Bilateral Hippocampal Pathology Impairs Topographical and Episodic Memory (PDF)

HIPPOCAMPUS 11:715–725 (2001) Bilateral Hippocampal Pathology Impairs Topographical and Episodic Memory But Not Visual Pattern Matching Hugo J. Spiers,1* Neil Burgess,1 Tom Hartley,1 Faraneh Vargha-Khadem,2 and John O’Keefe1 1 Institute of Cognitive Neuroscience and Department of Anatomy and Dev...

HIPPOCAMPUS 11:715–725 (2001) Bilateral Hippocampal Pathology Impairs Topographical and Episodic Memory But Not Visual Pattern Matching Hugo J. Spiers,1* Neil Burgess,1 Tom Hartley,1 Faraneh Vargha-Khadem,2 and John O’Keefe1 1 Institute of Cognitive Neuroscience and Department of Anatomy and Developmental Biology, University College London, London, UK 2 Developmental Cognitive Neuroscience Unit, Institute of Child Health, University College London, London, UK ABSTRACT: A virtual reality environment was used to test memory addition of verbal and temporal inputs. This hypothesis performance for simulated “real-world” spatial and episodic information predicts that hippocampal damage in humans should im- in a 22-year-old male, Jon, who has selective bilateral hippocampal pa- thology caused by perinatal anoxia. He was allowed to explore a large- pair both topographical and episodic memory. scale virtual reality town and was then tested on his memory for spatial Both topographical and episodic memory can be layout and for episodes experienced. Topographical memory was tested tested in many ways. Topographical memory was previ- by assessing his ability to navigate, recognize previously visited locations, ously evaluated by observation of route learning (Habib and draw maps of the town. Episodic memory was assessed by testing the and Sirigu, 1987; Katayama et al., 1999), landmark rec- retrieval of simulated events which consisted of collecting objects from characters while following a route through the virtual town. Memory for ognition (Whiteley and Warrington, 1978; Incisa della the identity of objects, as well as for where they were collected, from Rochetta et al., 1996; Maguire et al., 1996a; Katayama et whom, and in what order, was also tested. While the first task tapped al., 1999), route learning on a tabletop maze (Semmes et simple recognition memory, the latter three tested memory for context. al., 1955; Milner, 1965; Bottini et al., 1990; Katayama et Jon was impaired on all topographical tasks and on his recall of the context-dependent questions. However, his recognition of objects from al., 1999), the ability to describe common routes (Incisa the virtual town, and of “topographical” scenes (as evaluated by standard della Rochetta, 1996; Teng and Squire, 1998), view dis- neuropsychological tests), was not impaired. These findings are consistent crimination of the same building (Whiteley and War- with the view that the hippocampus is involved in navigation, recall of long rington, 1978, Suzuki et al., 1998), orientation and dis- term allocentric spatial information and context-dependent episodic mem- tance judgements (Maguire et al., 1996a, Teng and ory, but not visual pattern matching. Hippocampus 2001;11:715–725. © 2001 Wiley-Liss, Inc. Squire, 1998), and map drawing (Habib and Sirigu, 1987; Bottini et al., 1990; Maguire et al., 1996a; Suzuki KEY WORDS: hippocampus; navigation; spatial memory; recognition; et al., 1998). Small-scale tasks such as remembering the virtual reality spatial location of an object or stimulus on a tabletop or a display have also been applied to test allocentric spatial memory (Smith and Milner, 1981; Morris et al., 1996; Abrahams et al., 1997; Nunn et al., 1998; Bobhot et al., INTRODUCTION 1998; Holdstock et al., 2000). Episodic memory has also been tested by a number of The human hippocampus has long been associated with episodic memory methods. One method is to test the ability to remember (Scoville and Milner, 1957), while the hippocampus in rodents has been autobiographical events that occurred prior to the lesion associated with spatial navigation (O’Keefe and Nadel, 1978). O’Keefe and (e.g., Sanders and Warrington, 1975). Another method Nadel (1978) proposed that a possible link between topographical and epi- is to test the recall or recognition of a list of words, set of sodic memory is the existence of an allocentric (world-centered) cognitive pictures (e.g., Warrington and Weiskrantz, 1968), paired map, stored in the hippocampus. They suggested that a spatial system in rats associates (Meyer and Yates, 1955), or prose passages might have developed into an episodic memory system in humans with the (Milner, 1958). While these tests have been successful in demonstrat- ing that topographical and episodic memory are complex Grant sponsor: Medical Research Council, UK. processes that involve many brain regions, and not just *Correspondence to: Hugo J. Spiers, Institute of Cognitive Neuroscience, 17 Queen Square, London WC1N 3AR, UK. E-mail: [email protected] the hippocampus (see Aguirre and D’Esposito, 1999), it Accepted for publication 4 April 2001 is not clear how these tests map onto the real-world be- haviors of actually navigating or remembering a real © 2001 WILEY-LISS, INC. DOI 10.1002/hipo.1087 716 SPIERS ET AL. event. Navigation requires more than solely the recognition of Following a parental questionnaire, three main areas of memory landmarks or the judgment of distance and angle. Equally, episodic problems were identified in Jon by Vargha-Khadem et al. (1997): memory can be distinguished from both semantic and recognition 1) spatial: Jon is unable to find his way in familiar surroundings, memory by the additional ability to retrieve the rich spatiotempo- remember where objects and belongings are usually located, or ral context of events (Tulving, 1983). Furthermore, testing pa- remember where he has placed them; 2) temporal: Jon is not well- tients with lesions of more than one structure, such as temporal oriented in date or time, and he must frequently be reminded of lobectomy patients, does not allow us to accurately identify the regularly scheduled appointments and events, such as particular structure responsible for a selective impairment (e.g., hippocampus classes or extracurricular activities; and 3) episodic: Jon is unable to vs. other medial temporal lobe structures). provide a reliable account of the day’s activities or reliably remem- Testing the actual navigation performance and retrieval of con- ber telephone conversations or messages, stories, television pro- text-rich episodic memory in controlled conditions is made possi- grams, visitors, holidays, and so on. ble by the use of virtual reality. Whereas real-world environments and events are difficult to control experimentally, virtual environ- ments allow controlled simulated events to happen, and actions or movements to be measured accurately. Simulated virtual reality NEUROPATHOLOGY environments have been used in a number of functional imaging studies examining topographical memory, showing the involve- ment of brain regions consistent with those found in clinical stud- Three methods were used to assess Jon’s neuropathology ies (Aguirre et al., 1996, 1998; Maguire et al., 1998a,b; Gron et al., (Vargha-Khadem et al., 1997). These were MRI volumetric mea- 2000). Two of these studies (Maguire et al., 1998b; Gron et al., surements (Van Paesschen et al., 1997), T2 relaxometry (Jackson 2000) imaged active navigation and found participation of both et al., 1993), and proton magnetic resonance spectroscopy (1H the right and left hippocampus. Maguire et al. (1998b) found that MRS, Gadian, 1995). These techniques were selected because of blood flow changes in the right hippocampus were related to ac- their sensitivity to temporal-lobe pathology. From volumetric curacy of navigation: the more accurate the performance, the more measurements, Jon’s hippocampi were found to be bilaterally active the right hippocampus. To our knowledge, neurological shrunken by approximately 50%. His hippocampal volume was patients have not been tested on their ability to navigate in a virtual found to be reduced along the length of both hippocampi, as reality environment or to remember events occurring within them. indicated by cross-sectional areas (Fig. 1). Furthermore, T2 relax- In this study, we tested the ability of a patient with early-onset ometry, which provides a means of quantifying abnormalities that selective bilateral hippocampal pathology to navigate, and to re- are responsible for signal increases seen on T2-weighted MRI member topographical information and different aspects of epi- (Jackson et al., 1993), showed elevated T2 values bilaterally, sug- sodic memory within a large-scale virtual environment. The pa- gesting that the remaining hippocampal tissue in Jon is compro- tient, Jon, is one of a group of patients previously described by mised. Finally, 1H MRS provides a noninvasive method of detect- Vargha-Khadem et al. (1997) and Gadian et al. (2000). ing diffuse temporal lobe pathology by examining the ratio of certain metabolites within the brain, such as N-acetylaspartate (NAA) to creatine plus phosphocreatine (Cr), and choline-con- taining compounds (Cho). A reduction in the ratio of the NAA ETIOLOGY AND SYMPTOMS signal to the Cr and Cho signals is commonly interpreted as a reflection of neuronal loss or damage. Spectra were obtained from Jon was born prematurely at 26 weeks of gestation, weighting a 2 ⫻ 2 ⫻ 2 cm volume within the medial region of the temporal 940 g and suffering from breathing difficulties that required intu- lobes, encompassing a small portion of the hippocampus. Al- bation at 15 min, although spontaneous breathing was established though this selected region of the temporal lobe contains a contri- 30 min later. Despite some brief apnoeic attacks, Jon did well until bution from hippocampal tissue, this contribution is so small that age 3 weeks, when more severe episodes of apnoea occurred requir- any spectral changes are considered to arise predominantly from ing intubation and positive pressure ventilation for 1 week. At this extrahippocampal tissue, i.e., the spectral abnormalities reflect rel- time, he was suspected to have enterocolitis and he suffered from atively diffuse or widespread pathology that is additional to any multiple episodes of severe apnoea, again requiring intubation and pathology detected on volumetric or T2 measures of the hip- positive pressure ventilation. He was transferred to an intensive pocampus (Gadian et al., 1999). The 1H MRS values obtained for care unit for a period of 3 weeks, after which he improved steadily Jon were in the normal range on the left and marginally below on and encountered no further medical problems. His developmental the right, suggesting that the extrahippocampal regions sampled milestones were slightly delayed for walking, but speech and lan- were largely preserved. guage functions emerged normally. At age 3 years and 10 months, A more recent analysis using voxel-based morphometry involv- Jon suffered an unconfirmed convulsive episode in association ing Jon and four other patients who had suffered perinatal or with a cold. From an early age he was considered clumsy, although infantile hypoxic-ischemic episodes further confirmed that within he developed no other motor abnormalities. Memory problems the medial temporal lobes, the damage is confined to the hip- were first noted by Jon’s parents when he was about 5– 6 years old. pocampus (Gadian et al., 2000). Outside the medial temporal  ____________________________________________ HIPPOCAMPUS, SPACE, AND EPISODIC MEMORY 717 individual, and a story, and to recognize line drawings of objects and faces previously seen. Jon correctly remembered only 3 out of 12 items, which is indicative of severe impairment. In addition to the tests described above, 12 computerized, two- choice recognition tests were developed. These tests consisted of three types: one-trial recognition of lists of items, one-trial associa- tive recognition of lists of paired items, and multitrial associative recognition of lists of items. Stimuli for both the one-trial recog- nition and one-trial associative recognition consisted of words, nonwords, familiar faces, and unfamiliar faces. One-trial recogni- tion tests consisted of five lists of 12 sequentially presented items each; the presentation of each set was followed by forced-choice recognition of the familiar items, with each item paired with a novel distractor. One-trial associative recognition tests consisted of 10 lists of six sequentially presented pairs of items each, followed by re-presentation of one item and forced-choice recognition of its FIGURE 1. Hippocampal cross-sectional area as a function of slice position, sectioned posterior to anterior. Connected lines are associate or an item from a different pair. In the four multitrial Jon’s cross-sectional hippocampal area (right hippocampus is darker- associative recognition tests, the stimuli consisted of 20 pairs each shaded). Dashed lines are 2 standard deviations above and below the of nonwords, unfamiliar faces, voices, and faces, or objects and mean hippocampal cross-sectional area of a group of 22 normal locations. Multitrial tests involved one presentation of the list of healthy subjects (Van Paesschen et al., 1997). Cross-sectional areas stimuli followed by successive recognition trials with feedback un- are uncorrected for intracrainial volume (see Van Paesschen et al., 1997). til a criterion (18/20 correct) was reached or 10 trials had been completed. Jon was not significantly worse than 11 control sub- jects on any of the recognition tests except the voice-face and lobes there was reduced grey-matter density in the putamen and object-location tasks, where his performance was equal to that of the ventral thalamus bilaterally. the worst control subject on the voice-face task but poorer than the worst control subject on the object location task (Vargha-Khadem et al., 1997). Jon’s recognition memory was studied further on two recent PREVIOUSLY REPORTED occasions. When tested on the Doors and People Test (Baddeley et NEUROPSYCHOLOGY AND al., 1994), his scores were within normal range on both verbal and EXPERIMENTAL TESTING visual recognition subtests but severely impaired on the recall subtests (Baddeley et al., 2001). When Jon’s event-related poten- When tested in 1996 using the Weschler Adult Intelligence tials (ERPs) were measured during the recognition of previously Scale-Revised, Jon’s verbal IQ was 108 and his performance IQ studied words, he was found to be lacking the late positive com- was 120. He was severely impaired on a range of tests of delayed ponent (Duzel et al., 1999a), an ERP index thought to be associ- recall. On the Rey-Osterrieth Complex Figure Test, he recalled ated with recollection, a fundamental component of episodic only two identifiable fragments (maximum, 18) of the figure after memory (Duzel et al., 1999b). By contrast, his ERP index of fa- a 40-min delay, and on the 90-min delayed recall of the Logical miliarity, a basic component of semantic memory, was well-pre- Memory Subtest of the Weschler Memory Scale (Wechsler, 1947), served. he recalled only 10% of the stories. On the Children’s Auditory Verbal Learning Test (Tally, 1993), he obtained a standard score of 72, close to the minimum possible of 60. However, his digit span was 7 forward and 6 backward, and his Corsi block span was 7 MATERIALS AND METHODS forward and 8 backward, performing better than the average of a group (n ⫽ 47) of normal controls (Vargha-Khadem et al., 1997). Standard Neuropsychological Tests Using a parental questionnaire (Sunderland et al., 1983) and the Rivermead Behavioural Memory Test (Wilson et al., 1985), his Jon was 22 years old when tested in the present study. Informed anecdotally described problems with spatial, temporal, and epi- consent for participation was obtained in accordance with guide- sodic memory problems were documented and tested more for- lines set by the Great Ormond Street Hospital for Children and the mally. In the parental questionnaire, Jon’s parents had to rate how Institute of Child Health Ethics Committee. His general intellect often he forgot 29 typical everyday events. The ratings were at the was assessed using Raven’s Advanced Progressive Matrices, Set I extreme end of the scale, i.e., often forgetting more than once a (Raven, 1976). Mental rotation was evaluated by the Little Man day. The Rivermead Behavioural Memory Test requires the sub- Test (Ratcliff, 1979). In this test, the subject is shown a sequence of ject to remember a route through a room, where a belonging was 32 drawings of a man holding a white ball in one hand and a black placed, the date, a message to be delivered, a name for a pictured ball in the other hand. The man is either upright facing the subject, 718 SPIERS ET AL. TABLE 1. Results of Standard Neuropsychological tests Raven’s Advanced Progressive Matrices Set I Jon 13 Control Subjects Mean (Standard Deviation) 11/12 9.5 (1.7)/12 The Little Man Test Jon Control data from Ratcliff, 1979 Total 26/32 27/32 Upright 13/16 14/16 Inverted 13/16 13/16 The Camden Memory Test Jon Percentile Interpretation Topographical Recognition Memory Test 25/30 25th–50th Average Pictorial Recognition Memory Test 30/30 100th Superior Paired associates First test 16/24 10th–25th Low-average Second test 16/24 ⬍5th Impaired upright facing away, inverted facing the subject, or inverted facing gess et al., 2001). The same town was used in Spiers et al. (2000), away. For each picture the subject has to say which hand is holding and is a modification of that used in Maguire et al. (1998b). The the black ball. The Camden Topographical Recognition Memory town consisted of a main street intersected by a cross-road. Test (CTRMT), the Camden Pictorial Recognition Memory Test Throughout the town there were various interior locations which (CPRMT), and the Camden Paired Associates Memory Test included a cinema, an arcade, a book shop, a bar, a sushi bar, an (CPAMT) from the Camden Memory Test Battery (Warrington, 1996) were used to provide standardized measures of recognition and recall at the time of the present testing. The Camden Topo- graphical Recognition Memory Test involves a three-way forced- choice recognition test of previously presented photographs of scenes. The Pictorial Recognition Memory Test follows a similar format to the CTRMT, but the stimuli consist of people, animals, and general objects, as well as places. The Camden Paired Associ- ates Memory Test consists of three sets of eight word pairs; each word pair is read aloud during the presentation. After the presen- tation of eight words, recall of the pairs is tested by the presentation of a single word from each pair. Feedback in the form of re-pre- senting the pair is given after each response. The test is given twice, without re-presentation of the word pairs the second time. Experimental Tests To systematically evaluate topographical memory and episodic memory, four new tests were developed. Jon’s performance on these tasks was compared with that of a group of 13 right-handed, age- and IQ-matched (Raven’s Advanced Progressive Matrices) control subjects. Two of these control subjects were from Jon’s peer group. All subjects gave their informed consent in accordance with UCL/UCLH Ethics Committees. The mean age of the con- trol group was 21.2 years (SD ⫽ 2.2 years, see Table 1 for IQ scores). All control subjects had had similar experience of video games to Jon. A virtual reality town provided the environment in which to test Jon’s memory. The town was designed by one of us (N.B.) using FIGURE 2. The virtual town. A: A subject’s view of the crossroads the commercially available video game “Duke Nukem 3D” (娀 3D with the cinema on the right. [Color figure can be viewed in the online Realms Entertainment, Apogee Software Ltd., Garland, TX) with issue, which is available at www.interscience.wiley.com.] B: Aerial the editor provided (Build 娀 3D Realms Entertainment) (see Bur- perspective of virtual town (which subjects never saw during testing).  ____________________________________________ HIPPOCAMPUS, SPACE, AND EPISODIC MEMORY 719 underground station, a karaoke bar, and a bank (Fig. 2). Each new location if the subject changed his mind. A view of the loca- interior had several entrances, providing overall a large number of tion represented by each icon could be displayed at any time by routes from one location to another. clicking on it. Subjects were instructed to use the full extent of the grid, A desktop PC with a 19-inch screen was used to display the town and not to cluster the icons in one area. When all 10 icons had been at a frame rate of 21 Hz. No auditory stimuli were used. To ma- placed, subjects could opt to stop or to continue arranging the icons. neuver within the town, subjects used the cursor keys of a key- The completed maps were compared with an ideal map, con- board. To match Jon and the control subjects for dexterity with the structed to reflect the true layout of the town (Fig. 2B). A subject’s keys, they were timed while they followed a short route of arrows map was scored by calculating the error in distance between all through the town. After this, the subjects were given between pairs of icons as a fraction of the mean distance between the pair of 15– 40 min to explore the town. During the exploration phase, icons in the ideal and the subject’s maps. This measure is indepen- subjects were instructed to explore until they felt they were ready to dent of differences in map orientation. Independence from differ- be tested on their memory for the town and had been observed to ences in map size was achieved by scaling each subject’s so as map visit all locations. to best match the ideal map. Specifically, the error measure re- Following exploration, subjects were tested on the following ported in Table 2 is: tasks in the order described. Subjects were told that they would be 冦冬 冭冧 tested on their navigation and memory for the town but not told s |␥d ab ⫺ d ab i | min any specific details of how they would be tested, such as recogniz- ␥ 1 共␥ d ab s ⫹ d ab i 兲 ing scenes from the town or drawing a map of the town. 2 ab s Navigation where d ab is the distance between icons a and b in the subject’s map, d ab is the distance between icons a and b in the ideal map, ␥ i In this task, subjects had to navigate between different locations is an arbitrary scaling factor, and 具.典ab represents the average over all in the town, using the most direct route available. They were pairs of icons ab. shown a picture of a target location which was continuously present during navigation to that location. When the subject reached the location they were shown a new picture and asked to Episodic Memory repeat the process until they had navigated to all 10 places. These This task assessed Jon’s memory for various aspects of events. target locations were spread evenly across the town and varied in The test was run twice. Each run consisted of two phases, a pre- their relative proximity to the start point and difficulty to find. To sentation phase and a test phase (Fig. 2). During presentation, the assess the accuracy of a subject’s navigation, the average path length subject followed a route through the town, indicated by pictures of was calculated from cursor key presses made by the subject during locations along the route. While following this route, subjects re- the task. peatedly encountered two solitary characters, at 16 fixed intervals along the route. Every time a character was encountered, the sub- Environmental Scene Recognition ject was required to press a key, causing the character to produce an object, which the subject collected. They collected objects in two of Paired forced-choice recognition of 20 pairs of scenes was used the rooms on each route, but not always in the same part of the to test memory for locations visited during exploration and navi- room. The objects were common, familiar objects, e.g., a light gation. One of the scenes in each pair was a view from inside the bulb. Before the subjects started the presentation phase they were virtual town. Note that the subject was unlikely to have experi- informed that their memory for the objects, who gave them the enced the exact scene used; indeed, some of these were taken from objects, where they received the objects, and the order in which viewpoints which the subjects had never been able to reach. How- they collected them would be tested. For the second run, different ever, all subjects would have experienced similar views, and only objects, characters, and places were used. these target views were consistent with the layout of the town. The Subjects’ ability to recall the various aspects of the event of other half of the scenes were made by creating new virtual loca- receiving an object was tested in a counterbalanced, paired forced- tions. These foil views included objects, surface textures, and sim- choice procedure immediately following the presentation phase. ilar geometry from the original town (but spatially rearranged so as Subjects re-entered the rooms in which they had previously col- to be inconsistent with it), as well as novel objects, novel textures, lected the objects. The room now contained one of the two char- and novel geometry. Subjects responded by using a mouse to select acters and, displayed on the nearest wall, two objects and a word one scene from each of the 20 pairs. Responding was self-paced. which represented one of four questions: Map Drawing Object: “which of the two objects displayed were you given?” A computer program was used to assess map drawing in a quan- Person: “which of the two objects did you receive from the char- titative manner. The program displayed a 10 ⫻ 10 grid and a set of acter next to this question?” 10 icons representing locations in the town. Subjects were required Place: “which did you receive in the room you are currently in?” to move icons onto the grid, and thus form a map of the town. and Having been placed in one location, the icons could be moved to a First: “which did you collect first?” 720 SPIERS ET AL. TABLE 2. Results of Experimental Measures 13 control subjects mean Measure Jon (standard deviation) Duration of exploration (minutes) 33a 19.8 (5.8) Topographical memory tests Navigational accuracy (mean path length/virtual meters) 151a 83.4 (10.5) Environmental scene recognition 12/20a 18.2 (1.1)/20 Map drawing accuracy (computed score) 0.45a 0.30 (0.06) Episodic memory test Object recognition Object 30/32 30 (1)/32 Context-dependent memory Average (Person, Place, and First) 50/96a 77 (11)/98 Breakdown of context-dependent memory Person 19/32 26 (4)/32 Place 14/32 24 (6)/32 First 17/32a 27 (3)/32 Average presentation duration (minutes) 8.75 8.15 (1.5) Retrieval duration (average question block time/minutes) 3.32a 2.32 (0.32) a Impaired scores (⬎2 standard deviations from control mean). For the Object question, the foil was an altered version of the is consistent with his performance on the Weschler Adult Intelli- original object or a very similar object from another source. Foil gence Scale reported by Vargha-Khadem et al. (1997). His perfor- objects in the other conditions were other objects from the set they mance on the Little Man Test was well within normal range (Rat- had received. For the Place question, the foil object was one of the cliff, 1979), consistent with the view that he does not have damage collected objects that had not been given to them in the current to the right posterior parietal regions implicated in egocentric spa- room. For the Person question, the foil object was one of the tial processing (e.g., Burgess et al., 1999; Karnath, 1997). objects not given to the subject by the character that was next to the Results from the Camden Memory Test show that Jon is within question. From the two replications of the test, a total of 32 ques- the average range on both the Topographical and the Pictorial tions of each type was asked, resulting in a grand total of 128 Recognition Memory Tests. On the Paired Associate Memory Test questions. The questions were given in the form of eight sets of his performance was at the low end of normal range for the first eight questions, and the time taken to answer each set of eight ques- trial, but impaired on the second trial. These results mirror the tions was recorded. The question order, and left-right response to each results of Vargha-Khadem et al. (1997), in that his recognition is at question, were counterbalanced across the test phase. the high average end of the normal scale in contrast to his recall, Prior to testing proper, subjects were given practice trials during which is impaired. which they followed each of the two routes and were presented with four practice objects and two practice characters. They were Experimental Tests asked whether they had used any particular associational or mne- monic strategies to aid their recall of these practice events. If they Topographical memory tests had, they were asked to avoid using these strategies and simply pay attention to the aspects of the events in the experimental test. This The results of experimental tests are shown in Table 2. The was done to avert the use of explicit strategies. average path length was used to assess the accuracy of navigation. Jon’s mean path length was over 6 standard deviations longer than those of controls (see Table 2), indicating that his ability to navi- gate is severely impaired. His ability to discriminate between scenes RESULTS from the virtual town and altered scenes, or scenes from a different town, was also significantly compromised (see Table 2). Jon’s map of the town appeared disordered compared with an ideal map, and Standardized Neuropsychological Tests was ranked worst of all by our measure of accuracy (see Fig. 4). His Results of the standardized tests are shown in Table 1. Jon’s computed score was impaired by 2.4 standard deviations relative to good performance on Raven’s Advanced Progressive Matrices Test the control subjects’ mean (see Table 2). Jon explored the town for  ____________________________________________ HIPPOCAMPUS, SPACE, AND EPISODIC MEMORY 721 longer than did the controls, and indeed had experience of a similar town during piloting. The two control subjects from Jon’s peer group performed no differently from the other 11 control subjects (average exploration time, 21.0 min; average navigation distance, 91.5 virtual m; average environmental scene recognition, 87.5% correct; and average map drawing error, 0.323). Paired forced-choice episodic memory tests As expected, Jon’s performance on the forced-choice recogni- tion of objects presented was normal (Table 2). Although the per- formance of the controls was high, only three scored at ceiling (32/32). However, Jon’s ability to retrieve contextual information about receiving objects (i.e., which object was given first, by whom, and in what location) was deficient. Of the 96 contextual ques- tions, 50 responses were correct (i.e., two more than chance), a standard which was significantly lower than controls who averaged 76 correct responses. Examining the individual categories, Jon was at chance on each (Table 2), whereas only 0/13, 1/13, and 2/13 controls were at chance on the First, Person, and Place questions, respectively. On the individual tests, Jon’s score was 2 standard deviations lower than controls on the First question but not on the other two context-dependent questions, due to the high variance of controls’ scores (Table 2). To stabilize variance with respect to the mean of the proportion correct for each type of question, an arcsine transformation was applied to the data, which did not alter the overall finding. On a number of occasions, when answering the Person questions, Jon would try to position himself in the room so that one of the two objects appeared next to the character while the other object was occluded by the character, i.e., creating a situation similar to presentation of an event. None of the controls did this. Jon’s two peer group controls’ performance did not differ from the other controls (average number of Object questions correct, 31.5/ 32; and average context-dependent questions correct, 78/98). Both Jon and the controls took the same amount of time to collect the objects at presentation (average of 8 –9 min), although Jon took significantly longer to answer the questions. However, there was no evidence of a decline in Jon’s performance or that of controls over the duration of retrieval. DISCUSSION FIGURE 3. Episodic memory task. A: presentation phase: a view of one of the events. B: Recall phase: example of an Object question. Here, the object on the left is a collected object, and the object on the When first interviewed, Jon presented with three main symp- right is a foil object. The person next to the question is irrelevant for toms: spatial, temporal, and episodic memory problems. Using a the Object questions. C: Recall phase: example of a context-depen- computer-simulated environment, we have characterized his spa- dent question, i.e., a Place question. In this question the subject has to tial and episodic memory problems systematically. To our knowl- remember which of the two objects they collected in the question edge, this is the first study to examine both topographical and location. For the other context-dependent questions the word Place was replaced by the words Person or First, and the subject had to episodic memory together within the same simulated lifelike but remember which object the person next to the question had given controlled environment. Our results indicate that selective bilat- them or which of the two objects had been collected first, respectively. eral hippocampal damage disrupts navigation, topographical [Color figure can be viewed in the online issue, which is available at memory, and the contextual aspects of episodic memory, but does www.interscience.wiley.com.] not affect recognition when the stimulus is re-presented exactly as it was at presentation, i.e., visual pattern matching is unaffected. 722 SPIERS ET AL. FIGURE 4. Map drawing results. A: Ideal map to which the patient and control maps were compared. Road sections were not analyzed. B: Map drawn by the median control subject, score ⴝ 0.28. C: Patient Jon’s computer-drawn map of the town. Jon’s map had an associated score of 0.45. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.] Topographical Memory large-scale space, thus enabling good performance solely in terms of the familiarity of the pictures or visual pattern matching. Jon’s Many previous studies have implicated the medial temporal performance clearly dissociates such a test from more active tests of lobes in topographical memory (e.g., Maguire et al., 1996a; Habib navigation and map drawing, suggesting the possibility that, while and Sirigu, 1987; Aguirre et al., 1996), although relatively few extrahippocampal medial temporal regions are sufficient for topo- specified the role of the hippocampus or simulated actual naviga- graphical recognition tests (perhaps the posterior parahippocampal tion (Bohbot et al., 1998; Maguire et al., 1998b; Gron et al., cortex; see Epstein and Kanwisher, 1998), active navigation re- 2000). Despite his greater exploration of the town, Jon’s naviga- quires the hippocampus. tion, environmental scene recognition, and map drawing were all Our test of environmental scene recognition is different from impaired, consistent with these previous studies. Due to the large- scale nature of the town, all the tasks required long-term storage of the Camden Topographical Recognition Memory Test because it the spatial layout of the town, i.e., memory consistent with having requires the identification of scenes that had not been explicitly stored a cognitive map of the virtual reality town. Whether tested studied. Control subjects found it relatively easy to identify these by navigation, map drawing, or identifying scenes of the town, Jon scenes from the information encoded during the exposure to the does not appear to have access to this type of representation of the many thousands of scenes experienced during exploration and nav- town’s layout. Thus bilateral hippocampal damage appears to igation. However, due to the design of the foils in this task, it is not compromise the ability to form this type of representation. possible to perform consistently by identifying single objects Complementary to our finding that bilateral hippocampal pa- within the scene or by matching the test scenes to a previously thology severely disrupts navigation in large-scale space, unilateral encoded template scene, as the exact angle of view may have been damage to the medial temporal lobe has been shown to disrupt different. Instead, successful discrimination of target views proba- way-finding abilities (Maguire et al., 1996a). In a previous neuro- bly requires retrieval of abstracted environmental information, imaging study on healthy individuals using a similar task, blood such as the layout. This may explain why the hippocampus is flow in both hippocampi was found to be significantly greater required to solve our environmental scene recognition task. In during successful navigation than when following a route of arrows support of this, Holdstock et al. (2000) found that bilateral hip- through the town (Maguire et al., 1998b). Furthermore, blood pocampal damage has a substantial effect on the recognition of flow within the right hippocampus was found to be correlated with allocentric information at a delay of 60 s, but not the recognition of accuracy of navigation. The present data on Jon are consistent with egocentric information at the same delay. The fact that Jon’s de- the results of this previous study and other neuroimaging studies layed recognition of verbal information is unimpaired (Baddeley et (Maguire et al., 1996b, 1997, 1998; Ghaem et al., 1997; Gron et al., 2001) suggests that it is unlikely that his impaired environmen- al., 2000), further demonstrating that the hippocampus is critically tal scene recognition performance is due to the delay between required for accurate navigation. encoding and recognition. Despite the use of “topographical” stimuli, Jon’s performance The ability to draw an accurate map of a previously explored was in the average range on the Camden Topographical Recogni- environment also requires many skills and is likely to be subserved tion Memory Test (CTRMT), thus illustrating the extent of his by many brain regions. In this study, we determined that one of preserved recognition memory. Although this test can be used to these regions is the hippocampus (Jon’s map drawing performance diagnose topographical amnesia (e.g., Whiteley and Warrington, was worse than that of all the control subjects). Given Jon’s im- 1978; Habib and Sirigu, 1987), it involves recognizing photo- paired performance on the navigation and environmental scene graphs and does not require subjects to form a representation of recognition tasks, his map-drawing deficit is likely due to an in-  ____________________________________________ HIPPOCAMPUS, SPACE, AND EPISODIC MEMORY 723 ability to accurately recall the relative abstracted locations of ele- was presented. Thus the results are consistent with view that the ments in the town, consistent with the cognitive mapping hypoth- hippocampus is involved in the retrieval of context-dependent in- esis. Unlike previous studies which used subjective measurements formation but not necessarily familiarity-based recognition (e.g., (e.g., Maguire et al., 1996a; Suzuki et al., 1998), we used an ob- Aggelton and Brown, 1999). These findings are also consistent jective measure of map drawing accuracy, rather than subjective with the observation of left hippocampal involvement in the Place ratings. Note that, in this test, the pictures indicating locations condition, but not in the Object condition, as reported in an fMRI were designed to be as salient as possible and were all highly dis- study (Burgess et al., 2001). Note that Jon’s poor context-depen- tinguishable from each other in contrast to the environmental dent memory cannot be explained by a failure to distinguish one scene recognition test, in which foils were designed to resemble object from another, as Jon was able to correctly recognize the targets. This result is also consistent with the finding that damage objects even among very similar foils. His unique strategy of trying to the medial temporal lobes can cause a deficit in map drawing to “line up” one of the objects with the character in the Person (Maguire et al., 1996a). questions is of interest. He reported that he was attempting to In addition to the impairments shown on the three topograph- recreate the situation at the time of presentation, which suggests ical tasks, Jon took significantly longer to explore the town than that he was trying to make use of his spared ability to match visual did control subjects. Given the impairments on the other tasks, this patterns. finding can be attributed to Jon’s inability to learn about the layout The finding that hippocampal damage impairs performance on of the town, as subjects were asked to finish exploration when they context-dependent questions is consistent with the relational the- thought they knew where everything was. ory of Cohen and Eichenbaum (1993) of hippocampal function, The finding that Jon is unimpaired on the Little Man Test in that hippocampal damage has disrupted the ability to bind (Ratcliff, 1979), performance on which is considered to be sensi- together associations between different stimuli. However, it is not tive to right posterior cortical damage, suggests that his spatial clear that Jon’s particular pattern of results (including his preserved navigation deficit is not attributable to some hidden damage in the recognition memory for pairs of words and pairs of faces, and right posterior cortex, or to more general problems with egocentric preservation of much semantic memory) is best described by a spatial processing such as mental rotation. This finding, together deficit in relational processing. It could be argued that the three with Jon’s unimpaired performance on the Camden Topographi- different associations in our episodic task, i.e., object-location, cal Recognition Memory Test, raises the possibility that spatial object-time (temporal order), and object-person, require integra- navigation as an example of a real-world function may not be tion of information represented in different cortical areas, although adequately tested by standard “spatial” neuropsychological tests. these different areas have not all been identified. However, it is not clear whether this explanation (or the relational theory) accounts for Jon’s impairment on the recognition of scenes from the town or Episodic Memory his more general recall failure shown in numerous tests evaluated Our test of episodic memory involved the recognition of certain through unimodal and cross-modal tests. aspects of events, such as receipt of objects from people in partic- It has been suggested that developmental hippocampal damage ular locations within a virtual reality town. When Jon’s memory of does not affect familiarity-based recognition in the same way as the objects he had received (Object questions) was tested, his score late-onset hippocampal damage (which can severely impair it; see was within normal range (the performance of 6/13 controls was patient R.B., Zola-Morgan et al., 1986; Manns and Squire, 1999). equal to or worse than Jon’s). Although 3 of 13 controls’ scores At present, it is difficult to know whether the differences in the were at ceiling, the number of correct responses made by Jon level of spared recognition memory between adult-onset damage exactly matched the mean of the control subjects, indicating that and developmentally sustained damage are primarily due to the his memory for the objects was unimpaired (Jon is also unimpaired selectivity of the hippocampal pathology, or the increased capacity when an arcsine transformation is applied to the data, showing the for reorganization and compensation of memory function in de- value of 0.52 standard deviations from the mean). Such spared velopmental cases, or a combination of both. However, a simple recognition for objects is consistent with the results from previous distinction between early-onset and late-onset damage seems un- tests of Jon (Vargha-Khadem et al., 1997) and with meta-analyses likely, given that patient Y.R., with late-onset selective bilateral of recognition in other amnesic patients (e.g., Aggleton and Shaw, hippocampal damage, shows a similar pattern of impairment to 1996; but for an alternative point of view see Manns and Squire, Jon: preserved recognition, but impaired recall (Holdstock et al., 1999). 1999, 2000). In contrast, the hippocampus appears to be required for remem- It could be argued that Jon’s topographical memory deficit bering the context-dependent aspects of events (i.e., Person, Place, forms part of a more generalized deficit in episodic memory, i.e., and First questions). Jon was impaired relative to control subjects topographical memory forms part of episodic memory. However, on these types of questions, and his performance was not signifi- there is reason to believe that topographical and context-depen- cantly different from chance. This is evidence that the hippocam- dent episodic memories are dissociable. In a recent study of pus is required for context-dependent memory. Whereas the Ob- adult patients with unilateral temporal lobectomy, using the ject questions can be solved without reference to context (using the same tasks (Spiers et al., 2000), right temporal lobectomy pa- visual familiarity of the objects as the cue), the other questions tients were impaired on all the topographical tests, whereas left demand access to the spatio-temporal context in which the object temporal lobectomy patients were impaired on their memory 724 SPIERS ET AL. for context-dependent episodic memory questions. Thus, it is Cohen NJ, Eichenbaum H. 1993. Memory, amnessia and the hippocam- possible that, rather than reflecting a general nonlateralized pal system. Cambridge, MA: MIT Press. Duzel E, Vargha-Khadem F, Heinze HJ, Mishkin M. 1999a. ERP evi- deficit, Jon’s impairments are due to the loss of what would be dence for recognition without episodic recollection in a patient with a right lateralized hippocampal spatial system and a left lateral- early hippocampal pathology. Soc Neurosci Abstr 259:11. ized hippocampal context-dependent episodic system in the Duzel E, Cabeza R, Picton TW, Yonelinas AP, Scheich H, Heinze HJ, normal brain. Of course, in Jon, the lateralization of the re- Tulving E. 1999b. Task-related and item-related brain processes of maining parts of these systems may have been reorganized (see memory retrieval. Proc Natl Acad Sci USA 96:1794 –1799. Maguire et al., 2001). Epstein R, Kanwisher N 1998. A cortical representation of the local visual environment. Nature 292:598 – 601. In summary, the performance of patient Jon on our virtual Gadian DG. 1995. N-acetylaspartate and epilepsy. Magn Reson Imaging reality tasks confirms and quantifies the real-life symptoms of im- 13:1193–1195. paired way-finding and episodic memory with which he originally Gadian DG, Mishkin M, Vargha-Khadem F. 1999. Early brain pathology presented. It also supports the view that the hippocampus is vital and its relation to cognitive impairment: the role of quantitative mag- for supporting topographical memory and context-dependent ep- netic resonance techniques. Adv Neurol 81:307–315. isodic memory. By contrast, it also indicates that familiarity-based Gadian DG, Aicardi J, Watkins KE, Porter DA, Mishkin M, Vargha- Khadem F. 2000. Developmental amnesia associated with early hy- recognition memory can be spared, even on tasks requiring “topo- poxic-ischaemic injury. Brain 123:499 –507. graphical” scene recognition. Ghaem O, Mellet E, Crivello F, Tzourio N, Mazoyer B, Berthoz A, Denis M. 1997. Mental navigation along memorized routes activates the hippocampus, precuneus, and insula. Neuroreport 8:739 –744. Acknowledgments Gron G, Wunderlich AP, Spitzer M, Tomczak R, Riepe MW. 2000. Brain We thank Mortimer Mishkin and Eleanor Maguire for useful activation during human navigation: gender-different neural networks as substrate of performance. Nat Neurosci 3:404 – 408. discussions, James Donnett for technical help, and Jon and his Habib M, Sirigu A. 1987. Pure topographical disorientation: a definition family for their help and participation in this study. N.B. is a Royal and anatomical basis. Cortex 23:73– 85. Society University Research Fellow. Holdstock JS, Isaac CL, Cezayirli E, Roberts N, Mayes AR. 1999. What kinds of memory do selective hippocampal lesions disrupt. Br Neuro- sci Assoc 15:86. Holdstock JS, Mayes AR, Cezayirli E, Isaac CL, Aggleton JP, Roberts N. 2000. A comparison of egocentric and allocentric spatial memory in a REFERENCES patient with selective hippocampal damage. Neuropsychologia 38:410 – 425. Incisa della Rochetta A, Cipolotti L, Warrington EK. 1996. Topograph- Abrahams S, Pickering A, Polkey CE, Morris RG. 1997. Spatial memory ical disorientation: selective impairment of locomotor space? Cortex deficits in patients with unilateral damage to the right hippocampal 32:727–735. formation. Neuropsychologia 35:11–24. Jackson GD, Connelly A, Duncan JS, Grunewald RA, Gadian DG. 1993. Aggleton JP, Brown MW. 1999. Episodic memory, amnesia, and the Detection of hippocampal pathology in intractable partial epilepsy: hippocampal-anterior thalamic axis. Behav Brain Sci 22:425– 490. increased sensitivity with quantitative magnetic resonance T2 relax- Aggleton JP, Shaw C. 1996. Amnesia and recognition memory: a re- analysis of psychometric data. Neuropsychologia 34:51– 62. ometry. Neurology 43:1793–1799. Aguirre GK, D’Esposito M. 1999. Topographical disorientation: a syn- Karnath HO. 1997. Spatial orientation and the representation of space thesis and taxonomy. Brain 122:1613–1628. with parietal lobe lesions. Philos Trans R Soc Lond [Biol] 352:1411– Aguirre GK, Dertre JA, Alsop DC, D’Esposito M. 1996. The parahip- 1419. pocampus subserves topographical learning in man. Cereb Cortex Katayama K, Takahashi N, Ogawara K, Takamichi H. 1999. Pure topo- 6:823– 826. graphical disorientation due to right posterior cingulate lesion. Cortex Aguirre GK, Zarahn E, D’Esposito M. 1998. Neural components of 35:279 –282. topographical representation. Proc Natl Acad Sci USA 95:839 – Maguire EA, Burke T, Phillips J, Staunton H. 1996a. Topographical 846. disorientation following unilateral temporal lobe lesions in humans. Baddeley AD, Emslie H, Nimmo-Smith I. 1994. Doors and people: a Neuropsychologia 34:993–1001. re-analysis of psychometric data. Bury St. Edmunds, UK: Thames Maguire EA, Frackowiak RS, Frith CD. 1996b. Learning to find your Valley Test Co. way: a role for the human hippocampal formation. Proc R Soc Lond Baddeley A, Vargh-Khadem F, Mishkin M. In press. Preserved recogni- [Biol] 263:1745–1750. tion in a case of developmental amnesia: implications for the acquisi- Maguire EA, Frackowiak RSJ, Frith CD. 1997. Recalling routes around tion of semantic memory? J Cogn Neurosci. London: activation of the right hippocampus in taxi drivers. J Neurosci Bohbot VD, Kalina M, Stepankova K, Spackova N, Petrides M, Nadel L. 17:7103–7110. 1998. Spatial memory deficits in patients with lesions to the right Maguire EA, Burgess N, Donnett JG, Frith CD, O’Keefe J. 1998a. hippocampus and to the right parahippocampal cortex. Neuropsycho- Knowing where things are: Parahippocampal involvement in en- logia 36:1217–1238. coding object locations in virtual large-scale space. J Cogn Neurosci Bottini G, Cappa S, Geminani G, Sterzi R. 1990. Topographical disori- 10:61–76. entation—a case report. Neuropsychologia 28:309 –312. Maguire EA, Burgess N, Donnett JG, Frackowiak RS, Frith CD, O’Keefe Burgess N, Jeffery KJ, O’Keefe J. 1999. The hippocampal and parietal J. 1998b. Knowing where and getting there: a human navigation net- foundations of spatial cognition. Oxford: Oxford University Press. work. Science 280:921–924. Burgess N, Maguire EA, Spiers HJ, O’Keefe J. In press. A temporoparietal Maguire EA, Vargha-Khadem F, Mishkin M. 2001. The effect of bilateral and prefrontal network for retrieving the spatial context of lifelike hippocampal damage on fMRI regional activations and interactions events. Neuroimage. during memory retrieval. Brain 124:1156 –1170.  ____________________________________________ HIPPOCAMPUS, SPACE, AND EPISODIC MEMORY 725 Manns JR, Squire LR. 1999. Impaired recognition memory on the Doors and episodic memory in a large-scale virtual reality town. Eur J Neu- and People Test after damage limited to the hippocampal region. rosci 12:79. Hippocampus 9:495– 499. Sunderland A, Harris JE, Baddeley AD. 1983. Do laboratory tests predict Meyer V, Yates AJ. 1955. Intellectual changes following temporal lobec- everyday memory? A neuropsychological study. J Verb Learn Verb tomy for psychomotor epilepsy: preliminary communication. J Neurol Behav 22:341–357. Neurosug Psychiatry 18:44 –52. Suzuki K, Yamadori A, Hayakawa Y, Fujii T. 1998. Pure topographical Milner B. 1958. Psychological defects produced by temporal lobe exci- disorientation related to dysfunction of the viewpoint dependent vi- sion. Res Publ Assoc Nerv Ment Dis 46:244 –257. sual system. Cortex 34:589 –599. Milner B. 1965. Visually-guided maze learning in man: effects of bilateral Tally JL. 1993. Children’s auditory verbal learning test. Lutz Psycholog- hippocampal, bilateral frontal, and unilateral cerebral lesions. Neuro- ical Assessment Resources. psychologia 3:317–338. Teng E, Squire LR. 1998. Memory for places learned is intact after hip- Morris RG, Pickering A, Abrahams S, Feigenbaum JD. 1996. Space pocampal damage. Nature 400:675– 677. and the hippocampal formation in humans. Brain Res Bull 40:487– Tulving E. 1983. Elements of episodic memory. Oxford: Clarendon Press. 490. Van Paesschen W, Connelly A, King MD, Jackson GD, Duncan JS. 1997. Nunn JA, Polkey CE, Morris RG. 1998. Selective spatial memory impair- The spectrum of hippocampal sclerosis: a quantitative magnetic reso- ment after right unilateral temporal lobectomy. Neuropsychologia 36: nance imaging study. Ann Neurol 41:41–51. 837– 848. Vargha-Khadem F, Gadian DG, Watkins KE, Connelly A, Van Paesschen O’Keefe J, Nadel L. 1978. The hippocampus as a cognitive map. Oxford: W, Mishkin M. 1997. Differential effects of early hippocampal pathol- Oxford University Press. Ratcliff G. 1979. Spatial thought, mental rotation and the right cerebral ogy on episodic and semantic memory. Science 277:376 –380. hemisphere. Neuropsychologia 17:49 –54. Warrington EK. 1996. The Camden memory tests. Hove: Psychology Press. Raven JC. 1976. Advanced progressive matrices, set I. Oxford: Oxford Warrington EK, Weiskrantz L. 1968. New method of testing long-term Psychologists Press. retention with special reference to amnesic patients. Nature 217:972– Sanders HI, Warrington, EK. 1975. Retrograde amnesia in organic am- 974. nesic patients. Cortex 11:397– 400. Wechsler D. 1947. A standardized memory scale for clinical use. J Psychol Scoville WB, Milner B. 1957. Loss of recent memory after bilateral hip- 19:87–95. pocampal lesions. J Neurol Neurosurg Psychiatry 20:11–21. Whiteley AM, Warrington EK. 1978. Selective impairment of topograph- Semmes J, Weistein S, Ghent L, Teuber HL. 1955. Correlates of ical memory: a single case study. J Neurol Neurosurg Psychiatry 41: impaired orientation in personal and extrapersonal space. Brain 575–578. 86:742–772. Wilson B, Cockburn J, Baddeley AD. 1985. Rivermead behavioural mem- Smith ML, Milner B. 1989. Right hippocampal impairment in the recall ory test. Reading, UK: Thames Valley Test Co. of spatial location: encoding deficit or rapid forgetting? Neuropsycho- Zola-Morgan S, Squire LR, Amaral DG. 1986. Human amnesia and the logia 27:71-81. medial temporal region: enduring memory impairment following a Spiers HJ, Burgess N, Maguire EA, Baxendale SA, Hartley T, O’Keefe J. bilateral lesion limited to field CA1 of the hippocampus. J Neurosci 2000. The effect of unilateral temporal lobectomy on topographical 6:2950 –2967.