On the Neural Bases of the Spatial Mapping ... - Semantic Scholar

HIPPOCAMPUS, VOL. 1, NO. 3, PAGES 236-239, JULY 1991

On the Neural Bases of the Spatial Mapping System: Hippocampus vs. Hippocampal Formation Leonard E. Jarrard D e p a r t m e n t of P s y c h o l o g y , W a s h i n g t o n a n d Lee University, L e x i n g t o n , VA 24450 U . S . A . As pointed out by Lynn Nadel in the lead article of this forum, since Thr Hippocumpus ns CI Cognitive M a p first appeared in 1978 many of the published studies have been concerned in some way with comparing the spatial mapping hypothesis with rival theoretical formulations. Basing their interpretations primarily on the results of lesion studies, investigators have suggested that the hippocampus is crucially involved in processes like working memory (Olton et al., 1979), the storage of intermediate-term memory (Rawlins, 1985), and configural associations ( Sutherland and Rudy, 1989). Why has it been so difficult to obtain convincing evidence 10 support one theory of hippocampal function to the exclusion of the others‘? While there are several reasons that could be given, most of them center around a failure on the part of many investigators to take into account the relevant neuroanatomy. Thus, one problem has been the confusion resulting from failing to always make a clear-cut distinction between the hippocampus and the hippocampal formation (see below). A second problem has been the variety of effects on behavior that are found when different structures and/or areas within the hippocampal formation are damaged. This is not surprising, since conventional lesion techniques (aTpiration, electrolytic, radio frequency) cause damage not only to cells in the area of interest, but usually to adjacent structures and the vasculature, as well as interrupt fibers of passage. In this essay, consideration will be given first to the relevant neuroanatomy. Selective hippocampal lesions obtained with neurotoxins will then be described. This will be followed by a brief description of the results obtained in a number of experiments designed to study the involvement of different components of the hippocampal formation in the processing of spatial information.

NEUROANATOMICAL CONSIDERATIONS A N D THE SELECTIVE LESION APPROACH The hippocampal formation is considered by most neuroanatomists to be comprised of the entorhinal cortex, dentate gyrus, hippocampus proper, and subiculum (Amaral and Witter, 1989). The term “hippocampus” is used to refer to the pyramidal cell fields of the hippocampus proper (CAI CA3), together with the hilar cells and the dentate gyrus (see Gottlieb and Cowan, 1973). Within the hippocampal formation there are two main interconnecting circuits that are especially relevant. One of the best understood, and probably

most important, is the intrahippocampal circuit, consisting of major projections from entorhinal cortex via the perforant path to dentate gyrus, through hippocampus, and to the subiculum (Amaral and Witter, 1989). The second circuit consists of direct projections from entorhinal cortex to the subiculunl and CA1 cell field, thus bypassing most of the hippocampus (Van Groen and Lopes da Silva, 1986; Witter et al., 1989). A consideration of relevant hippocampal formation afferents indicates that cortical inputs are primarily to the entorhinal cortex. The main subcortical projections being to hippocampus (with some to subiculum) from medial septum, diagonal band, and brainstem. Major efferents include entorhinal cortex projections back to all cortical areas that send projections: hippocampus projections to the septa1 region; subicular subcortical projections, by way of fimbria-fornix. to basal forebrain, thalamic and hypothalamic areas; and subicular-cortical projections to cingulate gyrus, entorhinal, and perirhinal areas (Swanson, 1979; Kohler, 1985: Witter et al., 1989). Givcn the complexity of the underlying neuroanatomy, one can see why damage produced by various conventional lesion techniques may be quite variable and produce changes in behavior that are difficult to interpret. For example, electrolytic and aspiration lesion approaches have been used to damage the hippocampus in many studies. but these surgical techniques interrupt extrahippocampal fibers of passage in alve~is and fimbria that run in both directions, cause extensive damage to the vasculature, and usually cause direct damage to the subiculum. Interruption of the axons in fimbria-fornix has been used to study functions of the hippocampus, yet this lesion not only affects many extrahippocampal fibers of pacsage but also spares the major output pathway from hippocampus that projects in a caudal direction toward subiculum and entorhinal cortex. In thc case of entorhinal cortex lesions, most investigators have uscd electrolytic o r radio-frequency lesions, resulting in extraentorhinal damage that includes the subiculum and, usually, damage to the caudal dentate gyrus. It is for the above reasons that we have devised surgical procedures in the rat that permit more selective removal of discreet, anatomically defined components of the hippocampal formation. By using multiple injections of small amounts of ibotenic acid (IBO), it is possible to remove the cells in the hippocampus and the subiculum with minimal damage to adjacent structures and/or fibers-of-passage, and without secondary damage to other brain structures (Jarrard, 1989). As a result, many of the problems associated with the use of electrolytic, aspiration, and radio-frequency lesion techniques are avoided.

SELECTIVE HIPPOCAMPAL FORMATION LESIONS A N D BEHAVIOR The selective ibotenate lesions described above have been used, together with conventional lesions, in several behavioral testing situations designed to study the involvement of the different components of the hippocampal formation in the processing of spatial and nonspatial information.

236

NEURAL BASES OF THE SPATIAL MAPPING SYSTEM / Jarrard O n the utilization of spatial and intramaze cue information: Preoperative vs. postoperative acquisition In a number of experiments we have employed complex place and intramaze cue memory tasks using an eight-arm radial maze together with a within-subjects design that involves two kinds of learning (place and cue) and two memory functions (working memory and reference memory) (Jarrard, 1983). In the place task the arms are similar, and four of the eight arms are consistently baited from trial to trial. In the intramaze cue task there are inserts of different materials (sandpaper, rubber matting, screen wire, etc.) that fit into the arms and change locations from trial to trial, and four cues are consistently baited. Using the terminology of Honig (19781, the limited baiting procedure permits a determination of both working memory (WM) (information that is useful for a relatively short time and is applicable to only a single trial) and reference memory (RM) (information that is applicable across many trials). Originally, we designed these testing procedures in order to compare and contrast the spatialicognitive map theory of O’Keefe and Nadel (1978) with the working memory theory of Olton et al. (1979). A variety of hippocampal formation lesions have now been used (both neurotoxin and conventional lesions) with the place and cue memory tasks, together with preoperative and postoperative training. The main results of these studies can be summarized as follows. (1) When the hippocampus has been selectively removed with IBO before acquisition, rats are unable t o learn the place task (making both RM and WM errors), but acquisition of the intramaze cue task is like that of controls (Jarrard, 1986; Leu et al., 1987). (2) If rats learn the tasks before the hippocampus is removed, postoperative testing indicates a slight, temporary impairment on the place task but normal performance on the cue task. Similar minimal effects on retention are found following removal of cells in the dentate gyrus with colchicine, CA3 cells with kainic acid, and CAI cells with IBO (Jarrard, 1983; 1986; Jarrard et al., 1984; Jarrard and Johnson, 1988). (3) Ibotenate lesions of subiculum cause a slight impairment in retention of the preoperatively learned place task but not the cue task (Jarrard, 1986). (Postoperative acquisition has not been studied in subiculum lesioned rats.) (4) Hippocampal lesions made with conventional aspiration and electrolytic approaches result in an inability both to learn the place task and to perform the place task, even with preoperative training; there is also a disruption of WM in performance of the cue task (Jarrard, 1986; Okaichi and Oshima, 1990). (5) Electrolytic lesions of the fimbria-fornix, medial septum, entorhinal cortex, and combined ibotenate lesions of hippocampus plus subiculum all impair retention of the preoperatively learned place task (resulting in both RM and WM errors). In addition, there is a separate, selective impairment in WM on the cue task similar to what is reported by Olton et al. (1979) in their spatial, radial-arm maze. In a test of the generality of the above findings obtained with the radial maze, we have recently completed a study that involved acquisition and postoperative retention of allocentric spatial learning in the open-field Morris water maze

237

(Morris et al., 1990). Both ibotenate hippocampal and subiculum lesions were found to result in impairments in initial learning of the place task, but with overtraining both groups eventually learned to find the safe platform. When the damage was more extensive and involved bath hippocampus and subiculum, overtraining did not result in significant learning. In rats trained before the hippocampus was removed, postoperative performance was impaired, but on probe trials the animals were able to demonstrate that spatial information regarding the location of the safe platform was retained-the problem appeared to be one of organizing their responses in order to reach the safe platform, that is, a problem in navigation. Forgetting of spatial vs. intramaze cue information Since the hippocampus is so often implicated as playing an important role in memory, and the results summarized above suggest a special involvement in acquisition and utilization of spatial information, we have designed experiments to study the forgetting of spatial as compared to nonspatial information. The basic task consists of an eight-arm radial maze, together with a win-stay procedure similar to the one described by Collier et al. (1982). In the spatial task a single arm is designated as being correct within a day and, following an information trial (the door to the correct arm is raised), the rat is given five trials in which the same arm is repeatedly baited and the rat must choose from among all eight arms. While the same arm is correct within a day, different arms are correct across days. Even though rats with ibotenate hippocampal lesions made more errors than controls and subiculum-lesioned rats in learning the basic task, hippocampals did learn this relatively simple spatial task eventually, and performance did not differ by the end of acquisition. Forgetting of spatial information was next studied by inserting delays of 0, 5 , 20, and 60 minutes between trials 4 and 5. Performance of hippocampal and control rats was similar at the 0-minute delay, but there was significant forgetting in hippocampal rats after 5 minutes and progressively greater forgetting with increasing delays (Jarrard et al., 1987). It is interesting that rats with subiculum lesions tended to make fewer errors than controls during both acquisition and tests for forgetting (the differences were not statistically significant). In order to see if animals without a hippocampus would experience accelerated forgetting of nonspatial, intramaze cue information, we have attempted to train rats on a comparable win-stay cue task using the eight-arm radial maze. In this task, cue inserts of different materials were placed in the arms and the locations of the cues were shifted between trials. A single cue was correct on all trials within a day but different cues were correct across days. As one might predict, even control rats had trouble learning the eight-arm, winstay, intramaze cue task. In a second study, task difficulty was decreased by using a six- rather than an eight-arm radial maze. Most rats learned the win-stay cue task with six arms. and, following removal of hippocampus with ibotenate injections and retraining, tests of forgetting (0-, 5-, 20-, 60-minute delays inserted between trials 4 and 5 ) were administered.

238

HIPPOCAMPUS VOL. I , NO. 3, JULY 1991

There was little forgetting of the intramaze cue information in either hippocampals or controls even with delays of 60 minutes (unpublished observations). The results of the spatial and cue forgetting studies seem to implicate the hippocampus as playing an important role in the forgetting of spatial but not cue information; however, further experiments are needed in which more similar procedures for the place and cue tasks are employed. O n the use of allocentric vs. taxon strategies in rats with hippocampal lesions

An unexpected result was obtained in an experiment designed to determine if removing the C A I pyramidal cells would cause a severe impairment in spatial learning similar to that reported following ischemic brain damage (Volpe et a l . , 1987). We used intrahippocampal injcctions of IHO aimed at the C A I field and tested the rats on the place and cue, limited baited, eight-arm radial maze tasks described above. Even though the loss of C A I cells was greater than that reported by Volpe et al. (1987) with ischemia, the effects on learning were minor and temporary. Specifically. rats with the CAI cells removed were slightly impaired early in acquisition on the place task (making both RM and WM errors). but learning of the cue task was jircilitritrd (Jarrard and Johnson, 1988). This may seem like a strange pattern of results but, in fact, O’Keefe and Nadel (1978) point out that the behavior of animals without a hippocampus (and without a locale system) would be largely determined by remaining brain properties (among them being the taxon system). Thus, it would appear that rats with the hippocampus removed were impaired in their use of the spatial cues because of an impaired locale strategy but employed a taxon strategy more readily than controls in the intramaze cue task. In a further test of the use of locale vs. taxon strategies, we have employed a 12-arm radial maze where six similar arms are designated place arms and six cue arms (M’Harzi and Jarrard, 1988). Three of the place arms are consistently baited over trials, and three cues are consistently baited (although the three cues change locations within the six cue arms over trials). As unoperated rats learn this task, the order of choosing arms changes from equally selecting place and cue arms to where with further training, the place arms are chosen first before cue arms. Thus, after training a place (locale) strategy is utilized first, followed by the usually more difficult cue (taxon) strategy. Preliminary data obtained from rats with hippocampal formation lesions indicate a preference for choosing cue arms over place arms, thus suggesting a predominant taxon strategy (unpublished observations).

CONCLUSIONS The results of our studies to date comparing the effects of the more selective neurotoxin removal of cells in the hippocampus with those obtained following conventional lesions that cause more extensive damage to hippocampal formation lead to the following general conclusions. First, the hippocampus seems to play an important role in the acquisition and utilization of spatial (but not cue) information. However, if the spatial information is acquired before the hippocampus

is removed, the information can be retrieved and utilized in subsequent performance, thus indicating that storage of spatial information is located outside the hippocampus. Second, more extensive lesions that involve damage to several hippocampal formation structures and related areas show that both pre- and postoperative acquisition of spatial information is impaired, while equivalent nonspatial information is minimally affected. Further, the results indicate that with the more extensive damage to hippocampal formation there is an additional impairment in WM. Our program of research has been “driven” by the underlying neuroanatomy of the hippocampus and related hippocampal formation structures, rather than being primarily concerned with the testing of existing theories of hippocampal function. However, as is apparent from the above brief summary, the pattern of behavioral results does not support most current theories. Contrary to the WM theory of Olton et ai. (IY79), WM is not selectively impaircd on spatial and nonspatial tasks when the damage is limited to hippocampus. Rather than being a “high-capacity intermediate-term memory store” for all types of memory, as proposed by Rawlins (1985), we find that cue (but not spatial) information is retained as well in hippocampal rats as controls even for periods as long as 60 minutes. In contrast with predictions of the configural association theory of Sutherland and Rudy ( 1989), acquisition and postoperative retention of complex, “ifthen” conditional discriminations is not impaired if the damage is limited to hippocampus, but impairments are Found if the damage is more extensive and involves several hippocampal formation structures (Jarrard and Davidson, 1991). Finally, our findings come closer to agreeing with the predictions of thc spatialicognitive mapping theory of O’Keefe and Nadel(1978). except that rats without a hippocampus are able, eventually, to learn a complex place task given sufficient overtraining (Morris et al., 1990) and can perform a complex place task if learning occurs before the hippocampus I S removed (Jarrard, 1986). So, “Is the hippocampal formation preferentially involved in spatial behavior?” If one is talking about the hippocurnpid formation it would appear from our findings and those reported by others that there is a preferential involvement in spatial behaviors-but it seems that other processes (like WM) may also be affected. If one is asking the question as it relates more specifically to the hippocrimpus, the existing data obtained with more selective hippocampal lesions indicate an even more clear-cut involvement in spatial behaviors without other processes being affected. Given the increased interest in computational modeling of the structure and function of the hippocampus, and continuing attempts to devise theories of hippocampal function, it seems important for those working in the area to make a clear-cut distinction between damage that is limited to the hippocampus and more extensive damage that includes hippocampus and related hippocampal formation structures. With greater concern for the underlying neuroanatomy and the nature and extent of the damage, and increased use of the convergent operations called for by Nadel in the lead article, it should be possible to obtain a better understanding of exactly how the hippocampus is involved in behavior.

NEURAL BASES OF THE SPATIAL MAPPING SYSTEM / Jarrard

ACKNOWLEDGMENTS T h e research described in this essay was supported by NSF grants BNS 85-07259 a n d 88-09208.

References Amaral, D. G., and M. P. Witter (1989) The three-dimensional organization of the hippocampal formation: A review of anatomical data. Neuroscience 31571-591. Collier, T. J., J . S. Miller, J. Travis, and A. Routtenberg (1982) Dentate gyrus granule cells and memory: Electrical stimulation disrupts memory for places rewarded. Behav. Neural Biol. 34:227-239. Gottlieb, D. I., and W. M. Cowan (1973) Autoradiographic studies of the commissural and ipsilateral association connection of the hippocampus and dentate gyrus of the rat. I . The commissural connections. J. Comp. Neurol. 149:393-422. Honig. W. K. (1978) Studies of working memory in the pigeon. In Cognitive Aspects ofAnirna1 Behavior, S. H. Hulse, H. F. Fowler, and W. K. Honig, eds., pp. 211-248. Erlbaum, Hillsdale, NJ. Jarrard, L. E. (1983) Selective hippocampal lesions and behavior: Effects of kainic acid lesions on performance of a place and cue task. Behav. Neurosci. 972373-889. Jarrard, L. E. (1986) Selective hippocampal lesions and behavior: Implications for current research and theorizing. In The Hippocumpus, Vol. 4, R. L. Isaacson and K. H. Pribram, eds., pp. 93126, Plenum, New York, NY. Jarrard, L. E. (1989) On the use of ibotenic acid to lesion selectively different components of the hippocampal formation. J . Neurosci. Methods 29:251-259. Jarrard, L. E., and T. L. Davidson (1991) On the hippocampus and learned conditional responding: Effects of aspiration versus ibotenate lesions. Hippocampus 1: 107-1 17. Javard, L. E.,and L.S. Johnson (1988) Effects of ibotenate lesions of hippocampal CAI cells on acquisition and retention of a complex place and cue memory task. Paper presented at the meeting of the European Brain and Behavior SocietyiEuropean Neuroscience Society, Zurich. Jarrard, L. E., H. Okaichi, 0. Steward, and R. Goldschmidt (1984) On the importance of the dentate gyrus and hippocampal connections in the performance of a complex place and cue task. Behav. Neurosci. 98:946-954. Jarrard, L. E., J. M. Ringland, and L. S. Johnson (1987) Increased

239

rate offorgetting of spatial information in rats with ibotenate lesions of hippocampus. Soc. Neurosci. Abstr. 13:802. Kohler, C. ( 1985)Intrinsic projections of the retrohippocampal region in the rat brain. I . The subicular complex. J . Comp. Neurol. 236:504-522. Leu, J . , L. S . Johnson, and L. E. Jarrard (1987) Effects of ibotenic acid lesions of the hippocampus on acquisition of a radial maze memory task. Soc. Neurosci. Abstr. 13:646. M’Harzi, M., and L. E . Jarrard (1988) Place and cue learning, reference and working memory: A new procedure for use in neurobehavioral research. Paper presented at the meeting of the European Brain and Behavior Society/European Neuroscience Society, Zurich. Morris, R. G. M., F. Schenk, F. Tweedie, and L. E. Jarrard (1990) Ibotenate lesions of hippocampus and/or subiculum: Dissociating components of allocentric 5patial learning. Eur. J . Neurosci. 2: 10 16-1028. Okaichi, H., and Y . Oshima (1990) Choice behavior of hippocampectomized rats in the radial arm maze. Psychobiology 18:416-421. O’Keefe, J . , and L. Nadel(l978) Hippocornpus as u Cognirive M a p . Oxford University Press, London. Olton, D. S., J. T. Becker, and G. E. Handelmann (1979) Hippocampus, space, and memory. Behav. Brain Sci. 2:313-365. Rawlins. J. N. P. (1985) Associations across time: The hippocampus as a temporary memory store. Behav. Brain Sci. 8:479-496. Sutherland, R. J., and J. W. Rudy (1989) Configural association theory: The role of the hippocampal formation in learning, memory, and amnesia. Psychobiology 17: 129-144. Swanson, L. W. (1979) The hippocampus: New anatomical insights. Trends Neurosci. 2:9-12. Van Groen, T., and F. H. Lopes da Silva (1986) The organization of the reciprocal connections between the subiculum and the entorhinal cortex in the cat. 11. An electrophysiological study. J . Comp. Neurol. 251:111-120. Volpe, B. T.,B. Waczek, P. Colombo, and J. P. Davis (1987) Ischemic induced hippocampal injury in rats causes dissociated working memory impairment on modified T-maze. Soc. Neurosci. Abstr. 13:802. Witter. M. P., J . J . Groenewegen, F. H. Lopes da Silva, and A. H. M. Lohman (1989) Functional organization of the extrinsic and intrinsic circuitry of the parahippocampal region. Prog. Neurobiol. 33: 161-253.