The Effect of Cul Length and Hippocampal Lesions on Maze Learning in the Rat PASCHAL N . STRONG, JR.
Psychology Department University of South Florida Tampa, Florida Abstract--Twenty-four rats, 8 with bilateral hippocampal lesions, 8 with cortical lesions and 8 unoperated rats, were tested on one of 2 mazes. Half of each group were run in a conventional 5 choice multiple Y maze. The other half were run in a symetrical, long cul maze in which distance and number of successive choice points were equal for a fight or wrong choice before the animal reached a cul end or the goal box. The correct path in the long cul maze was identical to that of the short cul maze. S's were run one trial a day for 15 days. On the symetrical, long cul maze there were no differences between groups. On the short cul mazej hippocampals were significantly worse than control S's and looked similar to control S's on the long cul maze. The results are interpreted in terms of frustration theory.
HIPPOCAMAL LESIONS CAUSE significant deficits in the learning of various mazes such as the L a s h l e y - I I I maze (Bender, Hostetter and Thomas, 1968; Jackson and Strong, 1969), the Hebb Williams mazes (Kimble, 1963); and multiple T-mazes (Hostetter and Thomas, 1967; Kaada, Rusmussen and Kviem, 1961). Jackson and Strong (1969) showed that the deficit in the Lashley-III maze was due to hippocampal animals' failure to enter the choice doors and not the inability to learn the alternation problem. They further suggested that increasing the cue prominence of the choice doors would improve performance. More recently, Winokur and Breckenridge (1973) tested this hypothesis and found evidence supporting it. Douglas (1967) and Kimble (1968) in reviewing the hippocampal literature point out that hippocampably ablated animals also show increased resistance to extinction, deficits in passive avoidance, and difficulty in learning reversal problems. Jackson and Strong (1969) demonstrated that learning alternation responses is easier for hippocampal animals than normals because of their resistance to extinction during the learning of complex, 3 bar alternation tasks. Recently, Franchina and Brown (1971) also demonstrated that rats with hippocampal lesions fail to respond to shift in reward magnitudes as do normals. While these data have been described in terms of response perseveration we feel that they can Presented in part at the 14th Annual Meeting of the Pavlovian Society, Sarasota, Florida, November 15-16, 1973.
be handled better using a frustration paradigm. We assume that once an animal has learned to expect a particular reinforcement following a given response, then a change in the reinforcement, (either no reinforcement or a reduction in reinforcement magnitude) arouses behaviors, in the presence of cues for the given response, which are imcompatible and thus disruptive of the learned behavior. This frustration paradigm has been used successfully to explain extinction in maze learning situations (Adelman & Maalsch, 1955; Stanley and Rowe, 1954; Kirkpatrick, Pavlik, and Reynolds, 1964), as well as other situations (for example, Holse, 1958). In looking at maze learning, particularly in the case of the multiple-T or multiple-Y maze, we hypothesize that two mechanisms are involved in learning. One is the backward chaining to each choice point of the consequences of reward (the re-so mechanism) and the second is active elimination of errors because of the frustrating effects of entering a cul-de-sac. Once an animal learns that food is available in a maze, entry into a cul-de-sac must be considered as frustrating. Since in most multiple T and Y mazes the animal comes to the non rewarded end of the cul soon after the wrong choice, the cues associated with that choice become aversive via condition of rrsf. Most multiple T and Y mazes, therefore, must be considered to be asymmetrical in that a wrong turn quickly leads to a dead end while a correct turn allows the animal to continue its ongoing movement, a condition reinforcing to the animal. If we made a multiple T or Y maze symmetrical such that the distance at each
0093-2213/78/0100/0246/$00.75 O J. B. Lippincott Company 246
Volume13 Number 4
MAZE LEARNING 1N HIPPOCAMPECTOMIZED RATS
Fzo. !. Floor plan of maze, shaded portion shows extent of short cul maze while the shaded plus the open section indicates the extent of the long cul, symmetrical maze.
choice to the end of the cul is as long as the correct path, and if there were also as many ftirther choice points before the end of the cul is reached as there are before the goal box is reached along the correct path, then it would be much more difficult to eliminate errors via the hypothesized frustration mechanism. This would be particularly true at earlier choice points, since there would be a greater spatial and temporal separation between frustration and the original choice point. We would thus predict that normal animals learning a 5 choice multiple-Y maze of the conventional, asymmetrical type should show faster learning than when learning the same correct choice sequences embedded in a symmetrical maze. If, however, hippocampally ablated animals are deficient in frustrative responses, they should be similar to normals on the symmetrical or long cul maze but inferior to them on the assymmetrical short cul maze. Method
Subjects Twenty-four male, Long Evans hooded rats between the ages of 80-90 days at the beginning of the experiment were used. Eight S's received bilateral hippocampectomies, eight had similar sized lesions of the cortex overlying the hippocampus, and eight were unoperated controls.
Apparatus The basic maze was a multiple choice-Y maze constructed as follows. Individual units were constructed from six pieces of plywood, 6" x 12" x 88 to make a hexagon 12" high. With the exception of the goal box, all hexagons had solid tops and bottoms. Twenty-one individual hexa-
gons were arranged as shown in Figure 1. The walls of adjacent hexagons were 4" apart thus forming a series of Y mazes with 6" runways, 4" wide. The crosshatched segment of Figure 1 shows the short cul, assymmetrical maze, while the crosshatched segment plus the open alleys constitute the long cul symmetrical maze. In the short cul maze, the blind end was around a corner so as to be out of sight at any choice point. There was a guillotine door at the start box and the goal box. The correct path was identical for S's in all conditions. A closed circuit T.V. camera over the maze allowed S's to be observed without E being in the experimental room.
Surgery Surgery was performed using sodium pentobarbital anaesthesia with clean technique. Aspiration was used for all lesions. Hippocampal and cortical S's had a minimum of 2 weeks recovery time before being placed on a deprivation schedule. All S's had recovered and exceeded their preoperative weight by this time.
Behavioral Testing Two weeks following surgery, or at the age of 110-I20 days for the unoperated controls, S's were reduced to 80% of their average body weight for the 3 days before deprivation. A two week period was used to reduce and stabilize their weight during the deprivation phase. Animals were handled every day and during the last three days before maze training were fed wet mash in the goal box for 30". They were then started on maze training receiving one trial a day for 15 days or until they had 3 successive errorless days. Except for the start box, S's were
Pav. J. Biol. Sci. October-December 1978
HIPPOCAMPALS : CORTICAL CONTROL o
NORMALS . . . .
LONG CUL MAZE
ERRORS (GROUP MEANS)
FIG. 2. Mean errors over days (one trial per' day) for each group of rats on the long cul, symmetrical maze.
DAYS 12 -
HIPPOCAMPALS e - - - - - . e CORTICAL CONTROLo
NORMALS . . . . .
ERRORS (GROUP MEANS)
Fio. 3. Mean errors for each group on the short cul, asymmetrical maze.
DAYS allowed to retrace until they reached the goal box. They were given 30" to eat wet mash and then returned to their home cage. S ' s were run in one of three different orders in squads of 3, one from each operative group, to minimize possible o r d e r effects. H a l f o f the animals from each operative group were run in the long cul maze and half in the short cul maze.
Anatomy At the end of the 15 days, lesioned animals were sacrificed by an overdose of sodium pen-
tobarbital and perfused with isotonic saline and 10% formalin. The brains were removed, embedded in gelatin and sectioned on a freezing microtome using 50 um sections. Each tenth section was mounted and stained with Thionin. Lesions were then reconstructed by projecting these slides on appropriate templates and tracing them. Results Analysis of variance of error scores indicates a significant difference between the two m a z e s
Volume 13 Number 4
MAZE LEARNING IN HIPPOCAMPECTOMIZED RATS
(F~.~s = 17.75, p < .001). There is no difference between operation groups on the long cul maze (F2.9 = 0.419) but a highly significant difference between operation groups on the short cul maze (F2..~ = 10.82 p < .01). A test of the means indicates that the hippocampal group is significantly different from the cortical controls and normal groups while these two groups do not differ from each other. Figures 2 and 3 Clearly demonstrate these effects and also show that the hippocampal animals on the short cul maze look almost exactly like the control animals on the long cul maze.
Histology Figure 4 shows the reconstructed lesion of the largest (crosshatched) and smallest (solid black) hippocampal lesions while Figure 5 shows the same data for the cortical controls. As can be seen, the hippocampal lesions were primarily in the anterior 2/3 of the hippocampus.
Discussion The results of this experiment clearly support the hypothesis and predictions detailed in the introduction. Subjectively there was no discernable difference between groups on the long cul maze while on the short cul maze, hippocampal animals looked clearly different. On the long cul maze the distribution of errors across choice points was almost identical for the three groups. In the evolution of the central nervous system there is no area in the reptile brain that looks remotely like the mammalian hippocampus. This structure is, however, highly developed in as primitive a mammal as the oppossum and as the cortex increases in size and importance, so does the hippocampus. We postulate that there is a cortical mechanism which serves as a comparator for evaluating incoming stimuli to expected stimuli. When a match between what is expected and what actually occurs, learned behaviors continue. A partial match leads to some disruption of behavior (generalization decrement) while a gross mismatch (such as no reinforcement when one is expected) causes a message to go via the hippocampus to subcortical reinforcement and arousal mechanisms. This causes disruption of the learned behavior, leading to other behaviors, one of which may be more adaptive. Gray (1970) has shown that the hippocampal-septal area shows rather specific theta responses of 7.5-8.5 Hz in the presence of novelty or frustrative nonreward. Low doses of sodium amobarbital considerably raises the threshold for electrically driving this range of theta but not other theta or nontheta frequencies. These low drug dosages also at-
FIG. 4. Reconstruction of the largest (crosshatched) and smallest (solid black) lesion in the hippocampai group. tenuate the behavioral effects of frustrative nonreward, while leaving other learned behaviors unaffected. Thus, for organisms less dominated by rigid instinctual behaviors, such a mechanism would be highly adaptive to changing environmental conditions. This conception of one of the roles of the hippocampus is as applicable to orienting responses (evaluation of novelty) as to reinforcement expectancies. This role of the hippocampus as a carrier of descending messages which lead to
pay. J. BioLSci.
FIG. 5. Reconstruction of the largest (crosshatched) and smallest (solid black) lesion in the cortical control group.
the increased arousal of frustration adequately covers a wide range o f deficits due to hippocampal damage. It also leads to certain predictions such as no inci'ease in post nonreward running speed in the Amsel and Roussel (1952) type of experiment. This experiment is now being carried out in our laboratory. References
Adelman, H. M., and Maatsch, J. L.: Resistance to extinction as a function of the type of response elicited by frustration. J. Exp. Psychol. 50, 61-65, 1955, Bender, R. M., Hostetter G. and Thomas, G. J.: Effects of lesions in hippocampus-entorhinal cortex on maze performance and activity in rats. Psychonomic Sc. 10, 13-14, 1968. Douglas, R. J.: The hippocampus and behavior. Psychol. Bull. 67, 416-442, 1967. Franchina, J. J. and Brown, T. S.: Reward magnitude shift effects in rats with hippocampal lesions. J. Compar. Physiol. Psychol. 76, 365, 1971. Gray, J. A.: Sodium amobarbital, the hippocampal theta rhythm, and the partial reinforcement extinction effect. Psychol. Rev. 77, 465-480, 1970. Hostetter, G. and Thomas, G. J.: Evaluation of enhanced thigmotaxis as a condition of impaired maze
learning by rats with hippocampal lesions. J. Cornpar. Physiol. Psychol. 63, 105-110, 1967. Hulse, S. H.: Amount and percentage of reinforcement and duration of goal confinement in conditioning and extinction. J. Exp. Psychol. $6, 48-57, 1958. Jackson, W. J., and Strong, P. N.: Differential effects of hippocampal lesions on sequential tasks and maze learning by the rat. J. Compar. Psysiol. Psychol. 68, 442-450, 1969. Kaada, B. R., Rasmussen, E. W. and Kviem, O.: Effects of hippocampal lesions on maze learning and retention in rats. Esp. Neurol. 3, 333-335, 1961. Kimble, D. P.: The effects of bilateral hippocampal lesions in rats. J. Comp Physiol. Psychol. 56, 273283, 1963. Kimble, D. P.: Hippocampus and internal inhibition. Psychol. Bull. 70, 285-295, 1968. Kirkpatrick, D. R., Pavlik, W. B. and Reynolds, W. F.: partial reinforcement extinction effects as a function of size of goal box. L Exp. Psychol. 48, 271-274, 1954. Stanley, W. C. and Rowe, M. I.: Extinction by omission of food as a function of goal box confinement. J. Exp. Psychol. 48, 271-274, 1954. Winokur, G. and Breckenridge, C. D.: Cue-dependent behavior of hippocampally damaged rats in a complex maze. J. Compar. Physiol. Psychol. 82, 512522, 1973.