- Email: [email protected]
Contribution of egocentric spatial memory to place navigation of rats in the Morris water maze
BEHAVIOURAL BRAIN RESEARCH
ELSEVIER
Behavioural Brain Research 78 (1996) 121-129
Research Report
Contribution of egocentric spatial memory to place navigation of rats in the Morris water maze Mojdeh Moghaddam, Jan Bures * Institute of Physiology, Academy of Sciences, Videnska 1083, P14220 Prague 4 Krc, Czech Republic Received 18 August 1995; revised 6 November 1995; accepted 6 November 1995
Abstract
Place navigation in the Morris water maze can be directed by memory of the target coordinates relative to remote landmarks (allocentric) or by the memory of the start-goal route (egocentric). When the start and goal positions remain constant and visual cues are eliminated by darkness, memory of the route may become decisive. This assumption was tested in 10 male hooded rats using an infrared television tracking system allowing navigation training in the dark. In Expt. 1, these animals were trained to swim in the dark from the start at the S rim of the pool to the goal position in the center of the NW quadrant of the pool. Mean escape latencies decreased from 47 s initially to 16 s during the 24 daily sessions. Another group of 10 male hooded rats learned the same task in the light. Mean escape latencies decreased from 20 s initially to 5 s during 4 daily sessions. In Expt. 2, possible allocentric location of the target was tested in the same rats by rotating both the start and goal positions by 90 ° counterclockwise (i.e., to E-SW and later to N-SE). Mean escape latency during 5 days after the first rotation increased to 24 s, but returned back to the asymptotic level of 18 s after the second rotation. The same change of the start and goal position (from S-NW to E-SW) in the light only increased escape latency in the first session. In Expt. 3, both the goal position and route direction were changed to N-SW. Surprisingly, the animals rapidly acquired a new heading angle at the start and mean escape latencies were not significantly changed. It is concluded that overtrained place navigation in darkness can be easily changed to a new direction. Keywords: Spatial memory; Path integration; Place navigation; Orientation in darkness; Allocentric map; Egocentric map; Rat
1. Introduction
Different explanations were proposed for the spatial abilities of rats in the Morris water maze [1,15,16]. Cognitive m a p theory [-19], computations based on local views [-11] also called 'snapshots' [2,4,31] and relational processes [-5] have been suggested to explain the rat's ability to navigate from any part of the pool to the invisible goal. The emphasis of the underlying research has been on allocentric navigation implemented by the m e m o r y of the coordinates of the target relative to remote landmarks. The allocentric coding system is based on spatial relationships between external landmarks that are independent of the location and position of the subject and lead to the coding of allocentric space within a spatial map. For instance, the hidden platform is north or 5 m * Corresponding author. Fax: (42) (2) 471-9517; E-mail: bures@sunl .biomed.cas.cz 0166-4328/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved PH S01 66-4328 (95) 00240-5
from the window and east or 3 m from the door. Successful place navigation can also be implemented, however, by an egocentric coding system which deals with spatial relationships between the subject and the goal and leads to coding of relative space. In this system, the rat turned with its back to the wall of the pool learns that the goal is 30 ° to the left and 100 cm distant. Close interaction of the two systems has been stressed by some theoretical [8], behavioral [29] and clinical [23] studies, For instance, when a rat learns to swim in a certain direction from a given starting point, egocentric coding can be responsible for this behavior, whereas repeated checking of the accuracy of such navigation must be done by an allocentric system. Association between the visual allocentric information and the kinesthetic egocentric information is an essential prerequisite for efficient coding of space [ 13 ]. Interpretation of visual signals in the three dimensional world must be supported by other sensory inputs determining the position of the organism with respect
122
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129
to the gravitational field and differentiating self-motion from the movement of the surroundings. Particularly important in this respect is the vestibular system which is essential for the concepts of horizontality and verticality as well as for detecting acceleration or deceleration of the body [25]. The theoretical role of vestibular signals is only rarely supported by experimental analysis of their relative contribution to a specific behavioral situation. Place navigation in the Morris water tank [15-17,26] represents a complex task requiring good orientation in space. According to O'Keefe [18], the cognitive map theory predicts that the animal can anticipate landmark configurations to be seen from any part of the charted environment, and that, even in absence of visual information, it will know its position relative to the starting point by path integration [7,14]. The latter mechanism reflects the capacity of a rat to integrate directional and distance information received during active or passive movement and to translate it into continuously updated azimuthal information about the direction to the starting point [7]. The importance of vestibular cues in this mechanism is demonstrated by the capacity of blind rats to display spontaneous alternation [22] and to solve radial maze tasks [32]. It is conceivable that path integration automatically monitors the position of the rat on the cognitive map between points whose coordinates can be established by observation of landmarks [ 12]. The two systems may also differ by their neural substrate. Whereas the hippocampus seems to be indispensable for allocentric navigation, egocentric navigation may require more expressed participation of cerebral cortex and basal ganglia (see Poucet [21] and Wiener and Berthoz [-30] for reviews). The aim of the present study was to assess the importance of egocentric coding for the acquisition and retrieval of the water maze task under conditions when the start-goal position was constant and the allocentric orientation was eliminated, or at least severely restricted, by total darkness. Cognitive maps increase the flexibility and efficacy of spatial behavior by allowing the animal to rapidly solve newly encountered spatial problems. Whereas this assertion is well supported by vast literature on allocentric orientation [-8] egocentric navigation over fixed start-target routes is sometimes explained as a motor skill or praxis [29]. Since motor skills are usually learned incrementally and cannot be easily reversed or otherwise modified, transfer of egocentric navigation to a different start-target route was used to test the cognitive map versus motor skill explanations of blind navigation.
animals weighed between 250 and 300 g at the start of the experiments and were housed 5 per cage in an animal room with a constant temperature (20°C) and natural lighting. Food and water were always freely available.
2.2. Apparatus The Morris water maze consisted of a uniformly blue circular pool (140 cm in diameter and 55 cm in height) filled to a depth of 25 cm with water (20°C). An ll-cmdiameter clear plexiglass platform stood 1 cm below the water surface in one of the 4 arbitrarily designated quadrants, 35 cm from the center of the pool. An infrared LED connected to a counterbalanced light cable was attached to the animal's body with a rubber band. A infrared-sensitive television (TV) camera was mounted over the center of the pool and used for monitoring the LED movements. The TV signal was digitized and fed into the computerized tracking system which monitored and stored the x - y coordinates of the animal's trajectory at 100-ms intervals. The swim path of each trial and the escape latency as well as the total length of the swim path and the time spent in any designated area of the pool was recorded.
2.3. Behavioral training Rats always navigated in complete darkness. A trial began by turning the light off, taking the animal from the waiting cage and placing it into the water while facing the wall of the pool. The tracking system was started at the time the animal was released. The maximum time the rat was allowed to swim, was 60 s and a dim light was switched on automatically either after this time had expired or after the animal had found and remained on the platform for more than 1 s. When the rat failed to find the platform it was led to it. Climbing onto the platform was the only way of escaping from the water. The animal was always allowed to rest there for 15 s before it was returned to a waiting cage. The light was switched off and the next trial was started approximately 30 s later. There were 8 trials per session. After the last trial, the animal was towel dried and put into another waiting cage under a heating lamp before being returned into the home cage.
2.4. Statistical analysis 2. General method
2.1. Animals Twenty male hooded rats of the Long-Evans strain were obtained from the Institute's breeding colony. The
Analysis of variance (ANOVA) with repeated measures followed by Tukey's test for multiple comparisons was used. Whenever necessary a paired t-test was used for within-group comparisons. The P<0.05 level was accepted as significant.
123
Mojdeh Moghaddam, Jan Bures/Behavioural Brain Research 78 (1996) 121-129
3. Experiment 1 Spatial orientation of animals in nature is implemented by two mechanisms: (1) the allocentric coding system based on the memory of the target coordinates relative to remote extra maze landmarks which leads to the coding of absolute space within a spatial map; and (2) the egocentric coding system based on the memory of the start position and of the start-goal route which leads to coding of relative space. These two systems are tightly inter-related under normal conditions: both are used simultaneously and both contribute to correct performance of the animal. In Expt. 1, we attempted to isolate the egocentric component of navigation between fixed start and goal positions by eliminating the allocentric one by conducting the experiment in total darkness. The importance of the allocentric component was assessed in a control group trained under the same conditions in the light. 3.1. Method
The procedure described in General method was followed. Naive rats (n= 10) were trained in 24 daily sessions (8 trials per session) to swim in complete darkness from the start in the south (S) to the goal platform in the center of the north-west (NW) quadrant of the pool. The combined allocentric and egocentric orientation was tested in another group of 10 animals trained for 4 days (32 trials) with the same start and goal position ( S - N W ) in the light. The same apparatus was used in all the phases of the experiment.
4. Results and discussion The control group trained in the light rapidly learned to search the platform in the inner part of the pool and reached asymptotic performance of 5 s on day 3. The rats trained in the dark persisted longer in attempts to climb the wall before they started random search all over the surface of the pool and gradually oriented toward the target. The mean escape latencies decreased more slowly and reached the asymptotic level of 17 s only in the fourth week. Fig. 1 compares mean escape latencies during the first 4 days of training in the light and in the dark. A two-way ANOVA (groups × days) with repeated measures on the last factor showed significant main effect of groups (F(1,18)=230.18, P<0.001) and days (F(3,54)= 15.24, P <0.001), but no significant interaction (F(3,54)=2.46, n.s.). A paired t-test showed a significant decrease of mean escape latencies in the group trained in the light between days 1 and 2 ( t ( 9 ) = 3.63, P < 0 . 0 1 ) and days 2 and 3 (t(9)=2.89, P<0.01), but not between days 3 and 4 (t(9)=0.32, n.s.) when performance became asymptotic. Training in the dark
- -
Darkness
.....
Light
eo F
4o ti "~
30
Li
t'--.....
2oi
'IF
.....
1
2
, 3
4
days
Fig. 1. Acquisition of the place navigation task under conditions of fixed start (S) and goal (NW) in rats trained in the light (dashed line) or in the dark (full line). Ordinate: mean (+ S.E.M.) escape latencies of 8 daily trials. Abscissa: days of training. Note that the performance improved to asymptoticlevel on day 3 in the light, but that remained almost unchanged in the dark. progressed much more slowly and did not yield a significant reduction of mean escape latencies between successive days. It was, therefore, continued for 5 weeks (5 days per week) until an asymptotic performance was reached. Fig. 2 shows the mean escape latencies on the last 3 days of every week of training. Paired t-tests revealed significant differences between the first and second week (t(9)=5.69, P<0.001), second and third week (t (9) = 3.65, P < 0.01 ), third and fourth week ( t (9) = 4.01, P < 0.01 ), but not between the fourth and fifth week (t(9)=0.15, n.s.). Acquisition of the egocentric navigation is better illustrated by the initial orientation of swimming trajectories than by the efficiency of the search after the rat has failed to locate the goal on the first approach. Fig. 3 shows examples of trajectories generated during the first 5 s after start by a typical rat on the last days of each week. No clear preferred direction is seen in the first week, but a prevailing orientation to the left appears in 60
5O
4O
oo
3O
~ 2o 10
1
3
4
5
woek8
Fig. 2. Acquisition of the place navigation task under conditions of the fixed start (S)-fixedgoal (NW) positions in the dark. The columns represent the mean (+ S.E.M.) escape latency during the last 3 days of every weekly block of 5 days. Note that the escape latencies reached the asymptotic level of 16 s in the fourth week.
Mojdeh Moghaddam, Jan Bures / Behavioural Brain Research 78 (1996) 121-129
124 A
B
was change of the starting point and goal position from S-NW to E-SW. In the animals trained in the dark, a second change from E-SW to N-SE followed after 5 days. The 10 rats trained in the light in Expt. 1 were used in Expt. 2 to check their performance after changing the start and goal position from S-NW to E-SW. 5.2. Results and discussion
C
D
Fig. 3. Computer printouts of the superimposed swimming trajectories in a typical rat on trained in darkness on days 5 (A), 10 (B), 15 (C) and 20 (D). The crossed lines correspond to the WE and NS diameters of the pool. The start mark is in S and the position of the goal is indicated by the double circle in the NW quadrant. Note that the widely dispersed trajectories on day 5 gradually contract to the compact form seen on day 20.
the second week and the trajectories become more compact in the third week and particularly in the fourth week. Note, however, that even at this stage of learning the platform is missed on some of the swims and the ensuing search accounts for the relatively long mean latency. The results demonstrate that acquisition of egocentric navigation (fixed start, fixed goal) in the dark is slower and less efficient than allocentric navigation in the light. Asymptotic performance is reached after 4 weeks in the dark and 4 days in the light with the corresponding escape latencies of 16 and 5 s as shown in Fig. 1 and Fig. 2, respectively.
The last day of training on the first task (S-NW) was compared with the first 2 days of the E-SW training in the dark and in the light. Average escape latencies shown in Fig. 4 were analyzed with a two-way ANOVA (groups x conditions) with repeated measures on the second factor which yielded significant main effects of groups (F(1,18)=48.72, P<0.001) and conditions (F(2,36)=8.50, P<0.001) as well as significant interaction (F(2,36)=4.30, P<0.05). A paired t-test showed a significant increase of escape latencies on the first day of E-SW training both in the dark (t(9)=3.2, P<0.01) and in the light (t(9)=4.2, P<0.01). The difference also remained significant on the second day in the dark (t(9)=3.27, P<0.01), but not in the light (t(9)=0.49, n.s.). E-SW training in the dark continued for 3 more days during which the escape latency decreased to 24.2+2.8 s. A new start-goal rotation from E-SW to N-SE increased the escape latency to 29.4+2.7 s. A paired t-test showed that this difference is not significant (t(9)= 1.27, n.s.). During a further 4 days of training, the escape latency dropped to 18.0+2.0 s on the last day, i.e., to a level not significantly different from the last week of S-NW training. The effect of the start-goal rotation can also be illustrated by the 5-s trajectories shown for a typical rat in Fig. 5. Although the rotation should not have affected egocentric navigation, its accuracy was considerably Darkness
~-~ Light
40
5. Experiment 2 3O
Navigation learned in the dark in Expt. 1 could have been due to allocentric memory for non-visual extramaze cues (acoustic beacons) or to egocentric route memory. The purpose of Expt. 2 was to test the relative significance of these possibilities by rotating the start and goal position by 90 ° counterclockwise (i.e., from S-NW to E-SW and to N-SE). Such rotation should leave egocentric memory unaffected, but would require relearning of allocentric navigation. 5.1. Method
The same overtrained rats and the same apparatus were used in the second experiment. The_only difference
20
S-NW
E-$W
E-SW
Fig. 4. Navigation performance of rats trained in the light (empty columns) or the dark (black columns) on the last day before and on the first 2 days after rotation of the start and goal positions by 90 ° counterclockwise (i.e., from S-NW to E-SW). Other descriptions as in Fig. 1. Note that the performance is back to the asymptotic level on the second day after change in the light, but is only slightly improved in the dark.
Mojdeh Moghaddam, Jan Bure~/BehaviouralBrain Research 78 (1996) 121 129
A
B
D
E
125
Fig. 5. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat trained in darkness on the last day of S-NW training (A), on the first (B) and fourth (C) day of the of E-SW training and on the first (D) and fourth (E) day of the NSE training. Note that the swimming trajectories became more dispersed on the first days after each change, but assumed, after another 3 days, the compactness seen after 3 4 weeks of training in Fig. 3. decreased in the first session after change with a tendency to visit the allocentric location of the previous goal. This is particularly clear in Fig. 5D where most swims pass through former position of the goals in the N W and SW quadrants of the pool. This type of error was, however, rapidly corrected and the animals returned to the somewhat modified start-target route. The above results suggest that the navigation performance at the end of the initial 5-week period of S - N W training was at least partly due to allocentric memory for the position of the goal with respect to acoustic beacons. It seems, however, that the significance of the allocentric component of navigation decreased after several changes of the position of the goal reduced the reliability of the allocentric, but not of the egocentric, orientation.
6. Experiment 3 When the start remains the same (N), but goal location is changed, both egocentric and allocentric navigation should be modified. It was expected that such double change is more difficult than the change of the allocentric component only. 6.1. Method
The same method of training was applied in this experiment, performed only in the 10 rats previously trained in the dark. The animals were started from N, but the goal was moved to SW. The rats were trained in total darkness under these conditions for 5 days.
6.2. Results and discussion
The results of the combined allocentric and egocentric change of the goal position are illustrated in Fig. 6. No deterioration of navigation performance was seen on the first day of the change and mean escape latency was 15.7__+1.4 s in the last 3 days of training. Quadrant preference already shifted on the first day from SE to SW and the overall time spent in the western half of the pool increased from 29% before to 60% after the change. Closer analysis of the first 4 trials of the sessions preceding and following the change indicated (Fig. 7) that the preference already appeared during the second trial and remained well expressed afterwards. The rapid change of the trajectory is shown in Fig. 8B: only the first two swims resembled attempts to find the goal in the area corresponding to its previous allocentric and egocentric location (Fig. 8A). The remaining trajectories of the first session resembled those generated at the asymptotic level of performance (Fig. 8C). Essentially similar results were obtained when, in a subsequent series of experiments, the same rats were started from the North, but the goal was moved to a point 20 cm from the South wall of the pool, i.e., to a location never used before. Fig. 9 shows typical trajectories of another rat before (A) and after the change (B). The first two swims were directed to previous position of the goal in SW, but the subsequent tracks were already aimed at the new goal which was, however, not reached during the 5-s interval displayed. This modification developed during the next 2 days into a faster new route perpendicular to the circumference of the pool (Fig. 9C).
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129
126
A
B
sec
20-
15-
10-
5Q. SW 0-
N-SE
N-SW
N-SE
N-SW
Fig. 6. Navigation performance of rats trained in the dark on the last day before (N-SE) and the first day after (N-SW) the change of the goal position. A: mean (_S.E.M.) escape latency. B: mean (±S.E.M.) percentage of the trial time spent in the 4 quadrants of the pool. Note significant preference against the random choice level (25%) for the goal quadrants before (39.8%, t(9)= 8.7, P < 0.001 for SE) and after the change (40.3%, t(9)=5.3, P<0.001 for SW).
7. General discussion
3C
]
2(~
[]
NWQ
IC
]
SWQ
]
SEQ
0
Trial 1
Trial 2
Trial 3
The main result of the present study is the finding that place navigation learning proceeds more slowly in the dark than in the light, even under conditions allowing egocentric solution of the task, but that overlearned egocentric navigation can be re-directed to new targets with the same ease as the allocentric navigation. This suggests that exploration of the egocentric space leads to formation of an egocentric cognitive map which can be used for goal-directed navigation with similar flexibility as the allocentric maps. Such a view is compatible with the assumption that the allocentric and egocentric maps represent the same space [8], but contrasts with the route learning explanation of the locomotion between fixed start-target points [-29] which assumes that the sequence of movements which carries the animal from the start to the goal is a special case of instrumental learning associating emitted behavior with reinforcement. No map is needed in this case but when the start-goal relationship is changed (as in Expt. 3) a new route has to be learned with almost the same effort as the first one. The present results are relevant in the
NEQ
Trial 4
Fig. 7. Mean (___S.E.M.) percentage of the first 10 s of the trial time spent in the 4 quadrants of the pool in the first 4 trials of the last session before (N-SE, above) and in the first session after (N-SW, below) the change of the goal position. Quadrant labeling is the same as in Fig. 6. Note that the rat preferred the East half of the pool before the change, that the preference for the new goal quadrant became significant since trial 2 (46.5%, t(9)=3.6, P<0.01 for SW) and that the preference for the NW quadrant only started to appear over trials 2 4.
A
B
C
i
Fig. 8. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat during the last day before (A) and the first day after the change of the goal location from SE to NW (B). Note that swims 1 and 2 were directed to the old position of the goal, but that swims 3-8 were already directed to the new goal place. The new route became fully effective after 2 more days of training (C).
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129 A
B
127
C
i
Fig. 9. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat during the last day before (A) and the first day (B) and third day (C) after the change of the goal location from SW to S. Note that the first two swims after the change were directed to the previous position of the goal, but that further swims gradually shifted to the South and developedinto a new route 2 days later. context of the more than 50-year-old dispute between Guthrie [9], an ardent proponent of the route explanation, and Tolman 1-27] who pioneered the cognitive map concept. They support the view that exploration in darkness (i.e., in egocentric space) does not produce individual start-target routes, but a cognitive map of the space entered from the start area which allows generation of a multitude of routes leading from the start to any point within this space. A better known version of flexible navigation of this type is homing [6,7,14] which allows the animal to return from any point reached on its outer journey directly back to the start.
7.1. Navigation in absence of visual cues While navigation to the hidden platform is usually guided by extramaze landmarks visible from the pool, the goal can be even found in the absence of such landmarks, e.g., in a pool surrounded by black curtains [26], in darkness [10] or by blind rats [3]. Two situations must be distinguished: (1) the target is in a fixed location and can, therefore, be located allocentrically by non-visual extramaze cues (e.g., acoustic beacons); or (2) the target is in one of several possible positions and can be found by a search strategy generating a route passing through all these locations, e.g., a circular route at a constant distance from the wall. Although this distance cannot be directly estimated in the absence of vision, tactile, kinesthetic and vestibular signals could guide the rat into a trajectory satisfying the above requirement. Buresova et al. [3] found, in highly overtrained blind rats, mean escape latencies of 41 and 33 s when using the randomly located goals and fixed goals, respectively. Liu et al. [10] trained rats released from starting points in N, E, S, or W to navigate to a hidden platform in the center of one quadrant of the pool. During 3 days of training in the dark, mean escape latencies decreased from 43 to 39 s, i.e., similar to that in Expt. 1. This indicates that the possible egocentric solution of the task cannot be applied immediately, but only after exploration of the pool in the dark has formed the egocentric cognitive map of the pool. On the other hand, egocentric navigation to a new target
can proceed rapidly when the egocentric cognitive map is already available, as in Expt. 3.
7.2. Dissociation of allocentric and egocentric components of place navigation Continued training of rats in the fixed start-fixed goal navigation task allows the rats to find the goal by combined use of the vestibular and kinesthetic cues in the egocentric space and of non-visual cues available in the allocentric space. The performance reached at the end of Expt. 1 was obviously due to the additive effect of these two factors. Simultaneous rotation of the start and goal positions dissociated these two components: whereas the egocentric component remained unaffected, the allocentric was now pointing in a different direction. The conflict between these two components accounted for an increase of escape latencies from 13 s to 31 s. Assuming that the allocentric component was added to the egocentric one before the change and subtracted from it after the change, the egocentric component alone should be between 13 and 31 s, i.e., 22 s. The next start-goal rotation elicited a less expressed prolongation of escape latencies which gradually approached the egocentric component. Purely egocentric navigation can also be obtained when allocentric orientation is prevented by everyday change of the goal position. This technique used by Save and Moghaddam [24] yielded, after 24 days of training, mean escape latencies of 27 s, uncontaminated by aUocentric components.
7.3. Flexibility of egocentric cognitive maps The difference between acquisition of the same navigation task in the light and in the dark is mainly due to the fact that the egocentric map is formed more slowly than the allocentric map of the same environment. Once the egocentric map is formed, however, it can be used for solving the navigation tasks with the same flexibility as the allocentric map. This is well documented for the working memory version of the place navigation task 1-15,20,28] in which the animal is exposed to a new position of the escape platform on the first swim and is able to find the platform much faster on the second
128
Mojdeh Moghaddam, Jan Bures/Behavioural Brain Research 78 (1996) 121 129
swim performed several tens of minutes or several hours later. When the goal position is changed and the animal does not find it in the expected place, the failure activates the search routine. After the goal is found, its egocentric coordinates are entered into the egocentric map and can support navigation to this place from the start. Since, on each trial, the animal either found the escape platform or was guided to it, the rat had all the information needed for entering the goal position into his egocentric cognitive map for the particular start and for plotting the putative start-goal direction by dead reckoning. The above explanation must be accepted with caution, however. Since the experimental conditions did not completely eliminate the use of non-visual allocentric cues (Expt. 2), allocentric mechanisms might have contributed to the rapid change of the route in Expt. 3. This possibility is enhanced by the fact that the new goal position in Expt. 3 corresponds to the location (SW) already used in the initial phase of Expt. 2 which may still be remembered by the animals on the basis of allocentric cues. Such an explanation is improbable, however, because allocentric orientation alone does not support efficient navigation in the dark [3,10]. Furthermore, the rats demonstrated similar flexibility when learning a new route to a goal position never used in the previous training (e.g., from North to South, Fig. 9). To rule out the contribution of allocentric orientation in the above situation, allocentric cues should be eliminated by an experimental design [24] in which the start-goal geometry with respect to the circular pool remains stable, but the allocentric position of the starting point changes randomly from trial to trial. In this case, allocentric coordinates of the goal become irrelevant and optimum solution of the task is only possible with the use of the egocentric map. Further research is needed to verify the postulated flexibility of egocentric mapping under the above conditions.
Acknowledgement The authors thank Mr. programming and Mr. A. tance. This research was AVCR 711401 and BMFT
Yu. Kaminsky for computer Zahalka for technical assissupported by Grants IGA 01VJ 920015/26-5a.
References [ 1] Brandeis, R., Brandys, Y. and Yehuda, S., The use of the Morris water maze in the study of memory and learning, Int. J. Neurosci., 48 (1989) 29-69. [2] Bures, J. and Buresova, O., Spatial memory in animals. In: R.E.
John (Ed.), Machinery of Mind, Birkhauser, New York, 1990, pp. 291-310. [3] Buresova, O., Homuta, L., Krekule, I. and Bures, J., Does nondirectional signalization of target distance contribute to navigation in the Morris water maze?, Behav. Neural Biol., 49 (1988) 240-248. I-4] Cartwright, B.A. and Collett, T.S., How honey bees use landmarks to guide their return to a food source, Nature, 295 (1982) 560 564. 1-5] Eichenbaum, H., Otto, T. and Cohen, N.J. The hippocampus what does it do?, Behav. Neural Biol., 57 (1992) 2-36. [-6] Etienne, A.S., Maurer, R. and Saucy, F., Limitations in the assessment of path dependent information, Behaviour, 106 (1988) 81 111. 1-7] Etienne, A.S., Maurer, R., Saucy, F. and Teroni, E., Short-distance homing in the golden hamster after a passive outward journey, Animal Behav., 34 (1986) 696-715. 1-8] Gallistel, C.R., The Organization of Learning, MIT Press, Cambridge, 1990. 1-9] Guthrie, E.R., Tolman on associative learning, Psychol. Rev., 44 (1937) 525 529. [10] Liu, Z., Francis Turner, L. and Bures, J., Impairment of place navigation of rats in the Morris water maze by intermittent light is inversely related to the duration of the flash, Neurosci. Lett., 180 (1994) 59 62. 1-11] McNaughton, B.L., Neural mechanisms for spatial computation and information storage. In: L. Nadel, L.A. Cooper, P. Culicover and R.M. Harnish (Eds.), Neural Connections, Mental Computations, MIT Press, Cambridge, MA, 1989, pp. 285-350. [12] McNaughton, B.L., Chen, L.I. and Markus, E.J., Dead reckoning, landmark learning and the sense of direction: a neurophysiological and computational hypothesis, J. Cognitive Neurosci., 3 (1991) 190 202. [13] McNaughton, B.L., Leonard, B. and Chert, L., Cortical-hippocampal interactions of cognitive mapping: an hypothesis based on reintegration of the parietal and inferotemporal pathways for visual processing, Psychobiology, 17 (1989) 230-235. [14] Mittelstaedt, H. and Mittelstaedt, M.-L., Homing by path integration. In: F. Papi and H.G. Wallraff (Eds.), Avian Navigation, Springer Verlag, Berlin, 1982, pp. 290-297. [15] Morris, R.G.M., Spatial localization does not require the presence of local cues, Learning Motiv., 12 (1981) 239-261. [16] Morris, R.G.M., An attempt to dissociate spatial mapping and working memory theories of hippocampal function. In: W. Seifert (Ed.), Neurobiology of the Hippocampus, Academic Press, New York, 1983, pp. 405-432. [17] Morris, R.G.M., Developments of a water-maze procedure for studying spatial learning in the rat, J. Neurosci. Methods, 11 (1984) 47-60. 1-18] O'Keefe, J., A computational theory of the hippocampal cognitive map. In: J. Storm-Mathisen, J. Zimmer and O.P. Ottersen (Eds.), Understanding the Brain through the Hippocampus, Progress in Brain Research, Vol. 83, Elsevier, Amsterdam, 1990, pp. 301 312. [19] O'Keefe, J. and Nadel, L., The Hippocampus as a Cognitive Map, Clarendon, London, 1978. [20] Panakhova, E., Buresova, O. and Bures, J., Persistence of spatial memory in the Morris water tank task, Int. J. Psyehophysiol., 2 (1984) 5-10. [21] Poucet, B., Spatial cognitive maps in animals: New hypotheses on their structure and neural mechanisms, Psychol. Rev., 100 (1993) 163-182. [22] Potegal, M., Day, M.J. and Abraham, L., Maze orientation, visual and vestibular cues in two-maze spontaneous alternation of rats, Physiol. Psychol., 5 (1977)414 420. [23] Ratcliff, G., Brain and space: some deductions from the clinical evidence. In: J. Paillard (Ed.), Brain and Space, Oxford University Press, Oxford, 1991. [24] Save, E. and Moghaddam, M., Effects of lesion of the associative parietal cortex on the acquisition and use of spatial memory in
Mojdeh Moghaddam, Jan Bures/ Behavioural Bra& Research 78 (1996) 121-129
[25] [26] [27] [28]
[29]
egocentric and allocentric navigation tasks in the rat, Behav. Neurosci., in press. Stoffregen, T.A. and Riccio, G.E., An ecological theory of orientation and the vestibular system, Psychol. Rev., 95 (1988) 3 14. Sutherland, R.J. and Dyck, R.H., Place navigation by rats in a swimming pool, Canad. J. Psychol., 38 (1984) 322-347. Tolman, E.C., Determiners of behavior at a choice point, Psychol. Rev., 45 (1938) 1 41. Whishaw, I.Q., Formation of a place-learning set by the rat: a new paradigm for neurobehavioral studies, Physiol. Behav., 35 (1985) 139-143. Whishaw, I.Q. and Mittleman, G., Visits to starts, routes, and
129
places by rats (Rattus norvegicus) in swimming pool navigation tasks, J. Comp. Psychol., 100 (1986) 422 -431. [30] Wiener, S. and Berthoz, A., Forebrain structures mediating the vestibular contribution during navigation. In: A. Berthoz (Ed.), Multisensory Control of Movement, Oxford University Press, Oxford, 1993, pp. 427 456. [31] Wilkie, D.M. and Palfrey, R., A computer simulation model of rat's place navigation in the Morris water maze, Behav. Res. Methods, Instr. Comput., 19 (1987) 400-403. [32] Zoladek, L. and Roberts, W.A., The sensory basis of spatial memory in the rat, Anita. Learning Behav., 6 (1978) 77 81.
ELSEVIER
Behavioural Brain Research 78 (1996) 121-129
Research Report
Contribution of egocentric spatial memory to place navigation of rats in the Morris water maze Mojdeh Moghaddam, Jan Bures * Institute of Physiology, Academy of Sciences, Videnska 1083, P14220 Prague 4 Krc, Czech Republic Received 18 August 1995; revised 6 November 1995; accepted 6 November 1995
Abstract
Place navigation in the Morris water maze can be directed by memory of the target coordinates relative to remote landmarks (allocentric) or by the memory of the start-goal route (egocentric). When the start and goal positions remain constant and visual cues are eliminated by darkness, memory of the route may become decisive. This assumption was tested in 10 male hooded rats using an infrared television tracking system allowing navigation training in the dark. In Expt. 1, these animals were trained to swim in the dark from the start at the S rim of the pool to the goal position in the center of the NW quadrant of the pool. Mean escape latencies decreased from 47 s initially to 16 s during the 24 daily sessions. Another group of 10 male hooded rats learned the same task in the light. Mean escape latencies decreased from 20 s initially to 5 s during 4 daily sessions. In Expt. 2, possible allocentric location of the target was tested in the same rats by rotating both the start and goal positions by 90 ° counterclockwise (i.e., to E-SW and later to N-SE). Mean escape latency during 5 days after the first rotation increased to 24 s, but returned back to the asymptotic level of 18 s after the second rotation. The same change of the start and goal position (from S-NW to E-SW) in the light only increased escape latency in the first session. In Expt. 3, both the goal position and route direction were changed to N-SW. Surprisingly, the animals rapidly acquired a new heading angle at the start and mean escape latencies were not significantly changed. It is concluded that overtrained place navigation in darkness can be easily changed to a new direction. Keywords: Spatial memory; Path integration; Place navigation; Orientation in darkness; Allocentric map; Egocentric map; Rat
1. Introduction
Different explanations were proposed for the spatial abilities of rats in the Morris water maze [1,15,16]. Cognitive m a p theory [-19], computations based on local views [-11] also called 'snapshots' [2,4,31] and relational processes [-5] have been suggested to explain the rat's ability to navigate from any part of the pool to the invisible goal. The emphasis of the underlying research has been on allocentric navigation implemented by the m e m o r y of the coordinates of the target relative to remote landmarks. The allocentric coding system is based on spatial relationships between external landmarks that are independent of the location and position of the subject and lead to the coding of allocentric space within a spatial map. For instance, the hidden platform is north or 5 m * Corresponding author. Fax: (42) (2) 471-9517; E-mail: bures@sunl .biomed.cas.cz 0166-4328/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved PH S01 66-4328 (95) 00240-5
from the window and east or 3 m from the door. Successful place navigation can also be implemented, however, by an egocentric coding system which deals with spatial relationships between the subject and the goal and leads to coding of relative space. In this system, the rat turned with its back to the wall of the pool learns that the goal is 30 ° to the left and 100 cm distant. Close interaction of the two systems has been stressed by some theoretical [8], behavioral [29] and clinical [23] studies, For instance, when a rat learns to swim in a certain direction from a given starting point, egocentric coding can be responsible for this behavior, whereas repeated checking of the accuracy of such navigation must be done by an allocentric system. Association between the visual allocentric information and the kinesthetic egocentric information is an essential prerequisite for efficient coding of space [ 13 ]. Interpretation of visual signals in the three dimensional world must be supported by other sensory inputs determining the position of the organism with respect
122
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129
to the gravitational field and differentiating self-motion from the movement of the surroundings. Particularly important in this respect is the vestibular system which is essential for the concepts of horizontality and verticality as well as for detecting acceleration or deceleration of the body [25]. The theoretical role of vestibular signals is only rarely supported by experimental analysis of their relative contribution to a specific behavioral situation. Place navigation in the Morris water tank [15-17,26] represents a complex task requiring good orientation in space. According to O'Keefe [18], the cognitive map theory predicts that the animal can anticipate landmark configurations to be seen from any part of the charted environment, and that, even in absence of visual information, it will know its position relative to the starting point by path integration [7,14]. The latter mechanism reflects the capacity of a rat to integrate directional and distance information received during active or passive movement and to translate it into continuously updated azimuthal information about the direction to the starting point [7]. The importance of vestibular cues in this mechanism is demonstrated by the capacity of blind rats to display spontaneous alternation [22] and to solve radial maze tasks [32]. It is conceivable that path integration automatically monitors the position of the rat on the cognitive map between points whose coordinates can be established by observation of landmarks [ 12]. The two systems may also differ by their neural substrate. Whereas the hippocampus seems to be indispensable for allocentric navigation, egocentric navigation may require more expressed participation of cerebral cortex and basal ganglia (see Poucet [21] and Wiener and Berthoz [-30] for reviews). The aim of the present study was to assess the importance of egocentric coding for the acquisition and retrieval of the water maze task under conditions when the start-goal position was constant and the allocentric orientation was eliminated, or at least severely restricted, by total darkness. Cognitive maps increase the flexibility and efficacy of spatial behavior by allowing the animal to rapidly solve newly encountered spatial problems. Whereas this assertion is well supported by vast literature on allocentric orientation [-8] egocentric navigation over fixed start-target routes is sometimes explained as a motor skill or praxis [29]. Since motor skills are usually learned incrementally and cannot be easily reversed or otherwise modified, transfer of egocentric navigation to a different start-target route was used to test the cognitive map versus motor skill explanations of blind navigation.
animals weighed between 250 and 300 g at the start of the experiments and were housed 5 per cage in an animal room with a constant temperature (20°C) and natural lighting. Food and water were always freely available.
2.2. Apparatus The Morris water maze consisted of a uniformly blue circular pool (140 cm in diameter and 55 cm in height) filled to a depth of 25 cm with water (20°C). An ll-cmdiameter clear plexiglass platform stood 1 cm below the water surface in one of the 4 arbitrarily designated quadrants, 35 cm from the center of the pool. An infrared LED connected to a counterbalanced light cable was attached to the animal's body with a rubber band. A infrared-sensitive television (TV) camera was mounted over the center of the pool and used for monitoring the LED movements. The TV signal was digitized and fed into the computerized tracking system which monitored and stored the x - y coordinates of the animal's trajectory at 100-ms intervals. The swim path of each trial and the escape latency as well as the total length of the swim path and the time spent in any designated area of the pool was recorded.
2.3. Behavioral training Rats always navigated in complete darkness. A trial began by turning the light off, taking the animal from the waiting cage and placing it into the water while facing the wall of the pool. The tracking system was started at the time the animal was released. The maximum time the rat was allowed to swim, was 60 s and a dim light was switched on automatically either after this time had expired or after the animal had found and remained on the platform for more than 1 s. When the rat failed to find the platform it was led to it. Climbing onto the platform was the only way of escaping from the water. The animal was always allowed to rest there for 15 s before it was returned to a waiting cage. The light was switched off and the next trial was started approximately 30 s later. There were 8 trials per session. After the last trial, the animal was towel dried and put into another waiting cage under a heating lamp before being returned into the home cage.
2.4. Statistical analysis 2. General method
2.1. Animals Twenty male hooded rats of the Long-Evans strain were obtained from the Institute's breeding colony. The
Analysis of variance (ANOVA) with repeated measures followed by Tukey's test for multiple comparisons was used. Whenever necessary a paired t-test was used for within-group comparisons. The P<0.05 level was accepted as significant.
123
Mojdeh Moghaddam, Jan Bures/Behavioural Brain Research 78 (1996) 121-129
3. Experiment 1 Spatial orientation of animals in nature is implemented by two mechanisms: (1) the allocentric coding system based on the memory of the target coordinates relative to remote extra maze landmarks which leads to the coding of absolute space within a spatial map; and (2) the egocentric coding system based on the memory of the start position and of the start-goal route which leads to coding of relative space. These two systems are tightly inter-related under normal conditions: both are used simultaneously and both contribute to correct performance of the animal. In Expt. 1, we attempted to isolate the egocentric component of navigation between fixed start and goal positions by eliminating the allocentric one by conducting the experiment in total darkness. The importance of the allocentric component was assessed in a control group trained under the same conditions in the light. 3.1. Method
The procedure described in General method was followed. Naive rats (n= 10) were trained in 24 daily sessions (8 trials per session) to swim in complete darkness from the start in the south (S) to the goal platform in the center of the north-west (NW) quadrant of the pool. The combined allocentric and egocentric orientation was tested in another group of 10 animals trained for 4 days (32 trials) with the same start and goal position ( S - N W ) in the light. The same apparatus was used in all the phases of the experiment.
4. Results and discussion The control group trained in the light rapidly learned to search the platform in the inner part of the pool and reached asymptotic performance of 5 s on day 3. The rats trained in the dark persisted longer in attempts to climb the wall before they started random search all over the surface of the pool and gradually oriented toward the target. The mean escape latencies decreased more slowly and reached the asymptotic level of 17 s only in the fourth week. Fig. 1 compares mean escape latencies during the first 4 days of training in the light and in the dark. A two-way ANOVA (groups × days) with repeated measures on the last factor showed significant main effect of groups (F(1,18)=230.18, P<0.001) and days (F(3,54)= 15.24, P <0.001), but no significant interaction (F(3,54)=2.46, n.s.). A paired t-test showed a significant decrease of mean escape latencies in the group trained in the light between days 1 and 2 ( t ( 9 ) = 3.63, P < 0 . 0 1 ) and days 2 and 3 (t(9)=2.89, P<0.01), but not between days 3 and 4 (t(9)=0.32, n.s.) when performance became asymptotic. Training in the dark
- -
Darkness
.....
Light
eo F
4o ti "~
30
Li
t'--.....
2oi
'IF
.....
1
2
, 3
4
days
Fig. 1. Acquisition of the place navigation task under conditions of fixed start (S) and goal (NW) in rats trained in the light (dashed line) or in the dark (full line). Ordinate: mean (+ S.E.M.) escape latencies of 8 daily trials. Abscissa: days of training. Note that the performance improved to asymptoticlevel on day 3 in the light, but that remained almost unchanged in the dark. progressed much more slowly and did not yield a significant reduction of mean escape latencies between successive days. It was, therefore, continued for 5 weeks (5 days per week) until an asymptotic performance was reached. Fig. 2 shows the mean escape latencies on the last 3 days of every week of training. Paired t-tests revealed significant differences between the first and second week (t(9)=5.69, P<0.001), second and third week (t (9) = 3.65, P < 0.01 ), third and fourth week ( t (9) = 4.01, P < 0.01 ), but not between the fourth and fifth week (t(9)=0.15, n.s.). Acquisition of the egocentric navigation is better illustrated by the initial orientation of swimming trajectories than by the efficiency of the search after the rat has failed to locate the goal on the first approach. Fig. 3 shows examples of trajectories generated during the first 5 s after start by a typical rat on the last days of each week. No clear preferred direction is seen in the first week, but a prevailing orientation to the left appears in 60
5O
4O
oo
3O
~ 2o 10
1
3
4
5
woek8
Fig. 2. Acquisition of the place navigation task under conditions of the fixed start (S)-fixedgoal (NW) positions in the dark. The columns represent the mean (+ S.E.M.) escape latency during the last 3 days of every weekly block of 5 days. Note that the escape latencies reached the asymptotic level of 16 s in the fourth week.
Mojdeh Moghaddam, Jan Bures / Behavioural Brain Research 78 (1996) 121-129
124 A
B
was change of the starting point and goal position from S-NW to E-SW. In the animals trained in the dark, a second change from E-SW to N-SE followed after 5 days. The 10 rats trained in the light in Expt. 1 were used in Expt. 2 to check their performance after changing the start and goal position from S-NW to E-SW. 5.2. Results and discussion
C
D
Fig. 3. Computer printouts of the superimposed swimming trajectories in a typical rat on trained in darkness on days 5 (A), 10 (B), 15 (C) and 20 (D). The crossed lines correspond to the WE and NS diameters of the pool. The start mark is in S and the position of the goal is indicated by the double circle in the NW quadrant. Note that the widely dispersed trajectories on day 5 gradually contract to the compact form seen on day 20.
the second week and the trajectories become more compact in the third week and particularly in the fourth week. Note, however, that even at this stage of learning the platform is missed on some of the swims and the ensuing search accounts for the relatively long mean latency. The results demonstrate that acquisition of egocentric navigation (fixed start, fixed goal) in the dark is slower and less efficient than allocentric navigation in the light. Asymptotic performance is reached after 4 weeks in the dark and 4 days in the light with the corresponding escape latencies of 16 and 5 s as shown in Fig. 1 and Fig. 2, respectively.
The last day of training on the first task (S-NW) was compared with the first 2 days of the E-SW training in the dark and in the light. Average escape latencies shown in Fig. 4 were analyzed with a two-way ANOVA (groups x conditions) with repeated measures on the second factor which yielded significant main effects of groups (F(1,18)=48.72, P<0.001) and conditions (F(2,36)=8.50, P<0.001) as well as significant interaction (F(2,36)=4.30, P<0.05). A paired t-test showed a significant increase of escape latencies on the first day of E-SW training both in the dark (t(9)=3.2, P<0.01) and in the light (t(9)=4.2, P<0.01). The difference also remained significant on the second day in the dark (t(9)=3.27, P<0.01), but not in the light (t(9)=0.49, n.s.). E-SW training in the dark continued for 3 more days during which the escape latency decreased to 24.2+2.8 s. A new start-goal rotation from E-SW to N-SE increased the escape latency to 29.4+2.7 s. A paired t-test showed that this difference is not significant (t(9)= 1.27, n.s.). During a further 4 days of training, the escape latency dropped to 18.0+2.0 s on the last day, i.e., to a level not significantly different from the last week of S-NW training. The effect of the start-goal rotation can also be illustrated by the 5-s trajectories shown for a typical rat in Fig. 5. Although the rotation should not have affected egocentric navigation, its accuracy was considerably Darkness
~-~ Light
40
5. Experiment 2 3O
Navigation learned in the dark in Expt. 1 could have been due to allocentric memory for non-visual extramaze cues (acoustic beacons) or to egocentric route memory. The purpose of Expt. 2 was to test the relative significance of these possibilities by rotating the start and goal position by 90 ° counterclockwise (i.e., from S-NW to E-SW and to N-SE). Such rotation should leave egocentric memory unaffected, but would require relearning of allocentric navigation. 5.1. Method
The same overtrained rats and the same apparatus were used in the second experiment. The_only difference
20
S-NW
E-$W
E-SW
Fig. 4. Navigation performance of rats trained in the light (empty columns) or the dark (black columns) on the last day before and on the first 2 days after rotation of the start and goal positions by 90 ° counterclockwise (i.e., from S-NW to E-SW). Other descriptions as in Fig. 1. Note that the performance is back to the asymptotic level on the second day after change in the light, but is only slightly improved in the dark.
Mojdeh Moghaddam, Jan Bure~/BehaviouralBrain Research 78 (1996) 121 129
A
B
D
E
125
Fig. 5. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat trained in darkness on the last day of S-NW training (A), on the first (B) and fourth (C) day of the of E-SW training and on the first (D) and fourth (E) day of the NSE training. Note that the swimming trajectories became more dispersed on the first days after each change, but assumed, after another 3 days, the compactness seen after 3 4 weeks of training in Fig. 3. decreased in the first session after change with a tendency to visit the allocentric location of the previous goal. This is particularly clear in Fig. 5D where most swims pass through former position of the goals in the N W and SW quadrants of the pool. This type of error was, however, rapidly corrected and the animals returned to the somewhat modified start-target route. The above results suggest that the navigation performance at the end of the initial 5-week period of S - N W training was at least partly due to allocentric memory for the position of the goal with respect to acoustic beacons. It seems, however, that the significance of the allocentric component of navigation decreased after several changes of the position of the goal reduced the reliability of the allocentric, but not of the egocentric, orientation.
6. Experiment 3 When the start remains the same (N), but goal location is changed, both egocentric and allocentric navigation should be modified. It was expected that such double change is more difficult than the change of the allocentric component only. 6.1. Method
The same method of training was applied in this experiment, performed only in the 10 rats previously trained in the dark. The animals were started from N, but the goal was moved to SW. The rats were trained in total darkness under these conditions for 5 days.
6.2. Results and discussion
The results of the combined allocentric and egocentric change of the goal position are illustrated in Fig. 6. No deterioration of navigation performance was seen on the first day of the change and mean escape latency was 15.7__+1.4 s in the last 3 days of training. Quadrant preference already shifted on the first day from SE to SW and the overall time spent in the western half of the pool increased from 29% before to 60% after the change. Closer analysis of the first 4 trials of the sessions preceding and following the change indicated (Fig. 7) that the preference already appeared during the second trial and remained well expressed afterwards. The rapid change of the trajectory is shown in Fig. 8B: only the first two swims resembled attempts to find the goal in the area corresponding to its previous allocentric and egocentric location (Fig. 8A). The remaining trajectories of the first session resembled those generated at the asymptotic level of performance (Fig. 8C). Essentially similar results were obtained when, in a subsequent series of experiments, the same rats were started from the North, but the goal was moved to a point 20 cm from the South wall of the pool, i.e., to a location never used before. Fig. 9 shows typical trajectories of another rat before (A) and after the change (B). The first two swims were directed to previous position of the goal in SW, but the subsequent tracks were already aimed at the new goal which was, however, not reached during the 5-s interval displayed. This modification developed during the next 2 days into a faster new route perpendicular to the circumference of the pool (Fig. 9C).
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129
126
A
B
sec
20-
15-
10-
5Q. SW 0-
N-SE
N-SW
N-SE
N-SW
Fig. 6. Navigation performance of rats trained in the dark on the last day before (N-SE) and the first day after (N-SW) the change of the goal position. A: mean (_S.E.M.) escape latency. B: mean (±S.E.M.) percentage of the trial time spent in the 4 quadrants of the pool. Note significant preference against the random choice level (25%) for the goal quadrants before (39.8%, t(9)= 8.7, P < 0.001 for SE) and after the change (40.3%, t(9)=5.3, P<0.001 for SW).
7. General discussion
3C
]
2(~
[]
NWQ
IC
]
SWQ
]
SEQ
0
Trial 1
Trial 2
Trial 3
The main result of the present study is the finding that place navigation learning proceeds more slowly in the dark than in the light, even under conditions allowing egocentric solution of the task, but that overlearned egocentric navigation can be re-directed to new targets with the same ease as the allocentric navigation. This suggests that exploration of the egocentric space leads to formation of an egocentric cognitive map which can be used for goal-directed navigation with similar flexibility as the allocentric maps. Such a view is compatible with the assumption that the allocentric and egocentric maps represent the same space [8], but contrasts with the route learning explanation of the locomotion between fixed start-target points [-29] which assumes that the sequence of movements which carries the animal from the start to the goal is a special case of instrumental learning associating emitted behavior with reinforcement. No map is needed in this case but when the start-goal relationship is changed (as in Expt. 3) a new route has to be learned with almost the same effort as the first one. The present results are relevant in the
NEQ
Trial 4
Fig. 7. Mean (___S.E.M.) percentage of the first 10 s of the trial time spent in the 4 quadrants of the pool in the first 4 trials of the last session before (N-SE, above) and in the first session after (N-SW, below) the change of the goal position. Quadrant labeling is the same as in Fig. 6. Note that the rat preferred the East half of the pool before the change, that the preference for the new goal quadrant became significant since trial 2 (46.5%, t(9)=3.6, P<0.01 for SW) and that the preference for the NW quadrant only started to appear over trials 2 4.
A
B
C
i
Fig. 8. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat during the last day before (A) and the first day after the change of the goal location from SE to NW (B). Note that swims 1 and 2 were directed to the old position of the goal, but that swims 3-8 were already directed to the new goal place. The new route became fully effective after 2 more days of training (C).
Mojdeh Moghaddam, Jan Bures/BehaviouralBrain Research 78 (1996) 121-129 A
B
127
C
i
Fig. 9. Computer printouts of the 8 superimposed swimming trajectories (first 5 s) in a typical rat during the last day before (A) and the first day (B) and third day (C) after the change of the goal location from SW to S. Note that the first two swims after the change were directed to the previous position of the goal, but that further swims gradually shifted to the South and developedinto a new route 2 days later. context of the more than 50-year-old dispute between Guthrie [9], an ardent proponent of the route explanation, and Tolman 1-27] who pioneered the cognitive map concept. They support the view that exploration in darkness (i.e., in egocentric space) does not produce individual start-target routes, but a cognitive map of the space entered from the start area which allows generation of a multitude of routes leading from the start to any point within this space. A better known version of flexible navigation of this type is homing [6,7,14] which allows the animal to return from any point reached on its outer journey directly back to the start.
7.1. Navigation in absence of visual cues While navigation to the hidden platform is usually guided by extramaze landmarks visible from the pool, the goal can be even found in the absence of such landmarks, e.g., in a pool surrounded by black curtains [26], in darkness [10] or by blind rats [3]. Two situations must be distinguished: (1) the target is in a fixed location and can, therefore, be located allocentrically by non-visual extramaze cues (e.g., acoustic beacons); or (2) the target is in one of several possible positions and can be found by a search strategy generating a route passing through all these locations, e.g., a circular route at a constant distance from the wall. Although this distance cannot be directly estimated in the absence of vision, tactile, kinesthetic and vestibular signals could guide the rat into a trajectory satisfying the above requirement. Buresova et al. [3] found, in highly overtrained blind rats, mean escape latencies of 41 and 33 s when using the randomly located goals and fixed goals, respectively. Liu et al. [10] trained rats released from starting points in N, E, S, or W to navigate to a hidden platform in the center of one quadrant of the pool. During 3 days of training in the dark, mean escape latencies decreased from 43 to 39 s, i.e., similar to that in Expt. 1. This indicates that the possible egocentric solution of the task cannot be applied immediately, but only after exploration of the pool in the dark has formed the egocentric cognitive map of the pool. On the other hand, egocentric navigation to a new target
can proceed rapidly when the egocentric cognitive map is already available, as in Expt. 3.
7.2. Dissociation of allocentric and egocentric components of place navigation Continued training of rats in the fixed start-fixed goal navigation task allows the rats to find the goal by combined use of the vestibular and kinesthetic cues in the egocentric space and of non-visual cues available in the allocentric space. The performance reached at the end of Expt. 1 was obviously due to the additive effect of these two factors. Simultaneous rotation of the start and goal positions dissociated these two components: whereas the egocentric component remained unaffected, the allocentric was now pointing in a different direction. The conflict between these two components accounted for an increase of escape latencies from 13 s to 31 s. Assuming that the allocentric component was added to the egocentric one before the change and subtracted from it after the change, the egocentric component alone should be between 13 and 31 s, i.e., 22 s. The next start-goal rotation elicited a less expressed prolongation of escape latencies which gradually approached the egocentric component. Purely egocentric navigation can also be obtained when allocentric orientation is prevented by everyday change of the goal position. This technique used by Save and Moghaddam [24] yielded, after 24 days of training, mean escape latencies of 27 s, uncontaminated by aUocentric components.
7.3. Flexibility of egocentric cognitive maps The difference between acquisition of the same navigation task in the light and in the dark is mainly due to the fact that the egocentric map is formed more slowly than the allocentric map of the same environment. Once the egocentric map is formed, however, it can be used for solving the navigation tasks with the same flexibility as the allocentric map. This is well documented for the working memory version of the place navigation task 1-15,20,28] in which the animal is exposed to a new position of the escape platform on the first swim and is able to find the platform much faster on the second
128
Mojdeh Moghaddam, Jan Bures/Behavioural Brain Research 78 (1996) 121 129
swim performed several tens of minutes or several hours later. When the goal position is changed and the animal does not find it in the expected place, the failure activates the search routine. After the goal is found, its egocentric coordinates are entered into the egocentric map and can support navigation to this place from the start. Since, on each trial, the animal either found the escape platform or was guided to it, the rat had all the information needed for entering the goal position into his egocentric cognitive map for the particular start and for plotting the putative start-goal direction by dead reckoning. The above explanation must be accepted with caution, however. Since the experimental conditions did not completely eliminate the use of non-visual allocentric cues (Expt. 2), allocentric mechanisms might have contributed to the rapid change of the route in Expt. 3. This possibility is enhanced by the fact that the new goal position in Expt. 3 corresponds to the location (SW) already used in the initial phase of Expt. 2 which may still be remembered by the animals on the basis of allocentric cues. Such an explanation is improbable, however, because allocentric orientation alone does not support efficient navigation in the dark [3,10]. Furthermore, the rats demonstrated similar flexibility when learning a new route to a goal position never used in the previous training (e.g., from North to South, Fig. 9). To rule out the contribution of allocentric orientation in the above situation, allocentric cues should be eliminated by an experimental design [24] in which the start-goal geometry with respect to the circular pool remains stable, but the allocentric position of the starting point changes randomly from trial to trial. In this case, allocentric coordinates of the goal become irrelevant and optimum solution of the task is only possible with the use of the egocentric map. Further research is needed to verify the postulated flexibility of egocentric mapping under the above conditions.
Acknowledgement The authors thank Mr. programming and Mr. A. tance. This research was AVCR 711401 and BMFT
Yu. Kaminsky for computer Zahalka for technical assissupported by Grants IGA 01VJ 920015/26-5a.
References [ 1] Brandeis, R., Brandys, Y. and Yehuda, S., The use of the Morris water maze in the study of memory and learning, Int. J. Neurosci., 48 (1989) 29-69. [2] Bures, J. and Buresova, O., Spatial memory in animals. In: R.E.
John (Ed.), Machinery of Mind, Birkhauser, New York, 1990, pp. 291-310. [3] Buresova, O., Homuta, L., Krekule, I. and Bures, J., Does nondirectional signalization of target distance contribute to navigation in the Morris water maze?, Behav. Neural Biol., 49 (1988) 240-248. I-4] Cartwright, B.A. and Collett, T.S., How honey bees use landmarks to guide their return to a food source, Nature, 295 (1982) 560 564. 1-5] Eichenbaum, H., Otto, T. and Cohen, N.J. The hippocampus what does it do?, Behav. Neural Biol., 57 (1992) 2-36. [-6] Etienne, A.S., Maurer, R. and Saucy, F., Limitations in the assessment of path dependent information, Behaviour, 106 (1988) 81 111. 1-7] Etienne, A.S., Maurer, R., Saucy, F. and Teroni, E., Short-distance homing in the golden hamster after a passive outward journey, Animal Behav., 34 (1986) 696-715. 1-8] Gallistel, C.R., The Organization of Learning, MIT Press, Cambridge, 1990. 1-9] Guthrie, E.R., Tolman on associative learning, Psychol. Rev., 44 (1937) 525 529. [10] Liu, Z., Francis Turner, L. and Bures, J., Impairment of place navigation of rats in the Morris water maze by intermittent light is inversely related to the duration of the flash, Neurosci. Lett., 180 (1994) 59 62. 1-11] McNaughton, B.L., Neural mechanisms for spatial computation and information storage. In: L. Nadel, L.A. Cooper, P. Culicover and R.M. Harnish (Eds.), Neural Connections, Mental Computations, MIT Press, Cambridge, MA, 1989, pp. 285-350. [12] McNaughton, B.L., Chen, L.I. and Markus, E.J., Dead reckoning, landmark learning and the sense of direction: a neurophysiological and computational hypothesis, J. Cognitive Neurosci., 3 (1991) 190 202. [13] McNaughton, B.L., Leonard, B. and Chert, L., Cortical-hippocampal interactions of cognitive mapping: an hypothesis based on reintegration of the parietal and inferotemporal pathways for visual processing, Psychobiology, 17 (1989) 230-235. [14] Mittelstaedt, H. and Mittelstaedt, M.-L., Homing by path integration. In: F. Papi and H.G. Wallraff (Eds.), Avian Navigation, Springer Verlag, Berlin, 1982, pp. 290-297. [15] Morris, R.G.M., Spatial localization does not require the presence of local cues, Learning Motiv., 12 (1981) 239-261. [16] Morris, R.G.M., An attempt to dissociate spatial mapping and working memory theories of hippocampal function. In: W. Seifert (Ed.), Neurobiology of the Hippocampus, Academic Press, New York, 1983, pp. 405-432. [17] Morris, R.G.M., Developments of a water-maze procedure for studying spatial learning in the rat, J. Neurosci. Methods, 11 (1984) 47-60. 1-18] O'Keefe, J., A computational theory of the hippocampal cognitive map. In: J. Storm-Mathisen, J. Zimmer and O.P. Ottersen (Eds.), Understanding the Brain through the Hippocampus, Progress in Brain Research, Vol. 83, Elsevier, Amsterdam, 1990, pp. 301 312. [19] O'Keefe, J. and Nadel, L., The Hippocampus as a Cognitive Map, Clarendon, London, 1978. [20] Panakhova, E., Buresova, O. and Bures, J., Persistence of spatial memory in the Morris water tank task, Int. J. Psyehophysiol., 2 (1984) 5-10. [21] Poucet, B., Spatial cognitive maps in animals: New hypotheses on their structure and neural mechanisms, Psychol. Rev., 100 (1993) 163-182. [22] Potegal, M., Day, M.J. and Abraham, L., Maze orientation, visual and vestibular cues in two-maze spontaneous alternation of rats, Physiol. Psychol., 5 (1977)414 420. [23] Ratcliff, G., Brain and space: some deductions from the clinical evidence. In: J. Paillard (Ed.), Brain and Space, Oxford University Press, Oxford, 1991. [24] Save, E. and Moghaddam, M., Effects of lesion of the associative parietal cortex on the acquisition and use of spatial memory in
Mojdeh Moghaddam, Jan Bures/ Behavioural Bra& Research 78 (1996) 121-129
[25] [26] [27] [28]
[29]
egocentric and allocentric navigation tasks in the rat, Behav. Neurosci., in press. Stoffregen, T.A. and Riccio, G.E., An ecological theory of orientation and the vestibular system, Psychol. Rev., 95 (1988) 3 14. Sutherland, R.J. and Dyck, R.H., Place navigation by rats in a swimming pool, Canad. J. Psychol., 38 (1984) 322-347. Tolman, E.C., Determiners of behavior at a choice point, Psychol. Rev., 45 (1938) 1 41. Whishaw, I.Q., Formation of a place-learning set by the rat: a new paradigm for neurobehavioral studies, Physiol. Behav., 35 (1985) 139-143. Whishaw, I.Q. and Mittleman, G., Visits to starts, routes, and
129
places by rats (Rattus norvegicus) in swimming pool navigation tasks, J. Comp. Psychol., 100 (1986) 422 -431. [30] Wiener, S. and Berthoz, A., Forebrain structures mediating the vestibular contribution during navigation. In: A. Berthoz (Ed.), Multisensory Control of Movement, Oxford University Press, Oxford, 1993, pp. 427 456. [31] Wilkie, D.M. and Palfrey, R., A computer simulation model of rat's place navigation in the Morris water maze, Behav. Res. Methods, Instr. Comput., 19 (1987) 400-403. [32] Zoladek, L. and Roberts, W.A., The sensory basis of spatial memory in the rat, Anita. Learning Behav., 6 (1978) 77 81.