Nature Vol. 273 25 May 1978 297 As staying up late at night is usually associated with some degree of sleep deprivation, the effects described here are probably not solely attributable to circadian rhythms. However, the similarity between these results and those of the schoolchildren study6 do suggest that circadian rhythms had an important role in producing them. Whatever their cause, these effects undoubtedly indicate that late-night studying is much less efficient than one's immediate impressions might imply. Much more will be forgotten later on than if the studying had taken place at a more normal time of day. We thank Dr Mary C. Lobban and Alison Edwards (Clinical Research Centre), and Giles Wynne and the other nursing staff (Northwick Park Hospital) for their help in this work. The study was carried out with the approval of the Northwick Park Hospital Ethical Committee. Timothy H. Monk Simon Folkard MRC Perceptual and Cognitive Performance Unit, Laboratory of Experimental Psychology, University of Sussex, Brighton, UK Received 3 March; accepted 6 April 1978. 1. Winch, W. H. /. educ. Psychol. 3, 18-28, 75-82 (1912). 2. Gates, A. i. Univ. Calif. Pubis Psychol. 2, 1-156 (1916). 3. Blake, M. J. F. Psychon. Sci. 9, 349-350 (1967). 4. Baddeley, A. D., Hatter, J. E., Scott, D. & Snashall, A. Q. 11 exp. Psychol. 22, 605-609 (1970). 5. Hockey, G. R. J. & Colquhoun, W. P. Aspects of Human Efficiency— Diurnal Rhythm and Loss of Sleep (ed. Colquhoun, W. P.) 1-23 (English Univ. Press, London, 1972). 6. Folkard, S., Monk, t. H., Bradbury, R. & Rosenthall, j. Br. 1. Psychol. 68, 45-50 (1977). , , 7. Folkard, S., Monk, t. H. & Lobban, M. C. Ergonomics (in the press). 8. Lobban, M. C. & Nessling, R. /. Physiol., Lond. 267, 24-25P (1977). 9. Folkard, S. & Haines, S. M. J. Physio!., Lond. 267, 25-26P (1977). Memory in monkeys severely impaired by combined but not by separate removal of amygdala and hippocampus The profound anterograde amnesia that has been attributed in the clinical literature to damage of the hippocampal system1'2 has not been observed in animals with such damage. Hippocampal-system lesions in animals do markedly impair some forms of spatial memory3,4, but the effects on other forms of memory have generally seemed minor5,8 compared with the dramatic disorder described in man7'8. This discrepancy between the clinical and animal literature could indicate a true evolutionary shift in the functions of the hippocampus9", or, at the other extreme, it could simply reflect the use of incommensurate measures across species1112. (Strong support for this second interpretation has been provided by Gaffan13.) A third possibility, however, is that the discrepancy points to inaccurate localisation of the neuropathology in man that is responsible for the profound amnesia. Support for this last alternative comes from new evidence in monkeys indicating that a striking impairment in visual memory can be produced by the combined ablation of the hippocampal formation and the amygdaloid body, but not by ablation of either of these structures alone. Twelve monkeys {Macaca mulatto) weighing from 3.5 to 5 kg were trained prebperatively in a. Wisconsin general testing apparatus to perform one-trial object recognition" for peanut rewards. The procedure involved presentation of a baited object over the central well of a three-well tray, followed 10 s later by re-presentation of that object (now negative) together with a new object (positive) over the lateral wells. Twenty such trials separated by 30 s intervals were given daily, each trial with a new pair of objects. The successive pairs were drawn randomly from a collection of several hundred unmounted junk objects of diverse sizes, colours, shapes and textures. The testing procedure exploits 0028-0836/78/0273—O297S01.O0 the monkey's natural tendency to choose the novel object in a pair even after only brief familiarisation, a few seconds earlier, with the other object of the pair"'1*; consequently, the criterion of 90 correct choices in 100 trials is achieved extremely quickly, in this case in an average of 99 trials and 28 errors (see Table 1), or about one-third the time required to learn matching, that is, to choose the familiar object in the pair". The animals were then divided into four lesion groups of three monkeys each and given either an amygdalectomy, hippocampectbmy, both combined, or neither. One-stage bilateral lesions were made aseptically while the animals were under Nembutal anaesthesia (35 mg per kg body weight). To remove the amygdaloid body, the frontotemporal junction was elevated slightly and the tissue medial to the anterior tip of the rhinal sulcus was entered with a small-gauge sucker; all gray matter rostral to the pes hippocampi (which afterwards lay exposed in the ventricle) and medial to the white matter of the temporal lobe was then ablated under visual control through an operating microscope. To remove the hippocampal formation, the occipitotemporal convexity was elevated slightly and the tissue medial to the rostral half of the occipitotemporal sulcus was entered with the sucker until the ventricle was opened; the entire hippocampal formation and much of the underlying fusiform-hippocampal gyrus was then ablated, again under visual control through the microscope, the upper ventricular surface serving as a readily identifiable dorsal boundary along the entire length of the removal. For the combined ablation, both of these surgical procedures were used in a single stage. Two weeks postoperatively the monkeys were retrained to criterion, with the outcome as shown in Table 1. Neither amygdalectomy alone nor hippocampectomy alone produced appreciable impairment, but the combination of the two removals yielded a severe effect: an abrupt drop to chance levels of performance followed by an average relearning score that was ten times the average for initial learning. When the monkeys had regained criterion, their recognition ability was tested further with procedures adapted from a study by Gaffan12. First, the delay between familiarisation and choice was lengthened in stages from the original 10 s delay to 30 s, then to 60 s, and finally to 120 s, each stage being tested for 100 trials. Second, the number of objects given for familiarisation before pairing each one with a novel object was increased in stages from the original single object to a series of three objects presented successively, then to a series of five objects, and finally to a series of ten objects; each of these stages was tested for 150 trials. The data in Table 1 indicate that, again, neither amygdalectomy nor hippocampectomy alone yielded more than a mild impairment. Each of these groups averaged 91% correct across the six testing conditions, as compared with an average of 97% correct for the normal controls. By contrast, the combination of the two lesions had a profound effect, yielding an average score of 60% correct, or just above chance. Since the animals had already regained the ability to perform the basic task, their sharp drop in performance with the longer delays and list lengths presumably represents a true memory loss rather than some other difficulty such as in visual perception or problem solving. This striking deficit in monkeys after amygdalo-hippocampal removals is not limited to recognition memory, where the animal must remember whether or not a test object had been seen before. An equally severe impairment has been found in associative memory (B. Spiegler and M.M., unpublished), where the animal must remember on the basis of a single trial whether or not the test object had been rewarded before12. On this more difficult, associative task, unlike the recognition task used here, amygdalectomy alone had a significantly greater effect than hippocampec- © Macmillan Journals Ltd 1978 298 Nature Vol. 273 25 May 1978 Table 1 Effects of removal of amygdala and hippocampus on memory Preoperative Postoperative Delays (%correct) Objects (% correct) Groups Trials Errors Trials Errors 30 s 60s 120 s 3 5 10 Normal control 1 100 26 0 0 97 98 96 96 93 91 2 80 28 0 0 99 100 98 97 97 94 3 40 19 0 0 98 99 98 97 96 92 Amygdalectomy 1 120 42 80 32 95 95 95 91 92 82 2 100 27 340 85 91 89 92 88 87 77 3 80 30 0 0 91 95 94 96 95 87 Hippocampectomy 1 60 17 80 22 98 93 94 95 92 84 2 100 26 120 32 85 89 83 89 85 71 3 120 31 20 4 98 99 95 95 92 88 Amygdalectomy + 1 210 49 760 179 79 65 65 62 64 59 hippocampectomy 2 100 26 1,500 429* 64 59 63 60 55 61 3 80 22 700 203 61 47 52 53 58 44 Group means Normal control 73 24 0 0 98 99 98 97 96 92 Amygdalectomy 100 33 140 39 94 93 94 92 91 82 Hippocampectomy 93 25 73 19 94 94 91 93 90 81 Amygdalectomy + 130 32 987 270 68 57 60 58 59 55 hippocampectomy Scores in preoperative and postoperative columns are the numbers of trials and errors preceding criterion of 90 correct choices of the novel object in 100 trials (delay following familiarisation with the other object in the pairs was 10 s). Scores in delays columns are percentage correct in 100 trials at each of three longer delays tested in succession at the rate of 20 trials per day, except for the longest delay (120 s) which was tested for 10 trials per day. Scores in objects columns are percentage correct in 150 trials for each of three multiple-object conditions tested in succession: in the first condition, three objects were presented for familiarisation, one at a time at 20 s intervals, and then re-presented in the same order, each being paired with a novel object, again at 20 s intervals; in the next condition, five objects were presented one at a time at 20 s intervals, and so on. Thirty trials were presented daily, ten sets of 3's or six sets of 5's or three sets of 10's. The minimum delay between familiarisation and choice was 60, 100, and 200 s for the three conditions, respectively. Scores for individual animals are shown in the upper part of the table, group means in the lower part. Histological examination indicated that the lesions were as intended except in animals 'hippocampectomy 2' and 'amygdalectomy -+- hippocampectomy no. 2', both of which sustained, in addition to the planned removals, bilaterally asymmetrical damage to the ventral part of inferior temporal ('visual') cortex. *Eailed; final score, 85 correct in 100 trials. tomy alone; but neither effect presaged the abrupt and permanent fall to chance levels' of performance that followed the combined ablation. The evidence that a severe memory disorder in monkeys can be produced only by the combined removal of amygdala and hippocampus seems at first surprising; it implies that these two structures, so unlike morphologically, may nonetheless serve as functional substitutes in a still undefined memory circuit. Yet this conclusion gains plausibilty from recent anatomical studies in monkeys which show that the amygdala and hippocampus have many more inputs and outputs in common than was previously supposed. Both of these structures receive projections from many of the same frontal and temporal association areas, the amygdala receiving them directly'617, and the hippocampus indirectly through the entorhinal area"""; and both of these structures send projections, in turn, to many of the same diencephalic and basal forebrain areas, the amygdala via both the ventral amygdalofugal pathway and the stria terminalis21, and the hippocampus by way of the fornix22. Thus, while the amygdala and hippocampus undoubtedly serve different functions, they could still constitute alternative relays through which a particular group of cortical association areas could interact with a particular group of subcortical targets, such that only the combined removal of both relays would cause a complete disconnection. Taken together, these behavioural and anatomical findings in monkeys open up a new possibility for helping to resolve the long-standing discrepancy between the animal and clinical literature on amnesia. To pursue this possibility, several kinds of additional investigations are needed. Among the most critical are: a determination of whether the severe memory impairment demonstrated here following combined amygdalo-hippocampal damage in monkeys transcends the visual modality, in keeping with the 'global amnesia' found in man; a reappraisal of the neuropathology in human amnestic cases with special attention to whether © 1978 Nature Publishing Group there is indeed damage to the amygdaloid system as well as the hippocampal system; and quantification of the memory impairment in clinical cases, to determine whether differences in degree of amnesia might correlate with differences in amount of pathological involvement of these two systems. I thank Mr Leon Dorsey for his assistance. Mortimer Mishkin Laboratory of Neuropsychology, National Institute of Mental Health, Bethesda, Maryland 20014 Received 5 December 1977; accepted 27 March 1978. 1. Barbizet, J. J. Neurol. Neurosurg. Psychiat. 26, 127-135 (19631 2. Brion, S. Int. J. Neurol. 7, 31-43 (1968). 3. Mahut, H. Neuropsychologic! 9, 409-424 (1971) 4. Olton, D. S. Scientific Amur. 236, 82-98 (1976). 5. Weiskrantz, L. in Cognitive Processes of Nonhumun Primates (ed. Jarrard, L. E.) (Academic, New York, 1971). 6. Iversen, S. D. in International Review of Neurobiology (eds Pfeiffer, C. C. & Smythies, J. R.) (Academic, New York, 1976). 7. Milner, B. in Amnesia (eds Whitty, C. W. M. & Zangwill, O. L.) (Butterworths, London, 1966). il. Zangwill, O. L. in Amnesia (eds Whitty, C. W. M. & Zangwill, O. L.) (Butterworths, London, 1966). 9. Woody, C. L>. & Ervin, F. R. Physiol. Behav. 1, 273-280 (1966). 10. Drewe, E. A., Ettlinger, O., Milner, A. D. & Passingham, R. E. Cortex 6, 129-163 (1970). 11. Weiskrantz, L. & Warrington, E. K. in The Hippocampus (eds Isaacson, R. L. & Pribram, K. H.) (Plenum, New York, 1975). 12. Gafl'an, D. J. comp, physiol. Psychol. 86, 1100-1109 (1974). 13. Gaifan. D. Q J. exp. Psychol. 29, 504-514 (1977). 14. Mishkin, M. & Delacour, J. J. exp. Psychol.: Anim. Behav. Proc. I, 326-334 (1975). 15. Mishkin, M, Prockop, E. S. & Rosvold, H. E. /. comp, phvsiol. Psvchol. 55, 178-181 (1962). 16. Herzog, A. G. & Van Hoesen, G. W. Brain Res. 115, 57-69 (1976). 17. Turner, B. H.. Mishkin, M. & Knapp, M. E. Neurosci Abstr. 2, 398 (1976). [8. Van Hoesen. CI. W. & Pandya, D. N. Brain Res. 95, 1-24 (1975). 19. Van Hoesen, G. W., Pandya, D. N. & Butters, N. Brain Res. 95, 25-38 (1975). 20. Van Hoesen. G. W. & Pandya, D. N. Brain Res. 95, 39-59 (1975). 21. Nauta. W. J. H. /. Anat. 95, 515-531 (1961). 22. Poletti, C. E. & Cresswell. G. J. comp. Neurol. 175, 101-128 (1977).