Experimental analysis of spatial learning in goldfish.
The "spatial map" theory (O'Keefe & Nadel, 1978) gives a good account of the mechanism of spatial learning. The theory is that animals form a spatial map of the environment representing relative locations. According to the spatial map theory, in the Olton radial maze, rats should form a spatial map representing the relative locations of extramaze cues and the arm with the bait. In the Morris water maze, rats should form a spatial map representing the relative locations of extramaze cues and the position of a hidden platform to escape from the water. There is a possibility, however, in the radial maze that animals learn the absolute locations of the goals and the extramaze cues. Specifically, the platform on which the animals can make a choice of arm is fixed, hence they might use the absolute location of the particular extramaze cue from the selecting point as a local cue and might not learn the relative locations. On the contrary, animals learn the location of platform relative to multiple extramaze cues in the Morris water maze because they can move freely in the maze. The radial arm maze may not fulfill the necessary conditions to force animals to make a "spatial map."
There was a study that was analogous to the radial maze in goldfish (Rodriguez, Duran, Vargas, Torres, & Salas, 1994), which compared the learning of allocentric (spatial relationship all around the maze) and egocentric strategies (spatial relationship from the subject) in a four-arm radial maze. In the allocentric strategy, the relationship between the starting arm, objects placed outside the maze (extramaze cue), and the goal arm was identified, while in the egocentric strategy, the directional relationship between the starting arm and the goal arm was identified. Goldfish were able to learn both egocentric and allocentric strategies.
To examine whether goldfish can form a "spatial map," an open-field maze for fish should be established like the Morris water maze. However, there are no fish experiments that are comparable to the Morris water maze or its "dry" version. We developed a new spatial learning task for fish that was compared to a dry version of the water maze for rodents (Kesner & Dakis, 1995; Kubie, Sutherland, & Muller, 1999). The dry version is different from the Morris water maze in the nature of the reinforcement used. In the dry version, the reinforcement is food, and in the water maze, it is avoidance of water. The important point is that learning in the dry version depends on the extramaze cues, which is similar to the Morris water maze.
In the present experiments we investigated whether fish could learn open-field tasks similar to those in the Morris water maze. The tasks are performed for a food reward. We examined the effects of local intramaze cues and of cues from several sensory modalities and tested whether learning in goldfish depends on extramaze cues.
Animals and Maintenance
Goldfish (Carassius auratus) from a local pet shop were used. The fish ranged from 6.5 to 7.5 cm in length and from 6.8 to 8.8 g in weight. They were kept in an aquarium for more than 2 weeks before the experiments. A 13L/11D artificial illumination cycle was employed. The fish received flaked fish food and fresh bloodworms twice a day, but they were food deprived for 2 days before behavioral training. The subjects were randomly separated and they lived in groups of 4 per aquarium. There were 4 aquariums.
The experimental maze was a blue polypropylene circular tank with a diameter of 94 cm and a depth of 10 cm. The water level was 5 cm below the top of the tank. The higher the water level, the narrower the visual field, but the longer the distance between the subject and the food in the baited hole. The shallowest level needed for fish to move freely was 5 cm, although we wanted to narrow the visual field of the fish as much as possible. The water temperature was kept at 25 [+ or -] 0.5 [degrees]C, and the water was changed every day. The experimental room was illuminated with fluorescent lamps, but covering sheets kept the direct light from the maze because it might have produced many unnecessary intramaze cues. The floor was white and made from an acrylic board with holes 0.65 cm in diameter and 0.5 cm deep. There were 16 holes in the apparatus. The minimum distance between the holes was 15 cm and the distance from the walls to the nearest hole was more than 15 cm. The start box was a white open-bottom pail with a handle, which was used as the starting position. The width of the bottom was 12 cm and the width of the upper opening was 13 cm. The height of the pail was 9.0 cm and the wall was 5 mm thick. When we lifted the pail, the subject was able to start exploring the pool. We did not touch the subjects directly. They were transferred from the aquarium to the start box with a nylon net. The behavior of the subjects was monitored through a CCD camera (Sony, XC-711) connected to a chromascan (Oyokeisoku, KWX-01) that detected the color of the subject and a videotracker (Oyokeisoku, G220) that digitized an image of the subject's movement with 6 Hz. Data from the tracker were sent to a computer (FMV-UV). The behavior was also monitored on a video display (SONY, PVM-MLJ).
The maze was centered in a 4-m square laboratory in which many cues, including shelves, songbird cages, and walls, were present.
Bloodworms (Chironomus plumosus) were used as reinforcers. They ranged in weight from 12 to 17 mg and in total length from 10.4 to 13 mm, and their widest parts of the breast ranged from 0.9 to 1.0 mm in thickness. They fit into the holes in the present apparatus so that the fish could not see them.
The bloodworms were frozen and they were removed from the freezer approximately 20 min before the experiment. By using a dead bloodworm, we could reduce visual cues, such as moving, and olfactory cues, such as metabolic discharge.
Half of the holes were baited with bloodworms. Each subject was released in the middle of the maze and was allowed to explore the holes and to eat food for 10 min. This procedure continued for one trial on each of 3 days. Subjects that did not eat food in the maze were not included in the study.
During training, only one hole was baited. In Experiments 1 to 3, without the changing group in Experiment 1, the position of the baited hole for each subject was always in the same position and was determined as the second hole from the end of 4 X 4 holes. The position of the baited hole in the laboratory is shown in Figure 1. The baited hole was west. The subject was released from the pail and directed toward the wall of the maze. The pail was placed at four points: northwest, southwest, northeast, and southeast; and the releasing points were randomly selected from the four points. The subject was released from the pail and oriented towards the wall of the maze, a procedure we used for controlling the field of the subject's vision and the subject's motion soon after release. If we did not control the subject's orientation, latency to reach the goal would vary widely because fish are not skillful at changing direction, and the path trajectory in releasing toward the wall is different from that in releasing toward the middle of the maze. Each subject was allowed to explore the maze for a maximum of 120 sec. When the subject found the food, it was allowed to eat for 30 sec and it was then returned to the aquarium. The fish were returned with a nylon net, which was used to transfer them from the aquarium to the pail. If the subject could not find the food within 120 sec, the experimenter held the abdomen of the subject with his or her hand and guided the fish to the baited hole. The subjects rarely spent more than 120 sec in reaching the goal. In the cases where the fish needed to be guided, there were no apparent aversive effects of the process on the performance in the next trial. This trial was repeated four times a day and continued for 5 days. The intertrial interval was 20-25 min. The behavior of the subject was tracked with a video tracking system and the latency to reach the food was measured with a stopwatch. There were two criteria for reaching the goal. The first was that the subject stopped at the baited hole, and the second was that the animal was observed pecking at the baited hole on the video tracker.
[FIGURE 1 OMITTED]
In the dry version of the water maze, latency to reach the food, distance traveled to find the food, and path trajectory have been used as dependent indexes (Kesner & Dakis, 1995; Kubie, Sutherland, & Muller, 1999; Whishaw, 1989; Whishaw & Tomie, 1996). In the present study, latency to reach the goal, distance traveled to find the food, and swimming trajectory were used as dependent measures.
We did not measure the amount of time that the subjects spent in the baited quadrants versus the nonbaited quadrants, because the amount of time did not reflect the spatial memory, as fish do not stay in either nonbaited or baited areas for a long time.
The mean of the index of the latency to reach the food site (goal) is not same as that of the swimming distance to find the food because there is a flaw in the data for latency. If the subject moved faster, the decrement of the latency did not show an improvement in learning when the distance to travel is used. The number of errors from visiting nonbaited holes could be considered to be learning criteria, but we did not apply the index because it is difficult to define a fish's error. Defining errors in the radial arm is easy, because an error is defined as entering the wrong arm. However, in the present apparatus, it is difficult to judge whether a fish is going through the wrong hole because passage through a nonbaited hole to get to a baited hole is not an error. Latency and swimming distance allow for quantitative judgment, while path trajectory allows for qualitative judgment.
If the latency exceeded 120 sec, it was recorded as 120 sec and the distance was set to the length of swimming within 120 sec.
Experiment 1: Spatial Learning vs. Searching Strategy
In Experiment 1, spatial learning in fish was examined in a maze with latticed holes. There was a possibility that fish would learn a strategy to reach the goal, but not the specific position of the food. The baited hole was fixed at the same position for one group, and the position was changed in each trial for the other group. In the fixed-hole group, fish could navigate themselves by referring spatial information from extramaze cues to the goal, but they can not do this in the changing-hole group. If the fish simply learn a strategy to search for food quickly without spatial learning, both groups should have the same performance, whereas if the fish in the fixed-hole group learn the position of the food, their performance should be better than that in the changing-hole group.
Twenty-two fish were used. The subjects were divided into two groups: 13 in the fixed-hole group and 9 in the changing-hole group. Each subject was experimentally naive.
Apparatus and Procedure
The maze had 4 X 4 latticed holes. The distance between the holes was 15 cm and the distance from the walls to the nearest holes was more than 15 cm.
For habituation trials the rewards were placed in every other hole and we changed the location of the baited holes every day. Thus, eight holes were baited. After habituation, subjects were trained for five sessions (four trials per session).
For one group, the subject was trained to search for one fixed position baited hole, and the position of the baited hole remained fixed over all training sessions. For the other group, the subject was trained to search for one changing-position baited hole, and the position of the baited hole was changed for each trial. The baited holes were assigned to 4 positions that were selected at random from the latticed 16 hole positions for every subject and for every training day. In both groups the releasing points were randomly selected from four points and no point was repeated within a session.
Results and Discussion
In the changing-hole group, 2 subjects were excluded from the study because they never ate food in the habituation session. In the fixed-hole group, 11 fish were included. Because of their whitish body color, 2 fish that could be not detected were excluded, and their distance data were discarded. There were 8 fish in the changing-hole group.
The mean latency to reach the goal and the mean swimming distance are shown in Figure 2, and the two samples of the path trajectories of both groups are shown in Figure 3.
The data were analyzed using two-way ANOVAs (split-design) and the variables were group (changing-hole/fixed-hole), sessions (1 to 5), and interaction of group and sessions. Two-way ANOVAs for latency showed significant main effects for group, F(1, 17) = 8.40, p < 0.01, and sessions, F(4, 68) = 52.27, p < 0.001, and a significant interaction of group and sessions, F(4, 68) = 1.71, p > 0.10.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
Subanalyses are shown below. One-way ANOVAs of the latency in the fixed-hole group showed a significant effect of sessions, F(4, 44) = 23.00, p < 0.0001, and one-way ANOVAs of the latency in the changing-hole group also showed a significant effect of sessions, F(4, 35) = 15.07, p < 0.001. In the fixed-hole group, there were significant differences in latency between Session 1 and Sessions 3, 4, and 5 and between Session 2 and Sessions 4 and 5 (Tukey's HSD test, p < 0.05), but no significant difference between other sessions. This suggests that fish learning within Sessions 1 and 2 was conspicuous. By comparison, a one-way ANOVA of latency during the training session in the fixed-hole and the changing-hole groups showed a significant effect of the groups after the 2nd session: 1st session, F(1, 17) = 0.312, p > 0.10; 2nd session, F(1, 17) = 19.52, p < 0.001; 3rd session, F(1, 17) = 11.53, p < 0.01; 4th session, F(1, 17) = 14.03, p < 0.01; and 5th session, F(1, 17) = 5.63, p < 0.05.
The mean swimming distance in both groups also decreased across sessions and indicated learning in both groups. The data for swimming distance were similar to those for latency. Two-way ANOVAs for swimming distance showed significant main effects of group, F(1, 15) = 15.32, P < 0.001, and sessions, F(4, 60) = 48.65, p < 0.001, and significant interaction of group and sessions, F(4, 60) = 11.28, p < 0.001. One-way ANOVAs of the swimming distances in the fixed-hole group showed a significant effect of sessions, F(4, 44) = 15.33, p < 0.0001. One-way ANOVAs of the swimming distances in the changing-hole group showed a significant effect of sessions, F(4, 35) = 15.63, p < 0.0001. A one-way ANOVA of the swimming distance in the training session of the two groups (fixed-hole and changing-hole) also showed a significant effect of the groups after the 2nd session: 1st session, F(1, 17) = 2.32, p > 0.10; 2nd session, F(1, 17) = 8.54, p < 0.05; 3rd session, F(1, 17) = 21.9, p < 0.001; 4th session, F(1, 17) = 6.05, p < 0.05; 5th session, F(1, 17) = 11.82, p < 0.01).
Improvement in performance of the changing-hole group suggests that the changing-hole group learned a strategy to find the goal efficiently through the training. The learning might be caused by an improvement of skill in swimming and searching in the maze.
Figure 3 demonstrates that the path trajectory in the changing-hole group clearly differed from that in the fixed-hole group. The data for the trajectory showed that the fixed-hole group swam directly to the baited hole, but the changing-hole group stopped at many holes until reaching the goal. The data for swimming distance showed a quantitative difference between the groups. In addition, the data for the trajectories showed a qualitative difference. These results demonstrate that the fish in the changing-hole group learned a strategy to search for the goal without learning a specific position. Thus, differences between the fixed-hole group and the changing-hole group indicated "net" spatial learning.
The 16-lattice configuration had three other holes that were geometrically equivalent to the baited hole. If fish used the latticed configuration as an intramaze cue, they would have shown a preference for accessing the three other holes. However, the results of the path trajectory did not show a preference for these holes.
There was a possibility that the fish could see the food and were thus able to reach the goal. We eliminated this possibility, as follows. We used bloodworms that were less than 1.0 mm thick. The holes in the floor were 5.0 mm deep and 6.5 mm wide. The depth of the water was 5.0 cm. The body of the subject was just under the surface of the water. The eye was located less than 4.0 cm from the floor. Therefore, we calculated that the subject could not see the food from more than 6.5 cm away from the hole. The distance between two contiguous holes was more than 15 cm. If the fish depended on seeing the bait, it had to travel randomly until it was within 6.5 cm of the food.
There was a further possibility that the fish could smell the food. If the olfactory cues of the food had prompted the performance of the fish, the fixed-hole and changing-hole groups might have had similar results. In fact, these groups showed different latencies and swimming distances, suggesting that no olfactory cues were used by the fish.
Another important point is that the data for the path trajectory in the changing-hole group did not support the idea that the fish could directly see the food. If the fish could see the food, they would have shown a straight trace to the goal on the trajectory.
Experiment 2: Examination of Intramaze and Extramaze Cues
The results of Experiment 1 suggested fish spatial learning, but there is a possibility that the fish used local intramaze cues and not extramaze cues. To examine possible intramaze cues, we rotated the wall or floor 180[degrees] after the training. If fish used local intramaze cues, they would perform poorly in the rotation test. To examine extramaze cues, the extramaze cues were directly manipulated by surrounding the maze with curtains after the training. If fish used extramaze cues, they would perform poorly with curtains.
Twenty experimentally naive fish were used for the examination of intramaze cues. Nine experimentally naive fish were used for the examination of extramaze cues.
Apparatus and Procedure
The habituation and training sessions were the same as those used for the fixed-group in Experiment 1.
In the intramaze control group, following 5 days of training, the subjects were tested in the part of the maze that was rotated. In the floor rotation test, the floor was rotated 180[degrees] relative to the wall of the maze and the experimental room; whereas in the wall rotation test, the wall was rotated 180[degrees] relative to the floor and the experimental room. The relative position of the baited hole to the experimental room was fixed throughout the training and the test.
In the extramaze control group, following 5 days of training, the subjects were tested in the maze after it was covered with curtains. The curtains were hung from the ceiling of the experimental room to the edge of the maze. They were parallel to the wall of the maze, and vertical to the ground. The ceiling of the experimental room was not covered and remained to give the effect of the extramaze cues. The curtains were made of opaque green cloth. Two new lights were set at a space between the wall of the maze and the curtain to avoid a decrease in the visual field of the fish because the fluorescent illuminations were also turned off and the maze might have been too dark without the new lights. Because of the position of the lights, there was no direct illumination. One test session consisted of four trials.
[FIGURE 4 OMITTED]
Results and Discussion
In the intramaze control group, behavior in the test sessions was compared with that in the last session of training. Figure 4 shows that mean latency to reach the goal in both rotation tests was similar to that on the last day of training. The data were analyzed using a repeated-measures ANOVA and the variables were session (5th training / test session with rotating intramaze cues) and subject. A repeated-measures ANOVA for the latency to reach the goal between the last day of training and the rotating floor test showed no significant main effect of session, F(1, 9) = 3.83, p > 0.05, and subject, F(9, 9) = 0.91, p > 0.10. And in the latency between the last day of training and the rotated wall test there was no statistically significant main effect of session, F(9, 9) = 2.95, p > 0.10, but there was a statistically significant main effect of subject, F(9, 9) = 3.57, p < 0.05. A repeated-measures ANOVA for the swimming distance to reach the goal between the last day of training and the rotating floor test also showed no significant main effects of session, F(1, 9) = 1.59, p > 0.10, and subject, F(9, 9) = 1.57, p > 0.10. And there was no statistically significant effect in the swimming distance between the last day of training and the rotated wall test in session, F(1, 9) = 1.16, p > 0.10, but a significant effect in subject, F(9, 9) = 4.69, p < 0.05. Thus, the results strongly demonstrated spatial learning without intramaze cues.
In the extramaze control group, behavior in the test session was compared with that in the last session of training. Figure 5 shows that mean latency to reach the goal in the test was not similar to that on the last day of training. The data were analyzed using a repeated-measures ANOVA and the variables were session (5th training / test session with curtains) and subject. A repeated-measures ANOVA for the latency to reach the goal showed a significant effects for session, F(1, 8) = 20.86, p < 0.0001, and for subject, F(8, 8) = 3.53, p < 0.05. A repeated-measures ANOVA for the swimming distance to reach the goal also showed a significant effects for session, F(1, 8) = 78.23, p < 0.0001, and for subject, F(8, 8) = 6.53, p < 0.01. Thus, the elimination of extramaze cues disturbed performance.
[FIGURE 5 OMITTED]
The increase in latency during testing might be caused by disruptive change in the environment. If the sudden environmental change affected motivational or attention factors, the fish performance improved across the four trials in the test. However, a one-way ANOVA showed no significant effect in latency in the four test trials, F(3, 32) = 0.92, p > 0.1.
This finding supports the conclusion in Experiment 1 that spatial learning in the present maze was based on extramaze cues.
Experiment 3: Examination of Visual vs. Olfactory Cues
The results in Experiment 2 indicate that the fish used visual extramaze cues. In order to support this suggestion, we examined the effect of sensory modalities. To determine sensory modalities that are crucial for spatial learning, we cut the olfactory tract to deprive the fish of olfactory senses or we performed eye enucleation to deprive them of visual senses.
Thirty experimentally naive fish were used. They were randomly divided into three groups.
Apparatus and Procedure
The habituation and training were the same as in Experiments 1 and 2. The locations of the holes were fixed at west. Following 5 days of training, the subjects were divided into the olfactory tract section (N = 10), eye enucleation (N = 10), and no-treatment groups (N = 10). The two sensory-deprived groups underwent surgery on the day following the training, and received a test session with four trials, 5 days after surgery. The no-treatment group received a test session 6 days after the training.
The subjects in the olfactory tract section group were anesthetized with a 1.0% solution of urethane (Aldrich) for 10 min. Then they were fixed in a stereotaxic apparatus that was modified for fish. A plastic tube mouthpiece, connected with a flexible tube providing a 0.5% solution of urethane or fresh water, was inserted into the mouth of the fish. The skin and skull were carefully opened and the olfactory tract was cut between the olfactory bulb and the telencephalon with a surgical knife under a microscope. The hole in the skull was filled with dental cement.
The subjects in the eye-enucleation group were anesthetized with iced water. The iris was cut through and the retina was stripped off with a small piece of cotton. The fish in the eye-enucleation group were able to eat by themselves.
Results and Discussion
The color of one subject in the control group could not be detected by the chromascan, so the swimming distance of that fish was not analyzed. Figure 6 shows latency to the goal. Every group showed a gradual decrease in latency and swimming distance during the training. An ANOVA in split-plot design for the latency to reach the goal in the three groups showed no significant main effect of group, F(2, 27) = 0.07, p > 0.10, but there was a significant main effect of sessions, F(4, 108) = 38.98, p < 0.001), and no significant interaction of group and sessions, F(8, 108) = 0.28, p > 0.10. An ANOVA in split-plot design for the distances in the three groups showed no significant main effect of group, F(2, 26) = 0.18, p > 0.10. There was a significant main effect of sessions, F(4, 107) = 31,78, p < 0.001, but no significant interaction of group and sessions, F(8, 107) = 0.82, p > 0.10. Thus there was no significant difference among the three groups during the preoperative learning.
[FIGURE 6 OMITTED]
There was no statistically significant difference between the control and the olfactory tract section groups, t = 0.34, p > 0.10. These results clearly showed that the fish used visual, not olfactory, cues to learn the spatial learning task. Although impairment in feeding behavior has been reported after olfactory tract section (Stacey & Kyle, 1983), the present results did not show a significant difference between the subjects with olfactory tract sectioning and the control groups.
As shown in Figure 5, no subject could reach the goal after the eye-enucleation while the subjects maintained their discrimination after the olfactory tract sections because the eye-enucleated fish ate flaked fish food and fresh bloodworms in the aquarium and survived 3 months after the experiment. Although eye-enucleated fish could use the taste sense and the possibility that the gustatory sense might be used as a sensory cue remains, the results in Experiment 3 showed that the taste sense was not used as local cues and is incapable of guiding navigation. Because the surgery of eye enucleation gave a great damage to the subject, it might disrupt the processing of sensory cues other than the visual sense. However, motivation to eat was not affected, and the fact that swimming distance after eye enucleation did not differ from that in the first session of training did not support the possibility that the poor performance could be attributed to motor or sensorimotor deficits. This supports the interpretation that the learning of the present task was based on visual extramaze cues.
Although there were two differences between the control and olfactory tract sectioned groups (surgical opening of the skull and injury to the skin and damage to the olfactory sense), there were no significant differences between the performances of the control and olfactory tract sectioned groups. The results showed that surgery in addition to olfactory impairment had no significant effect on subject's performance and suggested that we should not provide a sham-operated group that surgical effect could be detected.
We tried to examine whether goldfish learned the task in the maze based on "extramaze" cues. In Experiment 1, the latency to reach the baited hole and the swimming distance of the changing-hole group differed quantitatively from those of the fixed-hole group. Moreover, the path trajectory of the changing-hole group differed qualitatively from that of the fixed-hole group. The results did not support the possibility that the fish could see the food directly and that they learned a searching strategy rather than spatial information. However, there remained a possibility that spatial learning in Experiment 1 was based on intramaze cues. We examined the effects of learning based on local intramaze and extramaze cues in Experiment 2. Results in Experiment 2 indicated that local intramaze cues did not affect the performance in the maze and the learning of the fish depended on extramaze cues. In addition, the sensory modality was examined in Experiment 3. The results in Experiment 3 also demonstrate that learning in the present task was spatial learning based on visual information.
Goldfish certainly used the extramaze cues in the present spatial navigation task, although we did not identify them. There is also the possibility that the fish used path integration, which is the ability of animals to retrace or navigate their routes by relying on internal cues or by reading the position of the sun or a magnetic field. However, the facts that the eye-enucleated fish did not reach the goal and that difference in the releasing points across trials had no effect suggested that the fish did not use path integration.
Spatial learning in teleost fish has been seen previously in the Y maze (Zerbolio & Wickstra, 1980) and the 4-arm maze (Rodriguez et al., 1994). We found that goldfish exhibited spatial learning in a maze, the hole board task, which resembles the Morris water maze.
The present experiments provide a reasonable method to compare the spatial learning of fish with other animals. One of the characteristics of our maze task is that fish learned the spatial task quickly. They showed improvement within three training sessions. Such fast learning is similar to that of rats in the Morris maze. The other characteristic is the method of orienting to the maze and to extramaze cues. In the Y maze or radialarm maze, the route and the choice point of orienting are fixed, whereas in the Morris maze and our maze task, the subjects themselves decide the route and the choice point.
The present study showed that goldfish, like rodents, learned not only the fixed-route task, as in the Y or radial maze, but also the free-orienting (open field) task. This suggests that the capacity for spatial learning in goldfish is comparable to that of rodents. Moreover, with the present apparatus, we expect to closely study the mechanism of spatial learning in goldfish.
Studies of spatial learning in fish are particularly interesting because fish have many of the characteristics that are thought to exist in common ancestors of birds and mammals. Because of obvious structural and gross behavioral differences, and environmental differences in which the fish live, comparison with other vertebrates may reveal much about fish themselves and about animals as a whole.
Our data present a method for lesion studies of fish spatial learning and for comparative neural studies. Salas, Rodriguez, Vargas, Duran, and Torres (1996) showed that telencephalic ablated goldfish were unable to perform spatial tasks. Further studies should examine the intra-network of the goldfish telencephalon.
DODSON, J. J. (1988). The nature and role of learning in the orientation and migratory behavior of fishes. Environmental Biology of Fish, 23, 161-182.
KESNER, R. P., & DAKIS, M. (1995). Phencyclidine injections into the dorsal hippocampus disrupt long-but not short-time memory within a spatial learning task. Psychopharmacology, 120, 203-208.
KUBIE, J. L., SUTHERLAND, R. J., & MULLER, R. (1999). Hippocampal lesions produce a temporally graded retrograde amnesia on a dry version of the Morris swimming task. Psychobiology, 27, 313-330.
O'KEEFE, J., & NADEL, L. (1978). The hippocampus as a cognitive map. Oxford: Oxford University Press.
QUINN, T. P., & BRANNON, E. L. (1980). The use of celestial and magnetic cues by orienting sockeye salmon fly. Journal of Comparative Physiology A, 147, 547-552.
REEBS, S. G. (1993). A test of time-place learning in a cichlid fish. Behavioral Processes, 30, 273-282.
REEBS, S. G. (1996). Time-place learning in gold shiners (Pisces: Cyprinidae). Behavioral Processes, 36, 253-262.
RODRIGUEZ, F., DURAN, E., VARGAS, J. P., TORRES, B., & SALAS, C. (1994). Performance of goldfish trained in allocentric and egocentric maze procedures suggests the presence of a cognitive mapping system in fishes. Animal Learning & Behavior, 22, 409-420.
SALAS, C., RODRIGUEZ, F., VARGAS, J. P., DURAN, E., & TORRES, B. (1996). Spatial learning and memory deficits after telencephalic ablation in goldfish trained in place and turn maze procedures. Behavioral Neuroscience, 110, 965-980.
STACEY, N. E., & KYLE, A. L. (1983). Effects of olfactory tract lesions on sexual and feeding behavior in the goldfish. Physiology & Behavior, 30, 621-628.
WARBURTON, K. (1990). The use of local landmarks by foraging goldfish. Animal Behaviour, 40, 500-505.
WHISHAW, I. Q. (1989). Dissociating performance and learning deficits on spatial navigation tasks in rats subjected to cholinergic muscarinic blockade. Brain Research Bulletin, 23, 347-358.
WHISHAW, I. Q., & TOMIE, J. (1996). Of mice and mazes: similarities between mice and rats on dry land but not water mazes. Physiology & Behavior, 60, 1191-1197.
ZERBLIO, D. J., & WICKSTRA, L. L. (1980). Instrumental avoidance acquisition by goldfish in a Y-maze using explicit and response-contingent cues. Animal Learning & Behavior, 8, 304-310.
KOTARO SAITO and SHIGERU WATANABE
Keio University, Tokyo
Correspondence may be sent to Kotaro Saito, Department of Psychology, Keio University, Mita 2-15-45, Minatoku, Tokyo 108-8345, Japan. (E-mail: firstname.lastname@example.org).
|Printer friendly Cite/link Email Feedback|
|Author:||Saito, Kotaro; Watanabe, Shigeru|
|Publication:||The Psychological Record|
|Date:||Sep 22, 2005|
|Previous Article:||Interactions of contingencies, instructional accuracy, and instructional history in conditional discrimination.|
|Next Article:||Are torsos the basis for infants' categorization of cats versus dogs? A reply to Vidic and Haaf (2004).|