Prefrontal cortical mediation of rats' place learning in a modified water maze

Brain ResearchBulletin,Vol.38, No. 5, pp. 425-434, 1995 Copyright© 1995ElsevierScienceInc. Printed in the USA.All rights reserved 0361-9230/95 $9.50 + .00

Pergamon 0361-9230(95)02009-G

Prefrontal Cortical Mediation of Rats' Place Learning in a Modified Water Maze JESPER

MOGENSEN,

1 THOMAS

KIRK PEDERSEN,

SQREN

HOLM AND LIA EVI BANG

Laboratory of Neuropsychiatry, Department of Pharmacology, Universityof Copenhagen and Rigshospitalet, the University Hospital, Denmark [Received 23 November 1994; Accepted 2 June 1995] ABSTRACT: The acquisition of a place learning task in a water maze modified from the "standard" set-up by restriction of distal cues and addition of "proximal" cues (ping-pong balls in fixed positions on the surface of the water) was tested in three groups of rats: (I) animals subjected to bilateral ablation of the anteromedial prefrontal cortex, (11) rats in which the parietal "association" cortex had been removed bilaterally, and (111)a sham operated control group. The task acquisition of the p ~ t a l l y ablated group was significantly impaired, whereas the animals in which the parietal cortex had been removed acquired the task as quickly as the contbrol group. Upon reaching criterion level performance all animals were tested on "challenge" sessions on which the cues were manipulated. Such "challenges" demonstrated that the animals of all three groups discriminated between the distal cues and utilized such a discrimination for nayigational purposes.

ery is a shift from the normally preferred solution strategy to an alternative strategy. At the neural level such a shift is accomplished by a shift in the pattern of neural activity. This shift in the activation pattern of the brain represents a "deselection" of the (damaged or inactivated) structures mediating the normally preferred solution strategy accompanied by an increased activation of neural systems mediating alternative behavioural strategies [13,15,16,30]. The nature of the normally selected solution strategy is determined by the behavioural demands of the task facing the rat. In brain-damaged individuals the dominating spatial strategy is dictated by an interaction between the demands of the task and the type of neural damage suffered by the animal. Consequently, attempts to study the neural substrate of various spatial strategies must address the ways in which the potentially "spatial" structures of the brain participate in mediation of a broad spectrum of "spatial" tasks. For each of the potentially relevant structures, for example, the hippocampus and the prefrontal as well as parietal "association" cortices, it should be established to what degree lesions are associated with impaired acquisition or retention of a spectrum of spatial tasks. The spectrum of tasks should be composed in such a way that the widest possible range of spatial strategies are represented as normally preferred solution strategies. In addition, the behavioural analysis within each task should preferably be able to identify the nature of the spatial strategies eventually selected by each experimental group. In situations where groups of lesioned animals are able to acquire a particular task it may be relevant to subject the performance of such animals to pharmacological [16,30] or surgical [15] challenges. The outcome of such challenges may be able to demonstrate whether a normal proficiency of task mediation is accomplished by an alternative neural substrate. Place learning in water mazes is one of the frequently studied "spatial" tasks. In the "standard" set-up of this procedure [e.g., 17], the swimming rat is allowed freely to observe the room. Such a set-up offers the animal an abundance of three-dimensionally arranged objects and patterns. In the "standard" set-up the normally selected solution strategy appears to be a "mapping" procedure [e.g., 17,18,23]. Acquisition of the "standard" place learning task is significantly impaired by pharmacological blockade of the cholinergic system [e.g., 25,26], lesions of the hippocampus [e.g., 2,19,23,24], and ablation of the prefrontal cortex

KEY WORDS: Prefrontal cortex, Parietal cortex, Place leaming, Water maze, Rat.

INTRODUCTION Numerous studies have addressed the neural substrate of spatial orientation and various types of "spatial memory." Although consensus has been reached on certain issues, for example, that the hippocampal formation appears to be involved in mediation of certain types of spatial functions, many issues remain unresolved. One such research area is the identification of the types of spatial tasks to which cortical areas of the rat contribute. Faced with a spatial task, rats may engage in a variety of behavioural strategies while attempting task solution [e.g., 27]. For any particular problem the majority of nonlesioned rats will rather rapidly select the same (and presumably most appropriate) behavioural strategy--the "normally preferred solution strategy" [16]. The individual strategies (or "hypotheses" in the terminology of Krechevsky [10]) seem to be mediated by relatively independent neural systems [e.g., 20; J. Mogensen, L. H. Saerup, & G. W6rtwein, in preparation]. When a lesion or a pharmacological blockade inactivates the neural substrate of the normally preferred solution strategy, the behavioural task may still be acquired and performed at a normal level of proficiency [ 13,15,16,30]. Apparently, the behavioural basis of such a recov-

i Requests for reprints should be addressed to Jesper Mogensen, Laboratory of Neuropsychiatry, Rigshospitalet-6102, Blegdamsvej 9, DK-2100 Copenhagen 0, Denmark. 425

426

[7,8,23]. Ablation of the parietal "association" cortex allows normal task acquisition in small water mazes [8], but a lesionassociated impairment can be detected in larger mazes [6,9]. Lesions have been able to provoke a selection of altemative behavioural strategies, for example, "praxis" strategies [e.g., 2,18]. Animals in which aspects of the synaptic plasticity within the hippocampus have been prevented pharmacologically are able to acquire the "standard" place learning task to a normal level of proficiency, but apparently this "recovery" process is associated with selection of an alternative neural substrate [ 13,15,16]. Studies utilizing place learning in the "standard" set-up of the water maze are obviously able to provide important information about the neural substrates of a number of spatial strategies. It is, howe v e r - - a s argued above--important to supplement such studies by experiments scrutinizing the neural substrate of place learning in water mazes that are modified from the "standard" set-up in such ways that the normally preferred solution strategy differs from the "mapping" seen with the "standard" set-up. We have recently [30; J. Mogensen, L. H. Christensen, J. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted] introduced a modified place learning set-up in which the normally preferred solution strategy apparently differs from the "mapping" strategy (as well as from the " c u e " and "praxis" strategies). In our modified place learning task the "extramaze" cues are offered in the form of a restricted number of geometrical patterns on the inner surface of a curtain spanning the circumference of the maze. The "two-dimensional" arrangement of a restricted number of distal cues is likely to prevent or impair the application of a "mapping" strategy. An indication that the normally preferred solution strategy of the modified place learning task differs from the "mapping" strategy is the rather long acquisition period required by normal animals [30; J. Mogensen, L. H. Christensen, J. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted]. This slow acquisition rate is in sharp contrast to the rapid task acquisition seen when normal rats are faced with a "standard" place learning task in a water maze of similar dimensions [e.g., 16]. No specific cue marks the position of the platform, and, consequently, "cue" learning is not likely to be the dominant strategy. A pure "praxis" strategy would cause the task solution of the animals to be independent by the positions of the distal cues. Because behavioural test procedures (see below) have established that normal animals discriminate between individual cues and utilize such discriminations for navigational purposes, the normally preferred solution strategy of the modified place learning task cannot be a pure "praxis" strategy [30; J. Mogensen, L. H. Christensen, J. Johansson, G. WOrtwein, L. E. Bang, & S. Holm, submitted]. Studies addressing the neural substrate of the modified place learning task have demonstrated that when compared to the substrate of place learning of the "mapping" type, the neural substrate of task mediation is modified in such a way that unimpaired task acquisition can occur in spite of both cholinergic muscarinic receptor blockade [J. Mogensen, L. H. Christensen, A. Johansson, G. WiSrtwein, L. E. Bang, & S. Holm, submitted] and lesions of the hippocampus [30]. Administration of pharmacological and surgical "challenges" [see 16] upon successful task acquisition demonstrated that the normal level of place learning accomplished by rats deprived of the cholinergic or hippocampal mechanisms was accomplished by compensatory mechanisms within the catecholaminergic systems and the prefrontal cortex [30; J. Mogensen, L. H. Christensen, A. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted]. As indicated by these results, the degree of cholinergic and hippocampal task mediation differs between the commonly studied "standard" ("mapping") place learning task and the modified task in which distal cues are re-

MOGENSEN ET AL.

stricted. A variant of the modified place learning task offers both the restricted distal cues and additional cues in the form of three "proximal" cues--three ping-pong balls marking the midpoint of three of the quadrants of the maze (the platform being located in the middle of the fourth quadrant of the maze). Although effects of hippocampal lesions have only been studied in the modified place learning task without proximal cues, it has been demonstrated that cholinergic muscarinic receptor blockade leaves the acquisition of both the above described modified place learning tasks unimpaired [J. Mogensen, L. H. Christensen, A. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted]. The acquisition of the two modified place learning tasks described above has so far never been studied in animals subjected to ablation of the prefrontal or parietal "association" cortices. Consequently, an important step toward clarifying the neural substrate of these newly developed tasks is to study the acquisition of either task variant upon lesions of the prefrontal or parietal cortex. In this study the focus was on the modified place learning variant in which the restricted distal cues are supplemented by the proximal cues. The purpose of the study was to scrutinize the acquisition of this modified task by three groups of rats: animals subjected to ablation of the entire anteromedial prefrontal cortex, animals subjected to ablation of the parietal "association" cortex, and a sham operated control group. The distal cues offered--four different, but equally sized symbols--were arranged in such a way that each discriminandum marked the midpoint of the perimeter of one of the four quadrants. Although the position of the distal cues remained constant throughout training, the task performance of the animals was challenged by counter-clockwise rotations of the distal cues on the 2 days subsequent to the session on which criterion proficiency of performance was reached. On the first postcriterion session the discriminanda were rotated 90 °, on the second postcriterion session they were rotated 45 °. Because center-tocenter distance between discriminanda was 90 °, the first rotation gave rise to a situation in which the four discriminanda were still situated at the midpoints of the perimeters of the four quadrants, but in such a way that the symbol normally designating the SE quadrant had been shifted to the NE quadrant, and so forth. After the 45 ° rotation the discriminanda no longer marked the midpoints of the perimeters of the quadrants. If the navigational strategy of an animal was based on a discrimination between the four discriminanda, the performance of such an animal would be impaired by both the 90 ° rotation and the 45 ° rotation. If, however, a rat would only utilize the fact that the discriminanda marked the midpoints of the perimeters of the quadrants, the performance of such an animal would be impaired by the 45 ° rotation, while being unaffected by the 90 ° rotation. To determine whether the animals relied upon the proximal cues for successful task performance, we included a final challenge session in which the proximal cues were removed. Following the tradition of our previous studies of place learning in the modified experimental set-up [30]; J. Mogensen, L. H. Christensen, A. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted] we selected a behavioural criterion that demanded that each of the five swims of the criterion session should be completed in less than 15 s. Even in a relatively large water m a z e - such as the one utilized in this s t u d y - - 15 s may appear to be a somewhat long duration of a criterion level swim. It should, however, be remembered that the rather short swim durations measured in large water mazes have been obtained in the "standard" place learning set-up [e.g., 16], where a "mapping" strategy is presumably employed. Our previous studies of the modified place learning tasks [30; J. Mogensen, L. H. Christensen, A. Johansson, G. Wtirtwein, L. E. Bang, & S. Holm, submitted] have

PREFRONTAL CORTEX AND MODIFIED PLACE LEARNING

indicated that in these tasks normal animals reach asymptotic quality of performance at a mean swim duration of approximately 15s. The neural substrate of place learning based on restricted distal cues seems to differ from the neural substrate of the "standard" place learning by reduced functional contributions from both the cholinergic and hippocampal systems (see above). Functional contributions from the prefrontal cortex appear to be amongst the compensatory processes allowing normal task acquisition in spite of pharmacological blockade of the cholinergic system [J. Mogensen, L. H. Christensen, A. Johansson, G. Wtirtwein, L. E. Bang, & S. Holm, submitted]. Consequently, it could be speculated that even the neural substrate of normal acquisition of this modified type of place learning receives a relatively strong contribution from the prefrontal cortex. We therefore hypothesized that ablations of the prefrontal cortex would impair task acquisition significantly. No particular prediction was made as to the performance of the group subjected to ablation of the parietal cortex. In the "standard" place learning task, such ablations fail to interfere with task acquisition in small water mazes [8], whereas a marginal impairment is seen in large water mazes of a size comparable to the apparatus of this experiment [6,9]. METHODS

Subjects The subjects were 34 experimentally naive, male Wistar albino rats weighing approximately 250 g at the beginning of the experiment. They were housed in single cages with commercial rat chow and water always available. The animals' living quarters were maintained on a 12 h light/dark cycle (on 6.00 h). The rats were randomly divided into three experimental groups: sham surgery (n = 10), and groups with ablation of the total anteromedial prefrontal cortex (n = 12) or the parietal "association" cortex (n = 12). Forty-eight hours prior to the first training session all animals were dyed black on the neck.

Apparatus The water maze consisted of a circular water tank measuring 1.85 m in diameter by 0.50 m in height. The tank was filled to a depth of 0.28 m with water that was rendered opaque and white. The inner walls of the water tank were painted white. In a basic design similar to that of Morris [17] the water maze was illuminated through a white, opaque "ceiling" 1.30 m above the surface of the water. Through a hole in the center of the "ceiling" the entire surface of the water could be observed via a video camera. A circular glider was attached along the outer edge of the "ceiling" of the maze. A black curtain spanned the distance from the glider to the upper edge of the water tank. Four points along the circumference of the water tank were arbitrarily designated North (N), South (S), East (E), and West (W), thus dividing the maze into four "quadrants." Four white discriminanda (55 cm × 55 cm) were attached to the inner surface of the circular curtain. The four discriminanda differed in neither material nor area. They were all positioned with their lower edge 0.30 m above the upper edge of the water tank, and in the standard setting the position of each symbol marked the midpoint of the perimeter of a quadrant (X: NE quadrant; square: SE quadrant; cross: SW quadrant; and diamond: NW quadrant). The circular arrangement of the glider allowed the arrangement of the distal cues (discriminanda) to be modified by rotation of the curtain around the center of the maze. Throughout all parts of the experiment one circular, submerged platform (diameter: 12.5 cm)

427

remained in a fixed position in the middle of the SE quadrant. The upper surface of the platform remained covered by 1.0-1.5 cm of water and consisted of a white plastic mesh--offering the animals an efficient grip of the platform. Three yellow plastic pingpong balls (diameter: 3.7 cm) were left in constant positions on the surface of the water (each ball was anchored to the bottom of the water tank). The balls were positioned in the middle of the NW, NE, and SW quadrants. The signal of the camera was processed in a VP 112-scanning unit (HVS Image Analysing, Kingston, UK), which, at a rate of 50 Hz, provided the XY-coordinates of the high contrast mark on the back of the animal's neck. The VP112 was connected to a computer that stored the data and controlled the experiment. The experimenter could provide direct input to the computer via the keyboard or via four push-buttons mounted on the outer surface of the water tank (in the positions N, S, E, and W). While all parameters involving time were measured in seconds, all distances were measured in arbitrary units ("pixels").

Behavioural Procedure The general behavioural procedures were similar to those described by Mogensen et al. [13,16]. Each animal was given five trials (swims) daily. Each trial had as its start position one of the locations N, S, E, or W. A given start position was not allowed to be selected on more than two consecutive trials, and the start positions were otherwise randomly selected. At the beginning of a trial the experimenter released the animal and pressed the pushbutton. The activation of the push-button provided the computer input marking the beginning of the trial. If the coordinates of the swimming animal at any point in time coincided with the group of coordinates describing the position of the submerged platform, the program would register "rat on platform." If an animal after 60 s of swimming had not reached the platform, the trial would be terminated by the program and the swimming animal immediately picked up by the experimenter. After having picked up such a rat the experimenter would lift it onto the platform and allow the rat 30 s of undisturbed activities on the platform. If a swimming rat reached the platform the animal was allowed 30 s of undisturbed activity on the platform. The training was initiated 28 days after the animals had been subjected to surgery and continued with the standard arrangement of distal cues (as described in the Apparatus section) until a behavioural criterion was reached or for 50 sessions. The criterion demanded that the animal in a given session on all five trials reached the platform after a swim duration of less than 15.0 s. On the first session following the day of criterion performance the animals were subjected to a test on which one feature of the experimental set-up had been changed: the curtain had been rotated 90° counter-clockwise. On the day following the "90 ° rotation" session another "rotation" test was administered. For the second "rotation session" the curtain had been rotated 45 ° counter clockwise when compared to the standard setting of the discriminanda. Following the "45 ° rotation" two training sessions with the standard configuration of distal and proximal cues were given, and on the day following the last of these two "normalization" sessions the final test was administered. During that session the constellation of distal cues remained in the standard positions while the proximal cues had been removed. For each parameter from each session the mean value for the five trials was calculated, and the further data analysis was performed on such mean values of five swims. The following parameters were considered: the total swim distance, the total duration of a swim, the average speed of a swim, and the "heading angle error." The "heading angle error" was calculated in the

428

following manner: the set of coordinates of the point at which the animal was first detected during a given swim was registered and named A; the set of coordinates of the point at which the rat was found 1.0 s later was registered and named B; the vector from A to B was calculated; the vector from A to the center of the platform was calculated; the heading angle error was calculated as the number of degrees of the angle between these two vectors. During all behavioural procedures the experimenter was kept ignorant about the group to which an individual rat belonged.

MOGENSEN ET AL.

the ablated animals but for the incision into the dura and removal of cortical tissue (incision into the dura was avoided to reduce the risk of inflicting damage to the pia and thus damaging the cortical tissue). The prefrontal cortical group suffered bilateral ablation of the entire anteromedial prefrontal cortex [for details, see 12,14]. The parietal cortical group underwent bilateral removal of the cortex within a rectangle between the coronal lines at - 2 mm and - 6 mm caudal to bregma and the parasagittal lines 2 mm and 6 mm lateral to the midline [for details, see 4,28].

Histology Statistical Analysis Nonparametric statistics were chosen for the statistical analysis because normal distribution of the data could not be expected. The parameters considered from the acquisition period were: the number of training sessions required to reach the behavioural criterion (see above) and all parameters registered (see above) from the following sessions: all sessions from the beginning of the experiment until the session on which the first animal reached the behavioural criterion (i.e., the sessions during which all subjects of the three groups still participated in training), the criterion session, and the two last sessions prior to the criterion session. All these parameters were subjected to the KruskalWallis one-way analysis of variance test [22] for comparisons between the three experimental groups. Wherever the analysis of variance revealed significant group differences the performance of individual groups were compared using the Mann-Whitney U-test, two-tailed (however, because our hypothesis--see Introduction-predicted that the performance of the prefrontally ablated group would be impaired relative to the sham operated controls, comparisons between the prefrontally lesioned group and the sham operated group on the parameters number of sessions to behavioural criterion, total swim distance, and total duration of a swim only considered the possibility of inferior quality of performance of the prefrontal group, and, consequently, onetailed tests were applied) [22]. The performance on the "90 ° rotation," "45 ° rotation," and "distal cues only" sessions was analysed for each of the three groups independently. Such analysis was based on comparisons between the performance on a given test and the performance of the same group on the previous day (the criterion, "90 ° rotation," and "second normalization" session, respectively). For all parameters on the test sessions comparison to the performance on the previous session was performed using the Wilcoxon matched pairs test, one-tailed (the one-tailed test was selected because manipulations of distal and proximal cues could not be expected to cause improvement of performance--reliance upon a particular class of cues would be associated with impaired performance, whereas nonreliance upon a class of cues would only be expected to allow a continuation of criterion level performance) [22]. If an animal failed to reach the behavioural criterion (and consequently was trained for 50 sessions), the data from such a rat was not included in the analysis of the performance on the subsequent test sessions. It should, however, be emphasized that even such an animal would be included in the analysis of the behaviour up to and including the criterion session.

Surgery Twenty-eight days prior to the first acquisition session all animals were subjected to surgery. Anaesthesia was achieved by intraperitoneal injection of Equithesin (3.3 ml/kg) and 1% Atropine sulphate (0.9 mg/kg). The ablations were made by subpial aspiration with the help of an operating microscope. The sham operated control group underwent procedures similar to those of

Upon termination of the behavioural tests the animals were deeply anaesthetized and transcardially perfused with saline followed by a 10% formalin in saline solution. After perfusion the brains were removed and allowed to sink in 10% formalin in saline solution containing 20% sucrose. When sunk the brains were rapidly frozen in isopentane cooled to -50°C, stored at -80°C, and cut in a cryostat. The Nissl-stained sections were examined, and the lesions were reconstructed with the aid of a microfiche reader. RESULTS

Anatomy The lesions are illustrated in Fig. 1. Histological examinations revealed no signs of edema or other neuropathological changes, and the groups did not differ in any other respect than the locus (and presence) of ablation. The lesions of the parietal cortex included the parietal "association" area as defined by neuroanatomical criteria [9,28]. These lesions were somewhat larger than the minimum indicated by Kolb and Walkey [9] and corresponded largely to the locus and size of parietal lesions in our previous studies [e.g., 4,28]. The lesions of the anteromedial prefrontal cortex were restricted to the anteromedial cortical area of both hemispheres. In all animals the prefrontal ablations removed the major portion of the anteromedial cortex--a lesion that leaves only the suprarhinal part of the prefrontal cortex undamaged [e.g., 3,14]. In both ablated groups the lesions occasionally invaded the subcortical white matter. Subcortical damage did, however, never include other subcortical structures. In neither lesion group did the cortical damage reach significantly beyond the intended cortical areas. Consequently, no animal had to be excluded from the experiment on the basis of the histological examinations. Four animals from the prefrontal group failed to reach the behavioural criterion within 50 sessions (see below). A comparison between the locus and extent of lesions in these four animals and in the remaining part of the prefrontal group failed to demonstrate any obvious differences. In general, within each of the lesioned groups comparisons between, on the one hand, the level of behavioural modifications and, on the other hand, the size and locus of lesion failed to demonstrate any "within lesion group" correlation between the behavioural and anatomical parameters.

Behaviour Aspects of the behavioural results are shown in Figs. 2-4. Although all animals from the parietal and sham groups reached the behavioural criterion in less than 50 sessions, four animals from the prefrontal group failed to satisfy the criterion within 50 sessions. The performance of all animals was included in the analysis of behaviour during the acquisition phase of the experiment (up to and including the criterion session), but the four prefrontal animals that did not satisfy the behavioural criterion within 50 sessions were excluded from the subsequent "chal-

PREFRONTAL CORTEX AND MODIFIED PLACE LEARNING

PC

429

PFC

12.0~

FIG. 1. The lesions of the two experimentalgroups subjected to cortical ablations within the parietal cortex (PC) and the prefrontal cortex (PFC), respectively. Black = the tissue removed in all animals. Horizontal stripes = the tissue removed in at least one rat. The diagrams show levels 12.0, 10.5, 9.5, 7.6, 6.8, 5.6, 5.3, 3.9, 2.9, 1.8, 1.0, and 0.1 in front of the interaural line [21]. lenge" sessions. The analysis of the number of sessions required by the three groups to reach the behavioural criterion (Fig. 2) demonstrated that the prefrontal group required significantly (p < 0.001) more sessions than both the parietal and sham groups. The number of sessions required by the parietal group to reach the behavioural criterion did not differ from the corresponding value of the sham operated control group. This analysis of the duration of acquisition periods clearly indicated that, although parietal lesions apparently did not interfere with the duration of the learning process, prefrontal lesions impaired task acquisition severely--in some animals such lesions even prevented criterion task proficiency to be acquired within the allotted number of sessions. The first rat reached criterion performance on the third session, and the analysis of potential group differences on individual sessions therefore focused on Sessions 1-3, on the two sessions preceding the criterion session, and on the criterion session itself.

Aspects of the behavioural performance during the acquisition period are illustrated in Fig. 3. As discussed elsewhere [e,g., 11,17], the swim distance is the parameter that most reliably demonstrates the "cognitive" aspects of task acquisition and task performance (swim duration can more easily be influenced by factors such as purely "motoric" symptoms). In this experiment the task performance on the second session demonstrated the swim distance of the prefrontally ablated group to be significantly longer than the corresponding parameters of the two other groups. This demonstration of a significantly impaired task acquisition by the prefrontal group was further emphasized by the swim durations of the same session: a parameter on which the value of the prefrontally lesioned animals differed significantly from the corresponding values of both the other groups. The swim distance on the second session of the parietally ablated animals was significantly shorter than the corresponding value of the sham group. This result shows that during certain phases of

430

C 0 .m I,...

MOGENSEN ET AL. 50

-X-*~IX XX

0 0 0 .m

25

! 0

1.

i=

z PC

i

I

PFC

Sham

FIG. 2. Number of sessions required by the three experimental groups (PC: the group subjected to ablations of the parietal cortex; PFC: the group in which the prefrontal cortex had been ablated; and Sham: the sham operated control group). Values are given as medians with ranges. ***Significantly (p < 0.001) different from the sham group. ~x~Significantly(p < 0.001) different from the parietal group. the acquisition process, the parietally lesioned animals were able to demonstrate a performance significantly superior to that of the normal control animals. Because both swim distance and swim duration demonstrated a significant impairment of the prefrontal group, it seems unlikely that this impairment would be a reflection of a purely "motoric" disturbance. This conclusion is further emphasized by the fact that the swim speeds of the three experimental groups never differed significantly during the acquisition period. In addition to the significant group differences illustrated in Fig. 3, the heading angle error of the prefrontally ablated group on the third session was significantly (p < 0.01) larger than the corresponding value of the control group. Informal observations of the swim patterns of individual animals during various phases of task solution indicated a rather consistent pattern across all three experimental groups. Initially the rats appeared mainly to swim along the periphery of the maze, but soon they initiated swim patterns that included all parts of the maze. A more consistent pattern of task solution did, however--in at least some animals--begin to develop even during the first couple of sessions. This pattern eventually was seen in practically all animals of all groups--although individual differences were conspicuous. The eventual search pattern seemed in most cases to include a strategy in which the animal, upon being released, headed for the proximal cue closest to the release position. Upon reaching this cue the animal conducted what appeared to be a sequence of "partial swims," which brought the animal from one proximal cue to the next and eventually from one of the proximal cues to the position of the platform. Upon reaching the behavioural criterion, the performance of all animals (except for the four prefrontal rats who failed to reach the behavioural criterion within 50 sessions) was challenged by the two "rotation" sessions and the "distal cues only" session. Aspects of the behavioural performance of the three experimental groups on these test sessions and the sessions preceding the individual test sessions are illustrated in Fig. 4. The 90 ° rotation of the distal cues significantly increased both the swim distance and the swim duration of all three experimental groups. This is a clear demonstration that the animals of all groups relied so heavily upon the discrimination between individual distal cues that the task performance deteriorated significantly when the original po-

sitions of individual cues no longer remained constant (see Introduction). Furthermore, the swim distance of the prefrontally ablated animals was significantly increased w h e n - - o n the "distal cues only" session--the proximal cues were removed. This may indicate that the prefrontal group relied more heavily than the two other groups upon the three proximal cues. It should, however, be noticed that for neither experimental group had the quality of task performance rereached criterion level quality on the second "normalization" session. In addition to the significant session-to-session differences demonstrated in Fig. 4, the group suffering ablation of the prefrontal cortex demonstrated a significantly (p < 0.01) higher swim speed on the "distal cues only" session when compared to the second "normalization" session. DISCUSSION The outcome of this experiment clearly demonstrated that the prefrontal cortex contributes to the mediation of the currently studied, modified place learning task to such an extent that the task acquisition of prefrontally ablated rats is significantly impaired (Figs. 2,3). Although compensatory processes are able to prevent significant acquisition impairments of this task in rats subjected to pharmacological blockade of the cholinergic system [J. Mogensen, L. H. Christensen, A. Johansson, G. Wtirtwein, L. E. Bang, & S. Holm, submitted], the "recovery" processes upon prefrontal cortical ablations are less extensive: they are only able to allow the prefrontally ablated rats a normal proficiency of task performance after a significantly prolonged task acquisition. For the currently studied modification of place learning, prefrontal contributions to task mediation appear to be of relatively greater significance than the comparable contributions from cholinergic mechanisms [present data and J. Mogensen, L. H. Christensen, A. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted]. This is in marked contrast to the relative importance of prefrontal and nonprefrontal contributions to mediation of the "standard" place learning task. In the latter case the acquisition impairments provoked by hippocampal lesions [2,19,23,24] and cholinergic receptor blockade [25,26] are at least as significant as those associated with prefrontal cortical ablations [1,6-8,23]. In fact, in a study of the "standard" place learning task conducted in the water maze utilized in this study (but without the curtain and proximal cues of this study) we found the task acquisition of hippocampally lesioned rats to be significantly inferior to that of prefrontally ablated animals [J. Mogensen, K. T. Lauritsen, S. Elvertorp, A. Hasman, & G. Wrrtwein, in preparation], de Bruin et al. [1] even found the acquisition of the "standard" place learning task to be unimpaired by lesions of the prefrontal cortex. In contrast to the consequences of the prefrontal ablations, the task acquisition of the animals subjected to ablation of the parietal "association" cortex failed to reflect any lesion-associated impairment. During the second session of task acquisition the parietally lesioned animals performed at a significantly higher level of proficiency when compared to the sham operated control group. Apparently, restriction of distal cues and/or the addition of proximal cues reduces or eliminates the importance of the parietal contributions to mediation of place learning. It should be remembered that the size of the currently utilized water maze is such that an impairment of task acquisition would be expected if a "standard" place learning task had been administered to rats deprived of the parietal cortex [6,9]. The most likely explanation for the surprisingly superior task acquisition of the parietally lesioned rats seems to be that ablation of the parietal cortex decreased the likelihood that the animals would attempt to utilize

PREFRONTAL CORTEX AND MODIFIED PLACE LEARNING

431

Late sessions

Initial sessions 4400

4400' ~

XX

;r-.

~2200

2200

I

~$2S22z=:.

B (n

......

0 80

0

1

2 XXX

8O

i

i

CR-2

CR-I

CR-O

d-,

c.'-o

" ' CR-1

x R' C -0

C O Q

30

30

80

80

80

60

o I

,_E

o c.'-2

, xi

O O ¢,n

I E *"

40 I ,

v

1

Sessio,

40

' CR-2

Sesiion

FIG. 3. Performanceof the three experimentalgroups (T, the group subjectedto ablation of the parietal cortex; &, the group in which the prefrontal cortex had been ablated; and [2, the sham operated control group) during the initial and final sessions of the acquisition period. The initial three sessions as well as the criterion (CR-0) session with the preceding sessions (CR-1 and CR-2) are shown. Values are given as medians. **Significantly(p < 0.01) different from the sham group. *Significantly(p < 0.05) different from the sham group. (*)Significantly(p < 0.05, one-tailed) different from the sham group. ~xSignificantly(p < 0.001) different from the parietal group. ~xSignificantly(p < 0.01) different from the parietal group. a particular--under the current circumstances maladaptive--behavioural strategy. A pattern of results at least partly parallelling those of this experiment was obtained by Kesner et al. [5], who not only found lesions of the parietal cortex to be associated with impairment of allocentric orientation and prefrontal lesions to be associated with impaired egocentric orientation, but even observed facilitated allocentric orientation after prefrontal ablations. Kesner et al. [5] suggested that prefrontal lesions might improve aUocentric orientation by eliminating the competing egocentric "strategy." Although Kesner et al. [5] did not observe a facilitated egocentric orientation after lesions of the parietal cortex, it could be imagined that the parietal lesions of this experiment facilitated task acquisition because such lesions dimin-

ished the likelihood that allocentric orientation strategies would be selected. Such a speculation rests on the assumption that the task of this study differs from the "standard" place learning task by being more readily solved by egocentric than allocentfic orientation strategies. This assumption receives some support from the observation that lesions of the prefrontal cortex, which presumably is of greater importance to egocentric than allocentric orientation [5], are more potent in impairing task acquisition in the modified task than in the "standard" place learning task (present data). In this discussion we follow the arguments of Kesher et al. [5] by associating the prefrontal cortex with egocentric orientation and the parietal cortex with allocentric orientation. This viewpoint disagrees with the conclusions of Kolb et al. [8],

432

M O G E N S E N ET AL.

PC

4000

40

)< ,u

PC

~J ~p w

o

.2

2000

20 I

I

.~_

._s

J

01

0 Crit.

Rot.90

Rot.45

(;tit,

1.Norm. 2.Norm. Dilt.Only

PFC

1800

I

i

Rot.90

40

Rot.45

1.Norm. 2.Norm. Dist.Ofdy

PFC

"r. U

,x_ o

l= 0 II

9oo

20

,,.-,

I

._E

Ill

!1

0 Cri,t.

Rot.90

l Rot.45

jI

0

i

1.Norm, 2.Norm. DisI.Only

Crit.

Shim

6000

RoZ.90

so

Rot.45

1.Norm. 2.Norm. Dlst.Only

Sham

x o

c

~aooo

25

w

=I

E

I E

Ps I

i

Crit.

Rot.m0

Rot.45

l.Norm. 2.Norm. DiIt.Only

Session

Crit.

Rot.90

Rot.45

1.Norm. 2.Norm. Dist,Only

Se|$ion

FIG. 4. Performance of the three experimental groups (PC, the group in which the parietal cortex had been ablated; PFC, the group in which the prefrontal cortex had been removed; and Sham, the sham operated control group) on the criterion (Crit.), 90° rotation (Rot. 90), 45 ° rotation (Rot. 45), and "distal cues only" (Dist. Only) sessions as well as the two "normalization" sessions separating the "45 ° rotation" session from the "distal cues only" session (1. Norm. and 2. Norm.). Values are given as medians with ranges. **Significantly (p < 0.01) different from the performance of the same experimental group on the previous session. *Significantly (p < 0.05) different from the performance of the same experimental group on the previous session. (*)Significantly (p < 0.05, one-tailed) different from the performance of the same experimental group on the previous session.

who associate the parietal cortex with the use of " p r a x i s " strategies while believing the prefrontal cortex to be involved in more "allocentric" types of orientation. In spite of this lack of consensus, the studies of Kesner et al. [5] and Kolb et al. [8] agree on the conclusion that dissimilar behavioural strategies are mediated by the prefrontal and parietal cortical areas, respectively. Our present data lend further support to the conclusion that dissimilar contributions to the mediation of place learning are yielded by the prefrontal and parietal "association" cortices. Additional indications that the prefrontal cortex contributes to the mediation of egocentric orientation may be found in a recent

paper by Kolb et al. [6]. These authors report prefrontal lesions to cause a major impairment in a radial arm maze task requiring egocentric orientation. In the same study lesions of the parietal "association" cortex were found to have no effects on the " e g o centric" radial arm maze task. Reduction of distal cues appears to be an essential component in the current task modification [J. Mogensen, L. H. Christensen, A. Johansson, G. WOrtwein, L. E. Bang, & S. Holm, submitted], and apparently the task modifications provide a bias toward application of egocentric orientation strategies (see above). Consequently, it might be assumed that what is acquired by the an-

PREFRONTAL CORTEX AND MODIFIED PLACE LEARNING

imals is in fact a search strategy that does not rely upon the available distal cues. This possibility, however, is clearly contradicted by the performance of the rats of this study on the "rotation" sessions. The performance of all three experimental groups was significantly impaired by the 90° rotation of the distal cues (Fig. 4). As discussed in the Introduction, this indicates that all groups discriminated between the individual cues and utilized such a discrimination for navigational purposes. When, on the "distal cues only" session, the animals were deprived of the proximal cues, the prefrontally ablated group was the only one to demonstrate a significantly impaired task performance (Fig. 4). Although the prefrontally lesioned rats--in contrast to both the normal animals and the rats subjected to ablations of the parietal cortex--seem to have utilized the proximal cues for navigational purposes, it should be noted that the prefrontally lesioned group was at least as impaired as the two other groups by the 90° rotation of the distal cues. The prefrontal ablations appear not to have induced a shift from a dependence upon distal cues to a higher dependence upon proximal cues, but rather to be associated with a "hyper-dependence" upon navigational cues of a proximal as well as a distal nature. The major source of cortical afferent projections to the prefrontal cortex of the rat is the parietal "association" cortex [9,28]. Consequently, the prefrontal cortex of the parietally ablated group of this experiment was deprived of most of its direct, cortical input. The parietally ablated animals performed at least as well as normals, and a prefrontal cortical dysfunction would be expected to manifest itself in significantly impaired task acquisition (see the prefrontally ablated group of this experiment). On this background it must be concluded that the prefrontal cortex does not depend upon cortico-cortical afferents for its successful mediation of the currently studied, modified place learning task. Even in the spatial delayed alternation task--another "prefrontal" task--the prefrontal cortex appears to be able to mediate successful task performance in the absence of input from the parietal association cortex [28,29]. In conclusion, this study has demonstrated that the combination of restriction of distal cues and addition of proximal cues not only causes the normally selected solution strategy to differ from the "mapping" strategy of the "standard" task, but even drastically modifies the relative importance of the strategies mediated by the prefrontal and parietal "association" cortices. The strategy or strategies mediated by the parietal "association" cortex normally provides significant although limited contributions to the "mapping" strategy, but this strategy or these strategies are rendered irrelevant or even maladaptive by the current task modifications. The importance of the strategy or strategies mediated by the prefrontal cortex have, on the other hand, been drastically increased by restriction of distal cues and/or addition of proximal cues. From our previous studies it is known that in parallel to this emphasis on prefrontal task mediation not only the parietal contributions to task mediation but even the cholinergic [J. Mogensen, L. H. Christensen, A. Johansson, G. Wrrtwein, L. E. Bang, & S. Holm, submitted] and presumably hippocampal [30] contributions to task solution are significantly decreased. In a series of ongoing studies we address the question whether restriction of distal cues and/or addition of proximal cues in isolation can produce the currently observed pattern of parietal "deselection" and "hyperdependence" upon prefrontal task mediation. When the results of these studies are available, interpretation of the prefrontal contributions to the currently studied task should become somewhat easier. At the present moment only the analysis of the strategies demonstrated by the animals of this study is available. As mentioned in the Introduction, a pure "cue" learning strategy is prevented by the fact that no individ-

433

ual cue directly marks the position of the platform. Furthermore, the outcome of the "rotation" tests demonstrated that all three groups of animals discriminated between the individual distal cues and utilized such discriminations for navigational purposes. Consequently, neither the normal animals nor the two groups subjected to cortical ablations solved the task by a pure "praxis" strategy. Although only the performance of the prefrontal group appeared to be significantly affected by elimination of the proximal cues, it cannot necessarily be concluded that the animals of the two other groups did not utilize the proximal cues. During the two "normalization" sessions separating the "rotation" sessions from the "distal cues only" session, the performance of the animals had failed to normalize fully, and, therefore, effects of proximal cue elimination might not be easily seen. Furthermore, the informal observations of swim patterns during task solution (see the Results section) seem to indicate that all groups developed a spatial strategy that included visits to one or more of the proximal cues. It could be imagined that the normally preferred solution strategy of the task investigated in this study involves a sequence of relatively independent "navigational responses." An animal might, for instance, starting from an arbitrary release-position, conduct a series of "cue responses" during which it sequentially visits the proximal cues. Such a sequence of responses will for any release-position eventually be able to bring the animal to a particular proximal cue from which it might be able to approach the position of the platform rather precisely. If the normally preferred solution strategy involves such a sequence of relatively independent "behavioural subroutines" (the individual "navigational responses"), this might provide a potential explanation for the high degree of prefrontal task mediation. The prefrontal cortex seems to be crucially involved in the mediation of tasks in which behavioural "subroutines" are to be conducted in a particular sequence [12,14]. Such an interpretation emphasizes that although currently prefrontal cortical ablations were associated with a substantial impairment of the acquisition of a "spatial" task, such a result need not automatically imply that the prefrontal cortex mediates any aspect of a narrowly defined "spatial orientation." It does, however, emphasize that at least certain "spatial strategies" cannot easily be mediated in the absence of the prefrontal cortex [for recent discussions of the potential roles of the prefrontal cortex in spatial orientation: see 6]. ACKNOWLEDGEMENTS The authors are grateful for the financial support received from the Danish Medical Research Council, Direktor E. Danielsen og Hustrus Fond, DirektorJacob Madsenog hustru Olga MadsensFond, Eli og Egon Larsens Fond, Fonden af 1870, Fonden til Forskning af Sindslidelser, Grosserer Sonnich Olsen og Hustrus Legat (Fakuitetsfonden),Ivan Nielsens Fond, Kong Christian den Tiendes Fond, Laege Eilif Trier-Hansen og hustru Ane Trier-Hansens legat, The Novo-Nordisk Foundation, P. Carl Petersens Fond, Skizofrenifondenaf 1986, and Vera og Carl Johan Michaelsens Legat. The authors would also like to thank Ulla Mogensen for secretarial and technical assistance and both Karen and Jorgen Mogensen for technical help.

REFERENCES

I. de Bruin, J. P. C.; S~nchez-Santed,F.; Heinsbroek, R. P. W.; Donker, A.; Postmes, P. A behavioural analysis of rats with damage to the medial prefrontal cortex using the morris water maze: Evidence for behavioural flexibility but not for impaired spatial navigation. Brain Res. 652:323-333; 1994.

434

2. DiMattia, B. D.; Kesner, R. P. Spatial cognitive maps: Differential role of parietal cortex and hippocampal formation. Behav. Neurosci. 102:471-480; 1988. 3. Divac, I.; Mogensen, J.; Petrovic-Minic, B.; Zilles, K.; Regidor, J. Cortical projections of the thalamic mediodorsal nucleus in the rat. Definition of the prefrontal cortex. Acta Neurobiol. Exp. 53:425429; 1993. 4. Holm, S.; Mogensen, J. Contralateral somatosensory neglect in unrestrained rats after lesion of the parietal cortex of the left hemisphere. Acta Neurobiol. Exp. 53:569-576; 1993. 5. Kesner, R. P.; Farnsworth, G.; DiMattia, B. V. Double dissociation of egocentric and allocentric space following medial prefrontal and parietal cortex lesions in the rat. Behav. Neurosci. 103:956-961; 1989. 6. Kolb, B.; Buhrmann, K.; McDonald, R.; Sutherland, R. J. Dissociation of the medial prefrontal, posterior parietal, and posterior temporal cortex for spatial navigation and recognition memory in the rat. Cerebral Cortex 6:664-680; 1994. 7. Kolb, B.; Pittman, K.; Sutherland, R. J.; Whishaw, I. Q. Dissociation of the contributions of the prefrontal cortex and dorsomedial thalamic nucleus to spatially guided behavior in the rat. Behav. Brain Res. 6:365-378; 1982. 8. Koib, B.; Sutherland, R. J.; Whishaw, I. Q. A comparison of the contributions of the frontal and parietal association cortex to spatial localization in rats. Behav. Neurosci. 97:13-27; 1983. 9. Kolb, B.; Walkey, J. Behavioural and anatomical studies of the posterior parietal cortex in the rat. Behav. Brain Res. 23:127145; 1987. 10. Krechevsky, I. "Hypotheses" in rats. Psychol. Rev. 39:516-532; 1932. 11. Mogensen, J. Animal models in neuroscience. In: Svendsen, P.; Hau, J. eds., Handbook of laboratory animal science, Volume II, Animal models. New York: CRC Press, Inc.; 1994:125-136. 12. Mogensen, J.; Divac, I. Sequential behavior after modified prefrontal lesions in the rat. Physiol. Psychol. 12:41-44; 1984. 13. Mogensen, J.; Hasman, A.; W0rtwein, G. Place learning during inhibition of nitric oxide synthase in the rat. Homeostasis 36:12-18; 1995. 14. Mogensen, J.; Holm, S. The prefrontal cortex and variants of sequential behaviour: Indications of functional differentiation between subdivisions of the rat's prefrontal cortex. Behav. Brain Res. 63:89100; 1994. 15. Mogensen, J.; W6rtwein, G.; Gustafson, B.; Ermens, P. L-Nitroarginine reduces hippocampal mediation of place learning in the rat. Neurobiol. Learn. Memory, in press; 1995.

M O G E N S E N ET AL.

16. Mogensen, J.; W6rtwein, G.; Hasman, A.; Nielsen, P.; Wang, Q. Functional and neurochemical profile of place learning after L-nitroarginine in the rat. Neurobiol. Learn. Memory 63:54-65; 1995. 17. Morris, R. Developments of a water-maze procedure for studying spatial learning in the rat. J. Neurosci. Methods 11:47-60; 1984. 18. Morris, R. G. M. Toward a representational hypothesis of the role of hippocampal synaptic plasticity in spatial and other forms of learning. Cold Spring Harbor Symposia on Quantitative Biology 55:161-173; 1990. 19. Morris, R. G. M.; Hagan, J. J.; Rawlins, J. N. P. Ailocentric spatial learning by hippocampectomised rats: A further test of the "spatial mapping" and "working memory" theories of hippocampal function. Quarterly J. Exp. Psychol. 38 B:365-395; 1986. 20. O'Keefe, J.; Nadel, L. The hippocampus as a cognitive map. Oxford: Clarendon Press; 1978. 21. Paxinos, G.; Watson, C. The rat brain in stereotaxic coordinates. Sidney: Academic Press; 1982. 22. Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill; 1956. 23. Sutherland, R. J.; Kolb, B.; Whishaw, I. Q. Spatial mapping: Definitive disruption by hippocampal or medial frontal cortical damage in the rat. Neurosci. Lett. 31:271-276; 1982. 24. Sutherland, R. J.; Whishaw, I. Q.; Kolb, B. A behavioural analysis of spatial localization following electrolytic, kainate- or colchicineinduced damage to the hippocampal formation in the rat. Behav. Brain Res. 7:133-153; 1983. 25. Sutherland, R. J.; Whishaw, I. Q.; Regehr, J. C. Cholinergic receptor blockade impairs spatial localization by use of distal cues in the rat. J. Comp. Physiol. Psychol. 96:563-573; 1982. 26. Whishaw, I. Q. Cholinergic receptor blockade in the rat impairs locale but not taxon strategies for place navigation in a swimming pool. Behav. Neurosci. 99:979-1005; 1985. 27. Whishaw, I. Q.; Mittleman, G. Visits to starts, routes, and places by rats (Rattus norvegicus) in swimming pool navigation tasks. J. Comp. Psychol. 100:422-431; 1986. 28. W6rtwein, G.; Mogensen, J.; Divac, I. Retention and relearning of spatial delayed alternation in rats after combined or sequential lesions of the prefrontal and parietal cortex. Acta Neurobiol. Exp. 53:357-366; 1993. 29. WiSrtwein, G.; Mogensen, J.; Divac, I. Retention and relearning of spatial delayed alternation in rats after ablation of the prefrontal or total nonprefrontal isocortex. Behav. Brain Res. 63:127-131; 1994. 30. W6rtwein, G.; Smrup, L. H.; Charlottenfeld-Starpov, D.; Mogensen, J. Place learning by fimbria-fornix transected rats in a modified water maze. Int. J. Neurosci., 82:71-81; 1995.