J., Pedersen, T. K., Holm, S., & Bang, L. E. (1995). Prefrontal cortical mediation of rats’ place learning in a modified water maze. Brain Research Bulletin, 38, 425–435. Morris, R. G. M. (1981). Spatial localization does not require the presence of local cues. Learning and Motivation, 12, 239–260. Morris, R. G. M. (1984). Developments of a water-maze procedure for studying spatial learning in the rat. The Journal of Neuroscience Methods, 11, 47–60. Morris, R. G. M., Garrud, P., Rawlins, J. N. P., & O’Keefe, J. (1982). Place navigation impaired in rats with hippocampal lesions. Nature, 297, 681–683. O’Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon. Packard, M. G., & McGaugh, J. L. (1992). Double dissociation of fornix and caudate nucleus lesions on acquisition of two water maze tasks: Further evidence for multiple memory systems. Behavioral Neuroscience, 106, 439–446. Packard, M. G., & McGaugh, J. L. (1996). Inactivation of hippocampus or caudate nucleus with lidocaine differentially affects expression of place and response learning. Neurobiology of Learning and Memory, 65, 65–72. Paxinos, G., & Watson, C. (1986). The rat brain in stereotaxic coordinates (2nd ed). San Diego: Academic Press. Petrides, M., & Iversen, S. D. (1979). Restricted posterior parietal lesions in the rhesus monkey and performance on visuospatial tasks. Brain Research, 161, 63–77. Pohl, W. (1973). Dissociation of spatial discrimination deficits following frontal and parietal lesions in monkeys. The Journal of Comparative Physiological Psychology, 82, 227–239. Poucet, B. (1989). Object exploration, habituation, and response to a spatial change in rats following septal or medial frontal cortical damage. Behavioral Neuroscience, 103, 1009–1016.
178
DE BRUIN ET AL.
Poucet, B. (1993). Spatial cognitive map in animals: New hypothesis on their structure and neural mechanisms. Psychological Review, 100, 163–182. Semmes, J., Weinstein, S., Ghent, L., & Teuber, H.-L. (1963). Correlates of impaired orientation in personal and extrapersonal space. Brain, 86, 747–772. Shaw, C., & Aggleton, J. P. (1993). The effects of fornix and medial prefrontal lesions on delayed non-matchingto-sample by rats. Behavioural Brain Research, 54, 91–102. Sutherland, R. J., & Rodriguez, A. J. (1989). The role of the fornix/fimbria and some related subcortical structures in place learning and memory. Behavioural Brain Research, 32, 265–277. Swanson, L. W. (1992). Brain maps: Structure of the rat brain. Amsterdam: Elsevier. Ungerleider, L. G., & Brody, B. A. (1977). Extrapersonal spatial orientation: The role of posterior parietal, anterior frontal, and inferotemporal cortex. Experimental Neurology, 56, 265–280. White, A. M., Matthews, D. B., & Best, P. J. (1995). Fimbria/fornix lesions enhance performance on a nonspatial trajectory task in the water maze. Society of Neuroscience Abstracts, 21, 1942. Whishaw, I. Q. (1985). Cholinergic receptor blockade in the rat impairs locale but not taxon strategies for place navigation in a swimming pool. Behavioral Neuroscience, 99, 979–1005. Whishaw, I. Q., Cassel, J. C., & Jarrard, L. (1995). Rats with fimbria-fornix lesions display a place response in a swimming pool: A dissociation between getting there and knowing where. Journal of Neuroscience, 15, 5779–5788. Whishaw, I. Q., & Jarrard, L. (1995). Similarities vs differences in place learning and circadian activity in rats after fimbria-fornix transection or ibotenate removal of hippocampal cells. Hippocampus, 5, 595–604. Winer, B. J. (1971). Statistical principles in experimental design. New York: McGraw–Hill. Winocur, G. (1991). Functional dissociation of the hippocampus and prefrontal cortex in learning and memory. Psychobiology, 19, 11–20.
Place and Response Learning of Rats in a Morris Water Maze: Differential Effects of Fimbria Fornix and Medial Prefrontal Cortex Lesions Jan P. C. de Bruin, Marta P. Moita,1 Hannie M. de Brabander, and Ruud N. J. M. A. Joosten Graduate School of Neurosciences Amsterdam, Netherlands Institute for Brain Research, Amsterdam, The Netherlands
The question examined in this study is concerned with a possible functional dissociation between the hippocampal formation and the prefrontal cortex in spatial navigation. Wistar rats with hippocampal damage (inflicted by a bilateral lesion of the fimbria fornix), rats with damage to the medial prefrontal cortex, and controloperated rats were examined for their performance in either one of two different spatial tasks in a Morris water maze, a place learning task (requiring a locale system), or a response learning task (requiring a taxon system). Performance of the classical place learning (allocentric) task was found to be impaired in rats with lesions of the fimbria fornix, but not in rats with damage of the medial prefrontal cortex, while the opposite effect was found in the response learning (egocentric) task. These findings are indicative of a double functional dissociation of these two brain regions with respect to the two different forms of spatial navigation. When the place learning task was modified by relocating the platform, the impairment in animals with fimbria fornix lesions was even more pronounced than before, while the performance of animals with medial prefrontal cortex lesions was similar to that of their controls. When the task was again modified by changing the hidden platform for a clearly visible one (visual cue task), the animals with fimbria fornix lesions had, at least initially, shorter latencies than their controls. By contrast, in the animals with medial prefrontal cortex damage this change led to a slight increase in escape latency. 䉷 2001 Academic Press Key Words: prefrontal cortex; fimbria fornix; place learning; response learning; Morris water maze; rat.
This research was supported by a Leonardo da Vinci training placement grant from FORBITEC (Associac¸a˜o para a formac¸a˜o te´cnica em biotecnologia) to Marta P. Moita. The authors thank Henk Stoffels for preparing the lesion reconstruction figures. Address correspondence and reprint requests to Jan P. C. de Bruin, The Netherlands Institute for Brain Research, Meibergdreef 33, 1105 AZ Amsterdam, The Netherlands. Fax: ⫹31 20 6961006. E-mail: [email protected]. 1 Current address: Instituto Gulbenkian Cieˆencia, 2781 Oeiras Codex, Portugal. 1074-7427/01 $35.00 Copyright 䉷 2001 by Academic Press All rights of reproduction in any form reserved.
164
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
165
INTRODUCTION Animals use two different orientation systems to locate a target in space (e.g., an invisible platform in a water maze), a place (or locale) system and a taxon system (O’Keefe and Nadel, 1978). A place system implies the development of a spatial cognitive map, an internal representation of the relations among distal (extramaze) cues. Since the spatial relations between places are independent of the animal’s location, it has also been termed an allocentric cognitive map (e.g., Kesner, Farnsworth, & DiMattia, 1989). The learning process involved is called place learning or allocentric learning. A taxon system refers to animals following a route and, in the absence of distal cues, orientation may take place by a repeated use of relatively fixed motor movements to locate the target (e.g., turn left, turn right, move at a particular angle, etc.). Dimattia and Kesner (1988) have termed this a praxic rule, while others refer to it as (position) response learning, or egocentric (spatial) learning (Kesner et al., 1989; Packard & McGaugh, 1996). Place learning and response learning are the subject of the present study. The concept of a spatial cognitive map together with its cerebral localization has drawn the attention of many researchers to place learning, while the neural correlates of response learning have been the subject of fewer studies. Searching for the neural mechanisms underlying place learning, several lines of evidence led O’Keefe and Nadel (1978) to propose the hippocampus as the main neural structure involved. Strong support came from electrophysiological recordings of hippocampal cells (place cells) which are activated when the animal is in a particular location in an open field (O’Keefe & Nadel, 1978). Additionally, data from hippocampal lesions, both in humans and in various other mammalian species (Morris, Garrud, Rawlins, & O’Keefe, 1982; O’Keefe & Nadel, 1978), have indicated that the hippocampus is involved in the processing of spatial information. This model has received wide support, and research on spatial navigation has focused on the hippocampus as the main neural structure involved in spatial learning and memory. More recently, evidence has accumulated pointing to a multiple memory system where different cortical regions (such as the posterior parietal cortex, frontal cortex) would also be involved in the processing of spatial information (Packard & McGaugh, 1992; Winocur, 1991). Posterior parietal cortical lesions have been shown to induce severe impairments in the classical place learning version of both the Morris water maze task (DiMattia & Kesner, 1988; Kolb, Buhrmann, McDonald, & Sutherland, 1994) and in the cheese board task, a “dry” version of the former (Kesner et al., 1989). Primates with parietal cortical damage are impaired in the learning of mazes and in making visual–spatial associations (Petrides & Iversen, 1979). Frontal brain damage has also been reported to disrupt the performance of tasks with a spatial component. In monkeys an impairment in right/left discriminations was observed (Pohl, 1973). In rats frontal cortical damage was shown to lead to deficits in spatial delayed alternation in a T-maze (De Brabander, de Bruin, & Van Eden, 1991) and to impaired performance in the (egocentric) adjacent arm task in a radial maze (Kesner et al., 1989; Kolb et al., 1994). With this new line of thought (a multiple memory system with different cortical regions involved) Poucet (1993) has proposed a model for spatial orientation which emphasises place navigation. The hippocampus and the posterior parietal cortex would function in a cooperative manner to form an allocentric cognitive map. According to this model the
166
DE BRUIN ET AL.
hippocampus would be essential during the learning period, but not for the retention of spatial information, i.e., after hippocampal damage an animal would still be able to orient itself in a well-know environment. Thus, this area would merely be responsible for the establishment of topological relations between places. The posterior parietal cortex would be involved in both acquisition and retention, playing a crucial role in processing the geometric structure of the environment. Data from a study on the effects of hippocampal and posterior parietal cortical damage in the Morris water maze task strongly support Poucet’s model (DiMatttia & Kesner, 1988). The frontal cortex (less prominent in this model) would be involved in solving problems with a spatial component. There has been some contradicting evidence concerning the involvement of the frontal cortical region in spatial learning. Some studies, using the Morris water maze task, have shown a disruption of place learning in rats with frontal lesions (Kolb, Pittman, Sutherland, & Whishaw, 1982; Kolb, Sutherland, & Whishaw, 1983; Kolb et al., 1994). Thus, it was proposed that the frontal cortex would form part of a system with some sort of supramodal function in the control of spatially guided behavior (Kolb et al., 1983). On the other hand data became available which pointed to the involvement of the (pre-)frontal cortex in spatial behavior guided by response learning rather than place learning. Lesions of the prefrontal cortex have been found to be associated with impairments in egocentric spatial learning, both in the adjacent radial arm maze task and in an egocentric version of the Morris water maze task, while failing to show impairments in spatial tasks requiring place learning, e.g., in the cheese board task and the allocentric Morris water maze task (De Bruin, Sanchez-Santed, Heinsbroek, Donker, & Postmes, 1994; De Bruin, Swinkels, & De Brabander, 1997; Kesner et al., 1989; Maaswinkel, Gispen, & Spruijt, 1996). Recently, the involvement of the prefrontal cortex was shown in a response working memory task (Kesner, Hunt, Williams, & Long, 1996). Taken together, these latter studies are indicative of a dissociation between neural systems required for spatial navigation based on either allocentric spatial learning (place learning) or egocentric spatial learning (response learning). The hippocampus and the posterior parietal cortex would be responsible for allocentric spatial learning, while the prefrontal cortex, possibly together with the caudate nucleus (Packard & McGaugh, 1996), would be responsible for egocentric spatial learning. We have conducted the present study to further investigate this concept of a double functional dissociation between the prefrontal cortex and the hippocampus. We have used two versions of the Morris water maze paradigm, its classical “allocentric” (place learning) version (De Bruin et al., 1994; Morris, 1981) and the “egocentric” (response learning) version (De Bruin et al., 1997; Whishaw, 1985). This has the advantage that the two systems (place and taxon) are investigated in the same testing apparatus, that food restriction is not required for either one of the two spatial tasks, and that for both versions the same “reinforcer” (hidden platform) is used, thus facilitating comparisons. Anatomically, bilateral lesions were inflicted in either the medial prefrontal cortex or the fimbria fornix. Rather than the hippocampus proper, this latter area was chosen because of its practical advantages and because its damage, which interferes with afferent and efferent connections of the hippocampal formation, has been shown to result in behavioral effects which are, in general, quite comparable with those of hippocampal damage (Aggleton, Keith, Rawlins, Hunt, & Sahgal, 1992; Cassel, Cassel, Galani, Kelche, Will, & Jarrard, 1998; Maren & Fanselow, 1997; Packard & McGaugh, 1992; Shaw & Aggleton, 1993; Whishaw & Jarrard, 1995).
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
167
MATERIALS AND METHODS Subjects A total of 72 male Wistar rats (Harlan CPB-WU) were used for this study. They were socially housed in groups of four per macrolon cage (54*33*20 cm) with a normal light–dark cycle (light on from 7.00 AM to 7.00 PM ). Food (standard rodent pellets, Hope Farms B. V., Woerden, The Netherlands) and water were freely available. Operations took place approximately 1 week after the arrival of the animals in our laboratory and their weights were then between 250 and 300 g (mean weight, 271.4 ⫾ 12.6). Following a 2week postoperative recovery period the animals were behaviorally tested in a Morris water maze either in a place (allocentric) task or in a response (egocentric) task. In the place task the animals with fimbria fornix (FF) lesions and those with medial prefrontal cortex (mPFC) lesions each had their own control group. In the response task one control group was used for both animals with FF lesions and those with mPFC lesions. Prior to and following surgery all animals were regularly transported to the water maze room, where they were handled and weighed and could acclimate to the testing room. Behavioral Apparatus The animals were either subjected to a place task or a response task in the same water maze. The water maze was a black, circular pool with an inner diameter of 140 cm and walls 34 cm high. It was filled with normal tap water to a depth of 30 cm. The rim of the pool was raised by a 27 cm high transparent screen preventing the animals from climbing out. The water was at room temperature (22⬚C). The pool was divided in four quadrants of equal size by two (imaginary) diagonal lines running through the center of the pool. The quadrants were designated NE, NW, SW, and SE. A removable circular escape platform (diameter, 10 cm) could be positioned in the middle of each quadrant with the center 30 cm away from the wall of the pool. Two types of platforms were used: an invisible one, painted black and always 1.5 cm below the water surface, and a visible one, painted white and always 1.5 cm above the water surface. Both platforms had a rough (metal grid) surface providing sufficient grip for the animal to climb on top of them. Release sites were marked on the outside of the pool, each one directly opposite to either one of the four possible platform positions. The pool was placed in a room dimly lit by white light. The walls of the room were equipped with a variety of spatial cues which remained unchanged over the whole experiment. A radio broadcasting popular music was 1.5 m away from the pool. A video camera was mounted above the pool to record the movements of the animal. They were analyzed with a computerized detection system, “Hawktrack,” a tracking system used with an Acom Archimedes Computer, originally developed and described by Morris (1984). The data were analyzed with the accompanying program “Water Maze,” which provides illustrations of the swimming path along with a number of measures, such as escape latency, path length, average swimming speed, and directionality. Place Task The animals were trained and tested using the following procedures. Spatial training (eight sessions). For half of the animals the invisible black platform
168
DE BRUIN ET AL.
was placed in quadrant NE; for the other half in quadrant SW. Training was conducted in four-trial sessions, with each animal being released into the pool from one of the four release sites. The sequence of the four release sites varied from session to session, but was identical for all animals within one session. Each animal was released into the pool with its head facing the wall. If the platform was not found within the maximal trial duration of 90 s, the animal was guided to the platform. In either case the animal was left on top of the platform for a 30-s period. In between the successive trials of one session the animal was put in a black plastic bucket for a 30-s intertrial interval. During this period fecal bolusses (if present) were removed from the pool, and the transparent wall was wiped clean. Following the last trial of a session the animal was dried with a cloth towel and placed in a clean cage. There were two sessions a day with an interval of approximately 3 h. Reversal training (three sessions). Following 2 days without behavioral testing, reversal training began. The platform was now placed in the quadrant opposite to the one used during spatial training; otherwise all training procedures were identical to the ones described for spatial training. A total of three reversal sessions (each consisting of four trials) was conducted. Visual cue task (three sessions). On the day after the completion of the reversal training the animals were subjected to the visual cue task. Instead of a black invisible platform, a white visible one, extending 1.5 cm above the surface, was used. While the release site of the animal remained the same (opposite to the training quadrant of the spatial training), the platform position was varied over the four quadrants from trial to trial in a random order. Response Task For the response task a total of 12 four-trial sessions was conducted. Both the release site and the platform position varied from trial to trial. The release site varied in the same way as for spatial training and the platform was positioned in a fixed spatial relation to the release site. The animal, facing the wall, was released into the pool. For half of the animals from each group the platform was placed at a fixed distance to the right of the release site; for the other half the platform was placed at the same distance to the left of the release site. Otherwise the procedures were the same as those for the place task, except for the intertrial interval, which was 60 s (instead of 30 s) to allow sufficient time to change the platform position. A pretraining session was conducted on the first testing day to adapt the animals to the demands of this task. Each rat was given some assistance to locate the platform by guiding the animal to the platform at the end of the trial and progressively reducing the maximum swimming time from 90 s (first trial) to 15 s (fourth trial). Surgery One week after arrival in the Institute’s animal facilities the animals were operated under Hypnorm anesthesia (0.1 ml/100 g, im); they either received a bilateral lesion of the fimbria fornix or the medial PFC or they were sham-operated. Following placement
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
169
in a David Kopf stereotaxic apparatus a longitudinal skin incision was made, exposing the skull, and two holes were drilled (2 mm in diameter) in the skull at either side of the midsagittal suture line. At this point in the sham-operated animals, bone pieces were replaced and skin and periosteum were sutured, and the animals were taken out of the stereotaxic apparatus. In the other animals bilateral lesions were made with a highfrequency lesion generator (Radionics Inc., Model RFG-4) with thermistor-type electrodes, with an uninsulated tip 1.5 mm in length and 0.7 mm in diameter. Following incision of the dura mater the electrode was positioned at four different sites in each hemisphere for mPFC lesions and one site in each hemisphere for FF lesions. The coordinates were based on the Paxinos and Watson atlas (1986); anterior posterior (AP) coordinates referred to bregma, lateral (L) to the midsagittal suture line, and ventral (V) to the dura mater. After placement of the electrode at its predetermined site, current was passed and a stable tip temperature of 60⬚C was maintained for a 60-s period, followed by an additional 60-s period with the electrode in situ. For the two types of lesions coordinates were as follows: mPFC (angle ⫹ 16⬚), L ⫽ 1.8, AP ⫽ 2.7, V ⫽ 4.5 and ⫺2.0; L ⫽ 1.8, AP ⫽ 4.2, V ⫽ ⫺3.6 and ⫺2.0; FF (angle 0⬚), L ⫽ 0.9, AP ⫽ ⫺1.3, and V ⫽ ⫺3.5. Following surgery the wound was sutured and each animal was given a postoperative injection of Temgesic (0.03 ml/100 g, sc). Histology At the end of the experiment animals were deeply anesthesized with Nembutal (60 mg/ ml, 0.1 ml/100 g, ip), perfused with saline, followed by 4% paraformaldehyde, and decapitated. Brains were removed from the skull and were fixed in a 4% paraformaldehyde/ 15% sucrose solution. Coronal 25-m cryostat sections, stained with thionin, were prepared for visual inspection, and reconstructions of the lesions were made on templates of the rat brain atlas of Swanson (1992). Data Analysis Mean values (⫾ the standard errors of the mean) of escape latencies (time between release into the pool and climbing on top of the platform) were calculated for each fourtrial session. These data were analyzed with analysis of variance (ANOVA), with GROUP and SESSION as main factors, and with repeated measures on the last factor. When appropriate, posthoc comparisons were made with the Newman–Keuls test (Winer, 1971). The p ⬍ .05 level was accepted as significant. Data on the path lengths (between release site and platform) were also recorded and analyzed. Since the outcome of these analyses was quite similar to those of the escape latencies, they are not further dealt with under Results. RESULTS Histology The largest and smallest mPFC lesions which were included for further analysis are portrayed in Fig. 1. The largest lesion begins at the frontal pole and extends posteriorly for just over 3 mm, slightly more than the smallest lesion. In all lesions damage was
170
DE BRUIN ET AL.
FIG. 1. Reconstructions of the largest (solid line) and the smallest (dotted line) lesions of the medial prefrontal cortex (mPFC) of the animals tested in the place task (left) and the response task (right). Drawings are based on templates of the rat brain atlas of Swanson (1992).
inflicted to areas Fr2 (frontal area 2), ACd (dorsal part of the anterior cingulate cortex), and PL (prelimbic cortex), with variable damage of IL (infralimbic cortex). In a few animals, there was some damage to the most dorsal part of the olfactory structures in the frontal pole region; at the most posterior level marginal damage to the forceps of the corpus callosum was occasionally seen. Figure 2 illustrates the degree of damage to the FF. The lesion had affected the major part of this fiber tract, with some variation in its most (ventro)lateral segment which was spared in some animals. The lesion tract also caused some damage to the cingular cortex. Based on the analysis of the reconstructed lesions 12 animals were discarded from further analysis, which (together with 2 animals that died postoperatively) resulted in the following group sizes: Place learning: FF, n ⫽ 10, and C, n ⫽ 7; mPFC, n ⫽ 9, and C, n ⫽ 8. Response learning: FF, n ⫽ 8, mPFC,
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
171
FIG. 2. Reconstructions of the largest (solid line) and the smallest (dotted line) fimbria fornix (FF) lesions of the animals examined in the place task (top) and the response task (bottom). Drawings are based on templates of the rat brain atlas of Swanson (1992).
n ⫽ 9, and C, n ⫽ 7. The figures illustrate that mPFC lesion size (Fig. 1) and degree of damage to the FF fiber tract (Fig. 2) were quite similar in the animals subjected to the place learning task and those subjected to the response learning task. Behavior Place task, reversal task, and visual task (experiment 1). The time the animals took to find the hidden platform (escape latency) during the training sessions is depicted in Fig. 3 (FF and C) and Fig. 4 (mPFC and C). All animals reduced their escape latencies, which is evidenced by the highly significant session effects for the analyses involving the FF and C animals (F(7, 105) ⫽ 23.83, p ⬍ .001) and the mPFC and C animals (F(7, 98) ⫽ 28.43, p ⬍ .001). However, although learning takes place during the place task, the FF animals were clearly slower than their controls in locating the platform (F(1, 15) ⫽ 16.17, p ⫽ .001). Posthoc analysis shows that during sessions 4 to 8 the escape latencies of the FF animals were significantly higher than those of their controls (Newman– Keuls test, p ⬍ .05). By contrast, there was no significant difference between the mPFC animals and their controls (F(1, 14) ⫽ 1.26, p ⫽ .281). After the change in platform location (first session of reversal task) the mean escape latencies significantly increased, both in the animals with lesions and in their respective controls. This increase was especially evident in the animals with FF lesions (Fig. 3), but
172
DE BRUIN ET AL.
FIG. 3. Escape latencies during the place task, reversal task, and visual task of animals with fimbria fornix lesions (FF, n ⫽ 10) and control-operated animals (control, n ⫽ 7). Group differences are present during the place task ( p ⫽ .001), reversal task ( p ⬍ .001), and visual task ( p ⫽ .038).
perhaps the most striking difference was that in control animals and animals with mPFC lesions there was a rapid drop in the values of the escape latencies: as early as the second reversal session the values were similar to those of the last spatial training session (session 8). In the FF animals this decrease was slower and more gradual, and they differed from
FIG. 4. Escape latencies during the place task, reversal task, and visual task of animals with medial prefrontal cortex lesions (mPFC, n ⫽ 8) and control-operated animals (control, n ⫽ 8). There are no significant group differences.
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
173
their controls (F(1, 15) ⫽ 28.98, p ⬍ .001). This group difference was present during all three reversal sessions (Newman–Keuls test, p ⬍ .05). By contrast, escape latencies during the three reversal sessions were not significantly different between animals with mPFC lesions and their controls (F(1, 14) ⫽ 0.33, p ⫽ .576). During the visual task the FF animals were at least initially, faster in locating the visible platform than their controls, and there was a significant group different (F(1, 15) ⫽ 5.12, p ⫽ .039). Posthoc analysis showed significant group differences during the first and second sessions of the visual task (Newman–Keuls test, p ⬍ .05), a difference which had disappeared by the third session. Although animals with mPFC lesions tended to be somewhat slower when the visual platform was present, the difference with their controls failed to be significant (F(1, 14) ⫽ 3.85, p ⫽ .07). Response task (experiment 2). The response task was clearly more difficult to learn than the place task, and acquisition was a slower process. Animals with FF lesions did not differ significantly from the controls (F(1, 13) ⫽ 0.37, p ⫽ .552). Although the training data of Fig. 5 suggest that the acquisition curves of the two groups were not the same (i.e., the FF animals being initially faster, but slower at the end) this impression is not corroborated by the stastistical analysis, since the group x session interaction was not significant (F(11, 143) ⫽ 1.81, p ⫽ .057). By contrast there was a clear group difference (Fig. 6) between the animals with mPFC lesion and the controls (F(1,14) ⫽ 7.55, p ⫽ .016). The group x session interaction was not significant (F(11, 154) ⫽ 0.57, p ⫽ .854). These statistical data indicate that the learning process in the mPFC animals was slower than but not different from that of the controls. DISCUSSION Both the place and response learning tasks require the retention of information in the form of reference memory. In the first, the platform remains in the same position throughout
FIG. 5. Escape latencies during the response task of animals with fimbria fornix lesions (FF, n ⫽ 8) and control-operated animals (control, n ⫽ 7). The groups are not significantly different.
174
DE BRUIN ET AL.
FIG. 6. Escape latencies during the response task of animals with medial prefrontal cortex lesions (mPFC, n ⫽ 9) and control-operated animals (control, n ⫽ 7). There is a significant group difference ( p ⫽ .016).
the training period. Therefore, in order to learn this task the animal has to remember its location (place or allocentric learning). In the second, the site of the platform varies, but the spatial relation between platform and release site remains unaltered (response or egocentric learning). This study has shown that place learning is affected by FF lesions (which interfere with hippocampal connectivity), but not by lesions of the mPFC. By contrast, response learning is impaired by mPFC lesions, but not by FF lesions. The findings on impaired place learning following FF lesions are in agreement with those of a number of other findings reported in litterature (Cassel et al., 1998; Hannesson & Skelton, 1998; Packard & McGaugh, 1992; Sutherland & Rodriguez, 1989; Whishaw, Cassel, & Jarrard, 1995; Whishaw & Jarrard, 1995). Our observations have shown that animals with FF lesions do show the ability to swim toward the area where the platform is situated, but are less acccurate in finding it. They swim around in its vicinity, before actually climbing on top of it, resulting in longer escape latency times. In fact, their swim patterns resemble the ones mentioned by Wishaw and Jarrard (1995, Fig. 6). By contrast, both control-operated animals and animals with mPFC lesions soon improve their latency times; after approximately three to four sessions they swim directly to the platform. The findings for animals with mPFC lesions are similar to previously reported ones with comparable mPFC lesions (De Bruin et al., 1994) and with more restricted mPFC lesions (Maaswinkel et al., 1996), but differ from those reported by Kolb et al. (1983). We have previously argued (De Bruin et al., 1994) that these differences are probably caused by slight differences in experimental procedures. Several studies with the cheese board task, considered a dry version of the Morris water maze task, showed that an intact medial PFC is not essential for a good performance (Kesner et al., 1989; King & Corwin, 1992). In a recent study, Kesner et al. (1996) have demonstrated that the mPFC mediates working memory for spatial response information but not for spatial location information. Although the Morris water maze task is a test for reference memory, the outcome from Kesner et al.’s experiments further supports the idea that the mPFC is not necessary for place
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
175
navigation (spatial location information processing). In a study on the responses to spatial changes, Poucet (1989) has demonstrated that rats with damage of the medial PFC are able to detect small changes in topographical relations between objects, which suggests an accurate knowledge of the surrounding space. In the visual cue task the animals of the various groups eventually performed at a similar level, which precludes that differences might have been due to other factors, such as impaired visual functions or differences in swimming speed. However, during the first session the animals with lesions were different from the controls, the animals with FF lesions were faster, while the animals with mPFC damage were somewhat slower in finding the visible platform. This latter observation is in agreement with previous results (De Bruin et al., 1994), where it was argued that the change in strategy (from allocentric to visually cued) resulted in this transient impairment. Animals with mPFC lesions are perhaps less flexible than control animals, therefore needing more time to adapt their strategy. It was remarkable that the opposite was seen in the animals with FF lesions. Their initial shorter escape latencies in the visual task could be due to an impairment of their allocentric place learning system, making them more susceptible to responding to other cues, such as a clearly visible platform. This finding is in agreement with results reported by White, Matthews, and Best (1995). In the response learning task, the animals with mPFC lesions were impaired, i.e., slower in reaching the platform than their controls throughout the whole training period, Thus, these findings indicate that the integrity of the mPFC is required for successful response (egocentric) navigation. This result confirms the outcome of previous studies using this task in the Morris water maze (De Bruin et al., 1997) and other egocentric tasks in a radial maze (Kesner et al., 1989; Kolb et al., 1994). Taken together, this evidence supports the role of the medial PFC in processing information based on a response learning based strategy. The animals with FF lesions did not differ significantly from the controls, nor was there a significant group x session interaction. Nevertheless, the data of Fig. 5 suggest that acquisition of the response task is different from that in control animals. The animals with FF lesions were initially faster than controls in reaching the hidden platform, while after a few training sessions the scores of the control animals were similar. By the end of the training control animals appear to be faster than the animals with FF lesions. Two mechanisms, acting either separately or together, could account for this effect. (i) At the beginning of training, the control animals try to locate the platform in the pool using a place strategy. Failing to succeed, they change their strategy and eventually learn the route to the platform by using a taxon strategy (response learning). The predominance of a place system would render the control animals slower than the ones with FF lesions, because place learning is impaired in the latter animals. They start off by searching for the platform using a taxon system, the one which brings the fastest solution. Such an explanation is in line with the mechanisms proposed by Mogenson, Pedersen, Holm, and Bang (1995). Their test paradigm was a modified version of the water maze, where distal cues were restricted in number, form, and dimensions. In this setting parietal cortical lesions even improved performance. This was thought to be due to the same type of mechanism which we offered for the performance of animals with FF lesions in our experiment, i.e., animals with an impaired place learning system (which is not adaptive in this version of the task) are “faster” than normal animals, which rely first of all on their place learning system. The predominance of the place system has previously been
176
DE BRUIN ET AL.
suggested in a study where normal rats used a place strategy to find a hidden platform in a water maze, while atropine-treated rats preferred a taxon strategy, and when taken off atropine, switched to a place strategy (Whishaw, 1985). Other support for this explanation comes from experiments using a cross maze: place and response learning occurred concurrently, with normal rats initially displaying place learning and subsequently shifting to (more succesfull) response learning (Packard & McGaugh, 1992). (ii) Normal rats may use place learning simultaneously with response learning to improve the search for the platform. This could explain why rats with FF lesions (which begin the response task with low latency scores) improve less with progressive training, since they cannot use a proper place strategy to ameliorate their accuracy in reaching the hidden platform. In a study using a Morris water maze Whishaw (1985) has shown that the simultaneous use of various strategies may result in a good performance (hippocampal rats used cued and response navigation quite succesfully to solve a place problem). This is no direct evidence for the use of the place system to improve performance in a response task (although visual observations, based on the inspection of swim paths, are in support, De Bruin et al., 1997), but it suggests a common mechanism whereby animals use a combination of strategies to solve a problem. In summary, a double functional dissociation between prefrontal cortex and fimbria fornix has been established, with the first one mediating response learning and the second one place learning. These findings were mainly compared with related results obtained in studies with rats as experimental subjects. In addition, they show similarity with data from human and (nonhuman) primate studies. In brain-damaged human patients Semmes, Weinstein, Ghent, and Teuber (1963) distinguished between two types of spatial disorientation, involving either “the body” or “the space external to the body.” The first type of deficiency was especially seen in patients with anterior (frontal) cortical damage and the second type in patients with posterior (parietal) cortical damage. Monkeys with parietal cortical lesions were impaired in extrapersonal spatial tasks, whereas frontal cortical damage did not affect performance of such tasks (Ungerleider & Brody, 1977). Using a different experimental approach, recording single neuron activity, Feigenbaum and Rolls (1991) have reported that in the primate hippocampus there is a preponderance of allocentric encoding cells, while those encoding egocentric information are scarce. Finally, although the present experiments were designed to study the two spatial navigation systems separately, it should be emphasised that they are likely to be used concurrently. REFERENCES Aggleton, J. P., Keith, A. B., Rawlins, J. N. P., Hunt, P. R., & Sahgal, A. (1992). Removal of the hippocampus and transection of the fornix produce comparable deficits on delayed non-matching to position by rats. Behavioural Brain Research, 52, 61–71. Cassel, J.-C., Cassel, S., Galani, R., Kelche, C., Will, B., & Jarrard, L. (1998). Fimbria-fornix vs selective hippocampal lesions in rats: Effects on locomotor activity and spatial learning and memory. Neurobiology of Learning and Memory, 69, 22–45. De Brabander, J. M., De Bruin, J. P. C., & Van Eden, C. G. (1991). Comparison of the effects of neonatal and adult medial prefrontal cortex lesions on food hoarding and spatial delayed alternation. Behavioural Brain Research, 42, 67–75. De Bruin, J. P. C., Sanchez-Santed, F., Heinsbroek, R. P. W., Donker, A. & Postmes, P. (1994). A behavioral analysis of rats with damage to the medial prefrontal cortex using the Morris water maze: Evidence for behavioral flexibility, but not for impaired spatial navigation, Brain Research, 652, 323–333.
SPATIAL LEARNING, PREFRONTAL CORTEX, FIMBRIA FORNIX
177
De Bruin, J. P. C., Swinkels, W. A. M., & De Brabander, J. M. (1997). Response learning of rats in a Morris water maze: Involvement of the medial prefrontal cortex. Behavioural Brain Research, 85, 47–55. DiMattia, B. D., & Kesner, R. P. (1988). Spatial cognitive maps: Differential role of parietal cortex and hippocampal formation. Behavioral Neuroscience, 102, 471–480. Feigenbaum, J. D., & Rolls, E. T. (1991). Allocentric and egocentric information processing in the hippocampal formation of the behaving primate. Psychobiology, 19, 21–40. Hannesson, D. K., & Skelton, R. W. (1998). Recovery of spatial performance in the Morris water maze following bilateral transection of the fimbria/fornix in rats. Behavioural Brain Research, 90, 35–56. Kesner, R. P., Farnsworth, G., & DiMattia, B. (1989). Double dissociation of egocentric and allocentric space following medial prefrontal and parietal cortex lesions in the rat. Behavioral Neuroscience, 103, 956–961. Kesner, R. P., Hunt, M. E., Williams, J. M., & Long, J. M. (1996). Prefrontal cortex and working memory for spatial response, spatial location, and visual object information in the rat. Cerebral Cortex, 6, 311–318. King, V., & Corwin, J. V. (1992). Spatial deficits and hemispheric asymmetries in the rat following unilateral and bilateral lesions of posterior parietal or medial agranular cortex. Behavioural Brain Research, 50, 53–68. Kolb, B., Pittman, K., Sutherland, R. J., & Whishaw, I. Q. (1982). Dissociation of the contributions of the prefrontal cortex and dorsomedial thalamic nucleus to spatially guided behavior in the rat. Behavioural Brain Research, 6, 365–378. Kolb, B., Buhrmann, K., McDonald, R., & Sutherland, R. J. (1994). Dissociation of the medial prefrontal, posterior parietal, and posterior temporal cortex for spatial navigation and recognition memory in the rat. Cerebral Cortex, 4, 664–680. Kolb, B., Sutherland, R. J., & Whishaw, I. Q. (1983). A Comparison of the contributions of the frontal and parietal association cortex to spatial localization in rats. Behavioral Neuroscience, 97, 13–27. Maaswinkel, H., Gispen, W.-H., & Spruijt, B. M. (1996). Effects of an electrolytic lesion of the prelimbic area on anxiety-related and cognitive tasks in the rat. Behavioural Brain Research, 79, 51–59. Maren, S., & Fanselow, M. S. (1997). Electrolytic lesions of the fimbria/fornix, dorsal hippocampus, or entorhinal cortex produce anterograde deficits in contextual fear conditioning in rats. Neurobiology of Learning and Memory, 67, 142–149. Mogenson, J., Pedersen, T. K., Holm, S., & Bang, L. E. (1995). Prefrontal cortical mediation of rats’ place learning in a modified water maze. Brain Research Bulletin, 38, 425–435. Morris, R. G. M. (1981). Spatial localization does not require the presence of local cues. Learning and Motivation, 12, 239–260. Morris, R. G. M. (1984). Developments of a water-maze procedure for studying spatial learning in the rat. The Journal of Neuroscience Methods, 11, 47–60. Morris, R. G. M., Garrud, P., Rawlins, J. N. P., & O’Keefe, J. (1982). Place navigation impaired in rats with hippocampal lesions. Nature, 297, 681–683. O’Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon. Packard, M. G., & McGaugh, J. L. (1992). Double dissociation of fornix and caudate nucleus lesions on acquisition of two water maze tasks: Further evidence for multiple memory systems. Behavioral Neuroscience, 106, 439–446. Packard, M. G., & McGaugh, J. L. (1996). Inactivation of hippocampus or caudate nucleus with lidocaine differentially affects expression of place and response learning. Neurobiology of Learning and Memory, 65, 65–72. Paxinos, G., & Watson, C. (1986). The rat brain in stereotaxic coordinates (2nd ed). San Diego: Academic Press. Petrides, M., & Iversen, S. D. (1979). Restricted posterior parietal lesions in the rhesus monkey and performance on visuospatial tasks. Brain Research, 161, 63–77. Pohl, W. (1973). Dissociation of spatial discrimination deficits following frontal and parietal lesions in monkeys. The Journal of Comparative Physiological Psychology, 82, 227–239. Poucet, B. (1989). Object exploration, habituation, and response to a spatial change in rats following septal or medial frontal cortical damage. Behavioral Neuroscience, 103, 1009–1016.
178
DE BRUIN ET AL.
Poucet, B. (1993). Spatial cognitive map in animals: New hypothesis on their structure and neural mechanisms. Psychological Review, 100, 163–182. Semmes, J., Weinstein, S., Ghent, L., & Teuber, H.-L. (1963). Correlates of impaired orientation in personal and extrapersonal space. Brain, 86, 747–772. Shaw, C., & Aggleton, J. P. (1993). The effects of fornix and medial prefrontal lesions on delayed non-matchingto-sample by rats. Behavioural Brain Research, 54, 91–102. Sutherland, R. J., & Rodriguez, A. J. (1989). The role of the fornix/fimbria and some related subcortical structures in place learning and memory. Behavioural Brain Research, 32, 265–277. Swanson, L. W. (1992). Brain maps: Structure of the rat brain. Amsterdam: Elsevier. Ungerleider, L. G., & Brody, B. A. (1977). Extrapersonal spatial orientation: The role of posterior parietal, anterior frontal, and inferotemporal cortex. Experimental Neurology, 56, 265–280. White, A. M., Matthews, D. B., & Best, P. J. (1995). Fimbria/fornix lesions enhance performance on a nonspatial trajectory task in the water maze. Society of Neuroscience Abstracts, 21, 1942. Whishaw, I. Q. (1985). Cholinergic receptor blockade in the rat impairs locale but not taxon strategies for place navigation in a swimming pool. Behavioral Neuroscience, 99, 979–1005. Whishaw, I. Q., Cassel, J. C., & Jarrard, L. (1995). Rats with fimbria-fornix lesions display a place response in a swimming pool: A dissociation between getting there and knowing where. Journal of Neuroscience, 15, 5779–5788. Whishaw, I. Q., & Jarrard, L. (1995). Similarities vs differences in place learning and circadian activity in rats after fimbria-fornix transection or ibotenate removal of hippocampal cells. Hippocampus, 5, 595–604. Winer, B. J. (1971). Statistical principles in experimental design. New York: McGraw–Hill. Winocur, G. (1991). Functional dissociation of the hippocampus and prefrontal cortex in learning and memory. Psychobiology, 19, 11–20.