Severe learning impairment caused by combined immunotoxic lesion of the cholinergic projections to the cortex and hippocampus in monkeys

Brain Research 836 Ž1999. 120–138 www.elsevier.comrlocaterbres

Research report

Severe learning impairment caused by combined immunotoxic lesion of the cholinergic projections to the cortex and hippocampus in monkeys Rosalind M. Ridley, Perdita Pugh, Catherine J. Maclean, Harry F. Baker

)

MRC ComparatiÕe Cognition Team, Department of Experimental Psychology, Downing Street, Cambridge CB2 3EB UK Accepted 18 May 1999

Abstract Monkeys with immunotoxic lesions of both the basal nucleus of Meynert and the vertical limb of the diagonal band of Broca ŽNBMq VDB. lost cholinergic innervation throughout the cortex and hippocampus. They were impaired at learning discriminations between objects differing in either few, or many, attributes and at learning visuospatial conditional discriminations. Monkeys with immunotoxic lesions of the NBM lost cholinergic innervation of the neocortex only. Initially, they were unable to learn a simple visual discrimination where the stimuli differed in a limited number of attributes but they were unimpaired at learning discriminations between objects that differed in more attributes. They were mildly impaired at learning a visuospatial conditional task. The impairment exhibited by monkeys with lesions of the NBM alone ameliorated with time but that following NBM q VDB lesions did not. Previous experiments have shown that monkeys with immunotoxic lesions of the VDB alone are impaired at learning visuospatial conditional discriminations but are unimpaired at learning simple visual discriminations. When monkeys with NBM lesions were given excitotoxic lesions of the CA1 field of the hippocampus the learning impairment on discriminations between objects which differed in few attributes was reinstated. Pretreatment with a cholinergic agonist improved learning ability on visual discrimination learning in all monkeys but this improvement was significantly greater in monkeys with lesions of the NBM. On conditional discrimination learning, which is particularly sensitive to hippocampal damage, pilocarpine produced a significant improvement in monkeys with NBM q VDB lesions Žwhere the hippocampal dysfunction was cholinergic. but not in monkeys with NBMq CA1 lesions Žwhere the hippocampal damage was structural.. q 1999 Elsevier Science B.V. All rights reserved. Keywords: Acetylcholine; Monkey; Saporin; Discrimination learning; Cortex; Hippocampus

1. Introduction Administration of low doses of the centrally-acting muscarinic cholinergic blocking agent scopolamine Žhyoscine. shortly before testing affects memory acquisition in humans w11,19x macaques w3x, and marmosets w52x. In marmosets, scopolamine administration impairs simple object discrimination learning. More recently, we have shown that scopolamine causes greater impairments on discriminations between perceptually similar visual stimuli and on visuospatial conditional discriminations than on discriminations between stimuli which are visually very different from each other w26x. Studies in monkeys, together with the earlier work with scopolamine in humans, point to the rising cholinergic projection systems as candidate neural substrates for supporting the acquisition and encoding of ) Corresponding [email protected]

author.

Fax:

q 44-1223-339014;

E-mail:

information into long term memory. Yet evidence from rodent studies involving lesions of the cell bodies of the cholinergic projections have led to conflicting interpretations of the role of these projections w17x. For example, rats with lesions of the basal forebrain Žthe origin of cholinergic projections to the neocortex in rodents. induced by stereotaxic injections of quisqualic acid or AMPA, were found to be less impaired on task performance than were rats whose basal forebrain lesions were produced by less specific toxins such as ibotenic or kainic acid. This suggested that some of the impairments in these latter animals were due to damage to non-cholinergic cells Žfor review, see Dunnett et al. w16x.. Some of these rodent experiments were concerned with more general aspects of behaviour or competence in performing previously acquired task requirements than with acquisition of associative information into long term memory. It has been argued that the cholinergic projections from the basal forebrain are more likely to be involved in attentional functions than in learning

0006-8993r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 6 - 8 9 9 3 Ž 9 9 . 0 1 6 4 1 - 8

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

w40,61x. In order to draw such a distinction it must be supposed that the cortical cholinergic projections have a specific role in cognition. The view that acetylcholine has a modulatory effect on cortical function, however, implies that the cortical cholinergic projections may affect all the cognitive processes in which cortex participates, including perception, attention, learning and intention. In marmosets, excitotoxic lesions of the basal nucleus of Meynert ŽNBM. result in a loss of cholinergic markers in the cortex, but not in the hippocampus, and modest, or transient, impairments on discrimination learning, retention, and reversal tasks w49,50,56,62,63x. Irle and Markowitsch w28x found longer-lasting learning impairments following such lesions in squirrel monkeys, although Voytko found no effect on simple discrimination tasks following ibotenic lesions of the basal forebrain cholinergic system which included the NBM in macaques w71x. Aigner and Mishkin w1x and Aigner et al. w2x found small or transient effects on performance of a recognition task following excitotoxic basal forebrain lesions in macaques. Excitotoxic lesions of the vertical limb of the diagonal band ŽVDB. in marmosets result in a loss of cholinergic markers in the hippocampus and entorhinal cortex, and severe impairments on visuospatial conditional learning but not visual discrimination learning w47,58x. More specific lesions of the cholinergic cell bodies can be achieved by targetting them with a non-specific toxin coupled to an antibody that recognises antigens expressed on those cells. The ribosome-inactivating lectin, saporin, has been conjugated with an antibody Ž192IgG. which recognises the p75 neurotrophin receptor protein Žlocated on cholinergic cells in rats., to produce such an immunotoxin, IgG-saporin. Several studies have examined the effects of IgG-saporin injection into the basal forebrain in rodents and have failed to find unequivocal evidence of cognitive impairment, although the tasks used in many of these studies did not assess acquisition of information about visual stimuli into long-term memory w7,67,73x. When IgG-saporin was injected into both the basal forebrain and the medial septum Žwhich is equivalent to the VDB in monkeys., mild impairments were found in a radial arm maze task but not in a water maze task w5,14x. Although modest effects have been found on tasks which depend on cognitive functions other than acquisition of information into long-term memory Žfor review, see Baxter and Gallagher w6x., the surprising feature of most of the studies of immunotoxic lesions of either the basal forebrain or medial septum in rodents is their inability to find significant cognitive effects. It may be inappropriate to attempt to ascribe a unitary cognitive function to the rising cholinergic projections; but if there is one, rodent studies have not established what it is. Intraventricular Ži.v.. injections of IgG-saporin in rats result in loss of cholinergic cells in the basal forebrain and the medial septum. This method has generally had greater effects than have intraparenchymal injections of IgG-

121

saporin, especially on spatial learning tasks w7,31,41x. Only Vnek et al. w70x have looked at the effect of i.v. administration of IgG-saporin in rats on object discrimination problems; these authors found that the rats were impaired at learning and relearning object discriminations. Although it is possible that i.v. injections of IgG-saporin produce additional damage, e.g., destroy non-cholinergic p75-positive Purkinje cells in the cerebellum w72x, this technique may be effective because it can produce more extensive lesions of the basal forebrain cholinergic system than intraparenchymal injections can achieve. It is also possible that the two target areas of the basal forebrain and medial septum, i.e., the neocortex and the hippocampus, co-operate in many cognitive functions by providing complementary but different contributions. In an earlier study using monkeys we found preliminary evidence to support this possibility; combined stereotaxic immunotoxic lesions of the NBM q VDB caused greater impairments on visual discrimination acquisition and retention than did lesions of either area alone w51x. In the present study we confirm a substantial and persistent impairment on visual discrimination learning relative to control performance following combined NBM q VDB lesions and demonstrate a transient and circumscribed effect of NBM lesions alone. Addition of an excitotoxic lesion of the CA1 field of the hippocampus in monkeys with immunotoxic lesions of the NBM restored the learning impairment in these animals, demonstrating a functional interaction between neocortex and hippocampus.

2. Materials and methods 2.1. Animals and Surgery All procedures were carried out under a UK Home Office Project Licence in accordance with UK legislation. Eighteen young adult common marmosets Ž Callithrix jacchus ., nine male and nine female, aged ; 12 months at the beginning of the experiment and weighing 300–350 g each were used. They were born and reared within our colony. Following pre-operative testing on a single simple visual discrimination task Žsee below. each monkey was assigned to one of three groups so that the groups were matched for learning ability prior to surgery: Group 1 Žfour males, two females.; Group 2 Žtwo males, four females.; Group 3 Žthree males, three females.. Group 1 was a control group and received no surgery. For the first stage of the experiment, Group 2 monkeys were injected into the NBM, and Group 3 monkeys were injected into both the NBM and the VDB, with ME20.4IgG-saporin at a concentration of 0.2 mgrml in saline. ME20.4IgG is an antibody that recognises p75 neurotrophin receptor protein on cholinergic cell bodies in marmosets w32x. It has a greater affinity for cholinergic cells of the NBMrVDB than for cholinergic interneurones in the basal ganglia.

122

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

ME20.4IgG-saporin was obtained commercially from Advanced Targeting Systems ŽSan Diego, CA.. Starting 1 week after surgery, the monkeys were tested on a series of learning tasks Žsee below.. When the first stage of cognitive testing was completed, the monkeys in Group 2 received bilateral excitotoxin injections within the CA1 region of the hippocampus. Following a 1 week post-surgical recovery period behavioural testing was resumed for all monkeys. At the end of this stage a drug study was carried out in which the effect of the direct cholinergic agonist, pilocarpine, on learning ability in the three groups of monkeys was assessed. For immunotoxin injections, the monkeys were premedicated with 0.05 ml intramuscular Ži.m.. ketamine Ž100 mgrml. and anaesthetised with 18 mgrkg i.m. alphaxalone–alphadolone ŽSaffan, Glaxovet, UK.. For the monkeys in Group 2 ŽNBM., the immunotoxin was injected stereotaxically using the frontal approach previously described w18x. The injection needle was positioned at 568 from the vertical, in the sagittal plane. A burr hole was made over the frontal cortex on each side. The needle approach came in above the orbits, passed beneath the basal ganglia, and ended within the NBM. Group 2 ŽNBM. monkeys each received four deposits of ME20.4IgGsaporin at the following coordinates Žaccording to the atlas of Stephan et al. w66x.: AP q 0.87 cm, L " 0.47 cm, V q 0.72 Ž1.3 ml.: AP q 0.92 cm, L " 0.43 cm, V q 0.72 cm Ž2.2 ml.; AP q 0.98 cm, L " 0.39 cm, V q 0.72 cm Ž2.2 ml.; AP q 1.03 cm, L " 0.35 cm, V q 0.72 cm Ž1.3 ml.. For Group 2 ŽNBM. monkeys the amount injected was, therefore, 1.4 mg ME20.4IgG-saporin per hemisphere. Following surgery, the monkeys were kept overnight in incubators. They were given 0.2 ml Ž24 mgrml. paracetamol syrup orally if necessary. Five of the 6 monkeys, which received bilateral immunotoxic lesions of the NBM in one surgical procedure, recovered uneventfully from surgery. They were returned to their home cage and usual partner in 1–2 days. One monkey showed a syndrome similar to that shown by the majority of monkeys with NBMq VDB lesions Žsee below.. The monkeys in Group 3 ŽNBMq VDB. were injected into the NBM with ME20.4IgG saporin in the same way as the monkeys in Group 2 ŽNBM.. They were also injected into the VDB during the same surgical session, using a conventional vertical approach. A burr hole was made over the central sinus and two deposits of 1.5 ml each of immunotoxin were made in each hemisphere at the following coordinates: AP q 1.15 cm, L " 0.08 cm, V q 0.83 cm: AP q 1.20 cm, L " 0.05 cm, V q 0.98 cm. Thus, each monkey in Group 3 ŽNBMq VDB. received a total of 2.0 mg ME20.4IgG-saporin per hemisphere. Since the VDB deposits were near to the midline, the lateral zero was taken as the midpoint of the central sinus. Group 3 ŽNBM q VDB. monkeys were kept overnight in incubators. They were given 0.2 ml Ž24 mgrml. paracetamol syrup orally if necessary. They recovered uneventfully from surgery and

were returned to their home cage and usual partner in 1–2 days. About 2 weeks after surgery, five of the six monkeys with this lesion Žand one of the monkeys in Group 2, see above. became hypothermic and were returned to the incubators for intermittent periods until their thermoregulation recovered spontaneously. For the next two weeks these monkeys were intermittently behaviourally disturbed Že.g., hyperactive, dyspraxic, dysphagic or disorientated., they ate slowly and their movements were frail. Nonetheless they fared well in a sheltered environment, recovered spontaneously and were returned to their home cage and usual partner in about 2 weeks. The monkeys that exhibited this syndrome were not tested while behaviourally disturbed though all the monkeys had begun post-operative testing 7 days after surgery. For excitotoxin injections, the monkeys in Group 2 ŽNBM. were anaesthetised with 18 mgrkg alpahaxalone– alphadolone. ŽKetamine was not used to premedicate the monkeys since it is neuroprotective against the excitotoxin used.. Dexamethasone Ž2 mgrkg. was given prior to surgery to minimise cerebral oedema since this might have inhibited uptake of excitotoxin. The excitotoxin used was 0.12 M N-methyl-D-aspartate ŽNMDA. in 0.85% saline, adjusted to pH 7.4 with 1 N NaOH. After making a small skin incision and burr holes to the right and left of the occiput, NMDA was delivered throughout the greater part of the CA1 field of the hippocampus along a vertically and laterally diagonal trajectory using the method and co ordinates described fully elsewhere w60x. NMDA Ž0.7 ml. was delivered at each of five equally spaced positions along the length of CA1 in each hemisphere and each injection was delivered over the course of 1 min. Injection of excitotoxin was carried out in each hemisphere on 2 separate days with 2 or 3 days intervening, in order to minimise the possibility of generalised seizures. Following each surgical procedure, the monkey was placed in an incubator for recovery. Most monkeys showed mild, ipsilesional signs of epileptiform excitation Že.g., facial twitching. as they recovered from anaesthesia, but no generalised seizures. Anti-epileptic drugs were not required. The monkeys were returned to their home cages and usual partners in 1–2 days. They were given 0.2 ml Ž24 mgrml. paracetamol syrup orally if necessary, and behavioural testing recommenced 7 days after surgery. 2.2. CognitiÕe testing The monkeys were tested in a Wisconsin General Test Apparatus ŽWGTA. using standard methods w4x. They were tested on acquisition and retention of various discriminations to a predetermined criterion Žusually 27 correct responses in 30 consecutive trials, 27r30.. About 40 trials were given each weekday and the intertrial interval was about 15 s. Correct responses were rewarded with either a small piece of marshmallow or a small piece of bread soaked in syrup, depending on each monkey’s preference.

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

In order to maintain motivation, monkeys were fed their normal daily diet of bread, fruit and monkey chow after testing each day. Various types of discrimination task were used. In simple coloured object discriminations ŽCO tasks., two small plastic coloured ‘junk’ objects Že.g., toy ballerina, toy soldier. were presented over two food wells in the WGTA and the monkey had to learn to displace the appropriate object to retrieve the reward underneath. The leftrright position of the objects varied randomly from trial to trial but the reward was always placed under the same object Že.g., the ballerina.. In simple black object discriminations ŽBO tasks., two junk objects painted black all over were used. The coloured object tasks and the black object tasks differed in the type of perceptual analysis required for task solution, in that the coloured objects could be distinguished by identifying many different individual features Že.g., coloured parts. or by overall appearance, while the black objects could only be distinguished on the basis of overall appearance. In simple ‘pebble object’ discriminations ŽPO tasks., small, somewhat similar but geologically distinct, pebbles were used as stimuli. These tasks were intermediate in task difficulty between discriminations involving simple coloured objects and black objects. The monkeys were also tested on visuospatial conditional discriminations ŽVS tasks.. In these tasks, two pairs of identical, coloured objects were used on different trials. When one pair of identical objects ŽAA. was presented over the food wells, the reward was placed under the object on the left. When the alternative pair ŽBB. was presented, the reward was placed under the stimulus on the right. The pairs of stimuli were presented in random order. In simple visual discrimination tasks the monkey learns to associate one of the two stimuli with reward. This strategy cannot be used to learn the visuospatial conditional task because on each trial both stimuli are identical and the stimulus pairs AA and BB are equally associated with reward. Correct choice between the identical objects on each trial is conditional on both the appearance of those stimuli ŽAA or BB. and the position of each stimulus in the spatial array. Normal monkeys find visuospatial conditional discrimination tasks more difficult to learn than simple visual discriminations, although the second and subsequent examples of both types of task are easier to learn than the first presented example since the general rules of responding will then be familiar. We also tested the monkeys on a ‘mixed object and spatial discrimination’ ŽMOS. in which trials of an object discrimination ŽA vs. B. are randomly presented on some trials while a spatial discrimination is presented on other trials Ž‘go left on every trial when stimuli C and C are presented’.. Choice between stimuli is conditional not only on the appearance of the stimuli but on the rule applicable to those stimuli. We were interested in assessing whether learning one type of task interfered with learning another type of task.

123

2.2.1. Pre-operatiÕe testing Prior to stage 1 surgery, all monkeys were tested on a simple visual discrimination ŽB vs. S. to a criterion of 90r100. This high criterion was used to stabilise performance in the test situation. Each monkey was then assigned to one of three groups on the basis of its learning score Žthe number of trials required to learn the task excluding those trials in criterion., so that the three groups were matched for learning ability. 2.2.2. Testing following stage 1 surgery (Group 1, unoperated controls; Group 2, NBM lesions; Group 3, lesions of NBM and VDB) Behavioural testing recommenced 7 days after stage 1 surgery and tasks were presented to a criterion of 27r30. The monkeys were first tested on retention of the simple discrimination learnt prior to stage 1 surgery ŽB vs. S Ret.. This was followed by acquisition Žnew learning. of three simple visual discriminations using black objects ŽBO1., pebble objects ŽPO1., and coloured objects ŽCO1.. For these tasks, the monkeys were deemed to have failed if they did not reach criterion in a total of 300 trials. This cut-off was used because it was supposed that some lesioned monkeys might never have reached criterion. The monkeys were then tested on a visuospatial conditional discrimination task ŽVS1. to 27r30. Since this was the first time these monkeys had encountered a visuospatial conditional task they were deemed to have failed at 400 trials rather than 300 trials. This was followed by the MOS. This was designed to assess the monkeys’ ability to acquire and follow two different rules at the same time. This stage of the experiment was completed by testing the monkeys on a further three simple visual discriminations using black objects ŽBO2., pebble objects ŽPO2. and coloured objects ŽCO2.. It was anticipated, on the basis of previous work w18,51x, that the effect of the NBM lesion would have substantially ameliorated by this time. 2.2.3. Testing following stage 2 surgery (CA1 lesion added to NBM lesion in Group 2 monkeys) In order to qualify to take part in the second stage of the experiment, monkeys in Group 2 ŽNBM. had to have shown substantial recovery of learning ability by the end of stage 1. Since one monkey in Group 2 ŽNBM. remained maximally impaired it did not receive stage 2 surgery although it continued to be tested behaviourally. The entire behavioural data from this monkey was removed from the statistical analysis such that Group 2 consisted of five monkeys. ŽThis monkey is described separately in Section 3.. Following a 1-week post-operative recovery period Žduring which no monkeys were tested., testing recommenced for all groups with retentions of the three simple visual discriminations learnt just before stage 2 surgery ŽBO2 Ret, PO2 Ret, CO2 Ret., all to a criterion of 27r30. Assessment of the monkeys’ new learning of BO tasks and

124

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

VS tasks was carried out as part of the pilocarpine study Žsee below.. 2.2.4. Effect of pilocarpine on learning ability in the lesioned monkeys The monkeys were tested on a series of four simple, black object visual discriminations Žto 27r30 each. using new black objects for each task. All monkeys were presented with the tasks in the same order, ensuring that any interference effects between stimuli, which happened to resemble each other, were equivalent between monkeys. The monkeys were pretreated, 20 min before testing, with an i.m. injection of 0.1 ml saline for two tasks and with an i.m. injection of 0.1 ml pilocarpine for the other two tasks. The mean dose of pilocarpine was 0.75 mgrkg. If a task under drug could not be completed within 1 day’s test session, the task was continued on the following day and the same drug dose given 20 min before testing. The monkeys were assigned randomly to a drug, saline, saline, drug, or a saline, drug, drug, saline design. Some monkeys showed mild salivation but no other autonomic or overt behavioural effects of the drug. The numbers of trials to criterion for the two tasks under saline ŽBO3. or drug ŽBO4. were added for each monkey.

The monkeys were then tested Žto 27r30 each. on a series of four visuospatial conditional tasks, using new pairs of coloured objects for each. Again, all the tasks were presented in the same order and the monkeys were pretreated 20 min before testing with either saline or pilocarpine as above. The numbers of trials to criterion for the two tasks under saline ŽVS2. or drug ŽVS3. were added for each monkey. 2.3. Histology When all behavioural testing was completed, the monkeys in Group 2 ŽNBMq CA1. and Group 3 ŽNBMq VDB. were perfused for histological examination, together with two untrained, unoperated monkeys. Following premedication with 0.05 ml ketamine Ž100 mgrml., the monkeys were deeply anaesthetised with 1.0 ml pentobarbitone Ž200 mgrml, i.m.. and perfused transcardially with 250 ml cold phosphate-buffered saline, pH 7.4 ŽPBS. followed by 300 ml cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4. The brains were removed and post-fixed for 2 h in cold 4% paraformaldehyde–PBS after which they were cut coronally into blocks and transferred to 30% sucrose in PBS at 48C. The sucrose medium was changed

Fig. 1. Excitotoxic lesion of hippocampus ŽGroup 2 animal, NBMq CA1.. All sections stained with cresyl violet. Ža. Temporal lobe in unoperated control marmoset. Arrows delimit the extent of CA1. Žc. Temporal lobe in marmoset with excitotoxic lesion of CA1 region of hippocampus. Arrows delimit the lesion. Note the absence of CA1 but lack of damage in CA3–4, dentate, subiculum and cortex. Žb. Pyramidal cells of CA1 region in unoperated marmoset. Žd. Gliosis and absence of pyramidal cells in CA1 region in lesioned marmoset. v Shows approximate position of high power picture below. Bar s 0.25 mm for Žb,d..

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

after 24 h. and the blocks were stored in sucrose at 48C for up to 4 days after which they were cut on a freezing stage microtome to give a series of 40-mm sections throughout each block. Sections were stained for acetylcholinesterase activity ŽAChE. using methods previously described w30,58x, or with cresyl violet. Further sections from each monkey were immunostained with ME20.4IgG and were visualised using biotinylated secondary antibodies and streptavidin–biotin peroxidase as previously described w32x.

125

3. Results 3.1. Histology (see Figs. 1–7) 3.1.1. ME20.4IgG-immunostaining All monkeys in Group 3 ŽNBM q VDB. had similar histology. ME20.4IgG-immunostained neurones were almost totally absent throughout the NBM, the VDB, and the horizontal limb of the diagonal band ŽHDB.. A few cells

Fig. 2. Immunotoxic lesion of the NBM ŽGroup 3 animal, NBM q VDB.. Ža. NBM region beneath anterior commissure in unoperated control marmoset, immunostained with ME20.4IgG. Žb. Picture of immunopositive neurones from area marked above with v. Žc. Cresyl violet stained section adjacent to that shown in Žb.. Žd. NBM region beneath anterior commissure in marmoset with NBMq VDB immunotoxic lesion, immunostained with ME20.4IgG. Že. Immunostained section from area marked above with v. Note the virtual absence of immunopositive neurones in Žd and e.. Žf. Cresyl violet stained section adjacent to that shown in Že.. Note mild gliosis but lack of tissue damage when compared with Žc.. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

126

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

Fig. 3. Immunotoxic lesion of the NBM ŽGroup 2 animal, NBMq CA1.. Ža. NBM region beneath anterior commissure in unoperated control marmoset, immunostained with ME20.4IgG. Žb. Picture of immunopositive neurones from area marked above with v. Žc. Cresyl violet stained section adjacent to that shown in Žb.. Žd. NBM region beneath anterior commissure in marmoset with NBM immunotoxic lesion Žand CA1 excitotoxic lesion., immunostained with ME20.4IgG. Že. Immunostained section from area marked above with v. Note the virtual absence of immunopositive neurones in Žd and e.. Žf. Cresyl violet stained section adjacent to that shown in Že.. Note mild gliosis but lack of tissue damage when compared with Žc.. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

could be seen in the most posterior extremity of the NBM, i.e., in the medial and lateral medullary lamellae between the putamen and the internal or external portions of the globus pallidus. The weakly immunopositive cells of the basal ganglia and the ME20.4IgG-immunopositive cerebellar Purkinje cells were intact. All monkeys in Group 2 ŽNBMq CA1. had similar histology. ME20.4IgG-immunostaining was almost completely absent in the NBM but was normal in the VDB and

the HDB. The excitotoxic lesions of CA1 did not cause retrograde damage to the cholinergic projection to the hippocampus. A few cells could be seen in the most posterior extremity of the NBM, i.e., in the medial and lateral medullary lamellae between the putamen and the internal or external portions of the globus pallidus. The weakly immunopositive cells of the basal ganglia and the ME20.4IgG-immunopositive cerebellar Purkinje cells were intact.

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

127

Fig. 4. VDB area in Group 2 animal ŽNBMq CA1.. Ža. VDB region in unoperated control marmoset, immunostained with ME20.4IgG. Žb. Picture of immunopositive neurones from area marked above with v. Žc. Cresyl violet stained section adjacent to that shown in Žb.. Žd. VDB region in marmoset with NBM q CA1 lesions, immunostained with ME20.4IgG. Že. Immunostained section from area marked above with v. Note the intact VDB. Žf. Cresyl violet stained section adjacent to that shown in Že.. Note that there is no difference when compared with Žc.. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

3.1.2. AChE staining The pattern of loss of the AChE staining in all lesioned monkeys was as predicted from the known target areas of the component parts of the cholinergic projections in primates w34,35x. Monkeys with immunotoxic lesions of the NBM Žand excitotoxic lesions of CA1. had undetectable, or barely detectable, levels of AChE staining in all neocortical areas Žincluding parahippocampal gyrus. but some staining was evident in medial frontal cortex, especially below the corpus callosum, and in the most ventral part of

cingulate cortex adjacent to the corpus callosum. There was marked reduction in AChE staining in the lateral amygdala. AChE staining was normal in the entorhinal cortex, dentate gyrus and subiculum. There was some loss of AChE staining in those parts of CA1 which remained extant Žthough gliotic and lacking pyramidal cells. following the CA1 lesion. Monkeys with immunotoxic lesions of the NBM q VDB had undetectable, or barely detectable, levels of AChE staining throughout the cortex and hippocampal formation.

128

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

Fig. 5. Immunotoxic lesion of the VDB ŽGroup 3 animal, NBMq VDB.. Ža. VDB region in unoperated control marmoset, immunostained with ME20.4IgG. Žb. Picture of immunopositive neurones from area marked above with v. Žc. Cresyl violet stained section adjacent to that shown in Žb.. Žd. VDB region in marmoset with NBMq VDB immunotoxic lesions, immunostained with ME20.4IgG. Že. Immunostained section from area marked above with v. Note the absence of immunopositive neurones in Žd and e.. Žf. Cresyl violet stained section adjacent to that shown in Že.. Note mild gliosis but lack of tissue damage when compared with Žc.. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

A very small amount of staining was detected in the medial frontal cortex below the corpus callosum and in a small section of the entorhinal cortex adjacent to the amygdala. There was marked reduction in AChE staining in the lateral amygdala.

but no damage in the basal ganglia. In monkeys with NBMq VDB lesions there was detectable gliosis in the NBM, VDB and anterior commissure but no consistent damage elsewhere. There was no evidence of tissue loss or necrosis in the NBM or VDB in any monkey.

3.1.3. Cresyl Õiolet staining Monkeys with NBMq CA1 lesions all showed tissue destruction in the CA1 field of the hippocampus. There was mild gliosis in the NBM and the anterior commissure

3.2. CognitiÕe testing Histograms of the mean learning scores Ži.e., the number of trials up to, but excluding those in, criterion"

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

129

Fig. 6. AChE staining in temporal lobe ŽGroup 2 animal, NBM q CA1.. Ža. Temporal lobe in unoperated marmoset. Arrows delimit CA1 region. Žb. High power picture of AChE staining in CA1 in region marked v above. Žc. High power of AChE staining in temporal cortex in region marked w above. Žd. Temporal lobe in animal with NBMq CA1 lesions. Arrows delimit the extent of excitotoxic CA1 lesion. Broad arrow delimits the boundary of entorhinal cortex Žmedial. and neocortex Žlateral.. e: High power picture of AChE staining in CA1 in region marked v above. Žf. High power picture of AChE staining in temporal cortex in region marked w above. In the lesioned animal, note the lack of AChE staining in temporal neocortex but intact staining in entorhinal cortex. There is shrinkage of the CA1 region and reduced AChE staining in the CA1 tissue that remains, although staining in the subiculum and dentate gyrus is normal. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

S.E.M.. for each group on each task are shown in Figs. 8 and 9. Data were analysed for within and between group effects using ANOVA Žsee Table 1.. Groups of tasks of the same type Že.g., post-operative learning of simple visual discriminations. were analysed in a repeated-measures general linear model design to obviate the effects of multiple comparisons, followed by a posteriori between group comparisons using Fisher’s LSD test with an accep-

tance level of p - 0.05 Žsee Figs. 8 and 9.. Between group effects on single tasks were compared using one-way ANOVA with a posteriori Fisher’s LSD tests and an acceptance level of p - 0.05 Žsee Figs. 8 and 9.. Since one monkey was removed from Group 2 for failing to qualify for progressing to stage 2 ŽSection 2., the group compositions were: Group 1, n s 6; Group 2, n s 5; Group 3, n s 6.

130

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

Fig. 7. AChE staining in temporal lobe ŽGroup 3 animal, NBMq VDB.. Ža. Temporal lobe in unoperated marmoset. Žb. High power picture of AChE staining in CA1 in region marked v above. Žc. High power of AChE staining in temporal cortex in region marked w above. Žd. Temporal lobe in animal with NBMq VDB lesions. Že. High power picture of AChE staining in CA1 in region marked v above. Žf. High power picture of AChE staining in temporal cortex in region marked w above. In the lesioned animal, note the lack of AChE staining throughout the entire temporal lobe. Bar s 1 mm for Ža,d.. Bar s 0.25 mm for Žb,c,e,f..

3.2.1. Pre-operatiÕe testing The three groups did not differ in their ability to learn the pre-operative simple visual discrimination task ŽB vs. S, Fig. 8A.. 3.2.2. Effects of stage 1 surgery Group 2 ŽNBM. and Group 3 ŽNBMq VDB. were significantly impaired on retention of the pre-operatively learnt simple visual discrimination ŽB vs. S Ret, Fig. 8A.,

compared with Group 1 ŽControl.. Group 3 ŽNBM q VDB. did not differ significantly from Group 2 ŽNBM.. Group 2 ŽNBM. and Group 3 ŽNBMq VDB. were severely impaired at learning the black object discrimination ŽBO1, Fig. 8B., compared with controls, with all lesioned monkeys failing at 300 trials. Group 3 ŽNBM q VDB. Žbut not Group 2 ŽNBM.. was impaired relative to controls on pebble object discrimination ŽPO1, Fig. 8B.. The mean learning score of Group 2 ŽNBM. fell between

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

131

Fig. 8. Learning scores for tasks used preoperatively and after stage 1 surgery. C s Control group, n s 6; N s NBM lesion group, n s 5; NV s NBM q VDB lesion group, n s 6. ŽA. B vs. S, simple cloured object discrimination Ž‘ballerina’ vs. ‘soldier’. learnt preoperatively; Ret, retention learning; VS, visuospatial conditional task; MOS, mixed object discrimination and visuospatial task. B: BO, black object discrimination task; PO, pebble object discrimination task; CO, coloured object discrimination task; given shortly after surgery ŽFirst time. or following intervening tasks ŽSecond time.. U p - 0.05; UU p - 0.01; UUU p - 0.001, comparing Groups N and NV with Group C, Fisher’s LSD test following ANOVA Žsee Table 1..

those of the other two groups and was not significantly different from either. Group 3 ŽNBMq VDB. was also impaired on coloured object discrimination ŽCO1, Fig. 8B. compared with controls and compared with Group 2 ŽNBM.. There was no difference in learning ability between Group 2 ŽNBM. and Group 1 ŽControl. on this task. From this series of tasks it is clear the black object discrimination is particularly sensitive to lesions of NBM alone. Both Group 2 ŽNBM. and Group 3 ŽNBM q VDB. were impaired compared with Group 1 ŽControl. on the visuospatial conditional task ŽVS1, Fig. 8A.. The difference between Group 2 ŽNBM. and Group 3 ŽNBM q VDB. was not significant. The impairment was particularly marked in Group 3 ŽNBM q VDB. where five of the six

monkeys failed at 400 trials. This is consistent with our earlier finding that monkeys with VDB lesions find visuospatial conditional discriminations very difficult to learn w58x. Group 3 ŽNBMq VDB. was significantly impaired compared with Group 1 ŽControl. and Group 2 ŽNBM. on the mixed simple and visuospatial conditional discrimination ŽMOS, Fig. 8A.. Monkeys in Group 3 were impaired on both components of the MOS task ŽA vs. B, p - 0.01; CC left, p - 0.05., but monkeys in Group 2 were not impaired on either component. This shows that monkeys in Group 2 were able to learn two different types of task concurrently. Group 3 ŽNBMq VDB. was significantly impaired on all of the second three post-operative simple visual discriminations ŽBO2, PO2, CO2, Fig. 8B. compared with

132

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

Fig. 9. Learning scores for tasks used after stage 2 surgery. C s Control group, n s 6; NCA1s NBM lesion group to which CA1 lesion has been added, n s 5; NV s NBM q VDB lesions group, n s 6. A. Retention of black object ŽBO., pebble object ŽPO. or coloured junk object ŽCO. discriminations. B. Effect of pilocarpine on learning impairments. BO3, sum of two black object discriminations learnt under saline; BO4, sum of two black object discriminations learnt under pilocarpine. VS2, sum of two visuospatial conditional discriminations learnt under saline; VS3, sum of two visuospatial conditional discriminations learnt under pilocarpine. U p - 0.05; UU p - 0.01; UUU p - 0.001, comparing Groups NCA1 and NV with Group C, Fisher’s LSD following ANOVA Žsee Table 1.. a p - 0.05; aa p - 0.01, comparing performance under saline with performance under pilocarpine for each group using matched-pairs t-tests.

Group 1 ŽControl. and Group 2 ŽNBM.. Group 2 ŽNBM. did not differ from Group 1 ŽControl.. So, by this stage in the experiment the monkeys with double lesions of NBM q VDB remained impaired at acquisition of simple visual discrimination tasks while those with NBM lesions alone had recovered their ability to learn these discriminations. One animal in Group 2 did not progress to stage 2 because it had shown learning scores comparable to those of the monkeys in Group 3 throughout testing and had failed seven of the nine tasks presented up until stage 2 surgery. This was the only animal in Group 2 to have shown the behavioural syndrome 2–4 weeks after surgery which was exhibited by five out of six of the monkeys in

Group 3. Nonetheless when the brain of this animal was examined it had a lesion of the NBM but not the VDB and showed a loss of AChE staining in the neocortex but not in the allocortex or hippocampus. Its pattern of ME20.4IgGimmunostaining and AChE staining was indistinguishable from that seen in the other animals in Group 2. 3.2.3. Effects of stage 2 surgery Following stage 2 surgery, the monkeys were tested on retention of the three simple visual discrimination tasks learnt prior to this surgery ŽBO2 Ret, PO2 Ret, and CO2 Ret, Fig. 9A.. Group 2 ŽNBMq CA1. was significantly impaired on BO2 Ret compared with Group 1 ŽControl.

R.M. Ridley et al.r Brain Research 836 (1999) 120–138 Table 1 ANOVA of cognitive tasks in order of task presentation Task

Effect

F

Pre-operatiÕe task B vs. S

No group effect

F Ž2,14. s 0.5, ps 0.6

Group Group Task Group=Task Group Group Group Task

F Ž2,14. s8.37, p- 0.01 F Ž2,14. s 36.74, p- 0.001 F Ž2,28. s80.49, p- 0.001 F Ž4,28. s 4.21, p- 0.01 F Ž2,14. s9.51, p- 0.01 F Ž2,14. s 4.1, p- 0.05 F Ž2,14. s9.7, p- 0.01 F Ž2,28. s 5.74, p- 0.01

Group Task Group=Task Group Group

F Ž2,14. s16.73, p- 0.001 F Ž2,28. s 22.11, p- 0.001 F Ž4,28. s12.74, p- 0.001 F Ž2,14. s 4.58, p- 0.05 F Ž2,14. s11.22, p- 0.001

Stage 1 surgery Retention B vs. S BO1, PO1, CO1

VS1 MOS BO2, PO2, CO2

Stage 2 surgery BO2, PO2, CO2 Žretentions. BO3 ŽSal. VS2 ŽSal.

133

treatment with saline Ž t Ž5. s 2.88, p - 0.05. Žsee Fig. 9B.. The pilocarpine had no significant effect on Group 1 ŽControl. or Group 2 ŽNBM q CA1.. 4. Discussion

and Group 3 ŽNBMq VDB.. Group 2 ŽNBMq CA1. was also significantly impaired on PO2 Ret compared with Group 1 ŽControl. but did not differ from Group 3 ŽNBM q VDB.. The groups did not differ on CO2 Ret. Groups 3 ŽNBMq VDB. and 1 ŽControl. did not differ on any task. Group 2 ŽNBMq CA1. and Group 3 ŽNBM q VDB. were significantly impaired relative to Group 1 ŽControl. on black object discrimination new learning ŽBO3 ŽSal. in Table 1 and Fig. 9B.. Group 2 ŽNBMq CA1. was significantly different from Group 1 ŽControl. and Group 3 ŽNBMq VDB. on visuospatial conditional discrimination new learning ŽVS2 ŽSal. in Table 1 and Fig. 9B.. Group 1 ŽControl. and Group 3 ŽNBM q VDB. did not differ significantly on VS2 using the restricted Fisher’s LSD test. The effects of administering pilocarpine shortly before testing on either simple visual discrimination learning using black objects, or on visuospatial conditional learning was assessed using a matched-pairs t-test for each group on each task. For the simple visual discrimination, there was a significant reduction in learning scores in all three groups following treatment with pilocarpine when compared with scores after treatment with saline: ŽGroup 1 ŽControl., t Ž5. s 2.8, p - 0.05; Group 2 ŽNBMq CA1., t Ž4. s 4.98, p - 0.01; Group 3 ŽNBMq VDB., t Ž5. s 2.79, p - 0.05. Žsee Fig. 9B.. One-way ANOVA of the difference scores between saline and pilocarpine pretreatment showed a significant group effect Ž F Ž2,14. s 4.1, p - 0.05.. A posteriori comparisons showed that the effect of pilocarpine was significantly greater in Group 2 ŽNBM q CA1. and Group 3 ŽNBM q VDB. than in Group 1 ŽControl. Ž p - 0.05, in each case., indicating the drug has a greater effect in lesioned than in unlesioned monkeys. For the visuospatial conditional task, there was a significant reduction in the mean learning score of Group 3 ŽNBMq VDB. when compared with the mean score after

4.1. The effect of NBM q VDB lesions This experiment shows that destruction of the cholinergic projections to the cortex and hippocampus results in severe learning impairments on object discriminations and on visuospatial conditional discrimination tasks. We have previously shown that destruction of the cholinergic projection from the VDB to the hippocampus and associated allocortical areas produces an impairment on visuospatial conditional learning but not on object discrimination learning, irrespective of whether the stimuli differ in few or many features w51x. Thus the impairment in this experiment on visuospatial conditional discrimination learning following NBMq VDB lesion can be ascribed largely to the contribution of the VDB lesion. In this and previous experiments w18,51x, NBM lesions produced impairments on object discrimination learning which were most marked on tasks where the stimuli differed in a limited number of attributes Že.g., black objects or pebbles rather than multicoloured plastic toys.. The effects of NBM lesions in this and the previous studies were transient. The substantial and persistent effect of NBM q VDB lesions on object discriminations is therefore a consequence of the additive effect of both lesions rather than attributable to the effect of either lesion alone. This additive effect was two-fold; it produced an effect on coloured object discrimination learning which was more marked than after NBM lesion alone and it prevented the recovery on black object discrimination learning seen after NBM lesion alone. This suggests that the NBM and VDB Žand, by implication, their target areas. act ‘in parallel’ to make complementary Žbut probably different. contributions to object discrimination learning. 4.2. Comparison of the cognitiÕe effects of lesions of the NBM and r or VDB with the effects of lesions of their target areas Immunotoxic lesions of the VDB w51x result in a pattern of impairments which is virtually identical to that seen following fornix transection w55,59x or excitotoxic lesions of the CA1 w57,60x, supporting the view that the rising cholinergic projections maintain the functions of their target areas. Similarly, immunotoxic lesions of the NBM might be expected to produce impairments that resemble those resulting from lesions of the target area of the NBM, namely the temporal or frontal neocortical areas. Few cortical ablation studies have been carried out in the marmoset although suction ablations or excitotoxic lesions of parts of the frontal cortex have not produced major

134

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

impairments on acquisition of object discriminations w13,53,54x. On the basis of studies in macaques Žfor review, see Dean w12x and Gross w25x. impairment on visual discrimination learning would be anticipated in marmosets with lesions of the inferotemporal cortex. We have yet to study the extent of this effect in marmosets. 4.3. Demonstration that the cognitiÕe effects of immunotoxic lesions occur because of cholinergic depletions to the neocortex and hippocampus That immunotoxic lesions of the NBM or VDB produce cognitive impairments that are dependent on the loss of the cholinergic projections to the target areas can be demonstrated by the ability of a cholinergic agonist to restore cognitive function. We predicted that the cholinergic agonist, pilocarpine, would restore visuospatial learning in monkeys with NBMq VDB lesions but not in monkeys with NBM q CA1 lesions. Visuospatial discrimination learning depends on the integrity of the hippocampus. VDB lesions produce a loss of cholinergic influence in the hippocampus, potentially restorable with pilocarpine treatment, whereas the CA1 lesions involve a loss of tissue in the hippocampus, which would not be treatable with pilocarpine. This prediction was borne out in this experiment since only the monkeys with combined NBM q VDB lesions showed significant improvement on visuospatial conditional learning when treated with pilocarpine. There was no significant difference between the learning scores of the monkeys with NBM q VDB lesions treated with pilocarpine and those of the control monkeys, with or without pilocarpine. In fact, the mean learning score for the NBM q VDB group, treated with pilocarpine, was below the mean learning scores for the control group, with or without pilocarpine. This completely restorative effect of pilocarpine indicates that, if the immunotoxin had had any effect on non-cholinergic systems, such an effect did not contribute to the visuospatial conditional learning impairment. In previous experiments, we found that pilocarpine produced a significant improvement in acquisition of the visuospatial task by monkeys with ibotenic acid lesions of the VDB w58x or fornix transection w59x, although performance remained somewhat worse than that of control monkeys. In the current experiment, pilocarpine also produced a significant improvement on object discrimination learning in all three groups. The effect of pilocarpine in control monkeys was unexpected but small. The improvement seen after pilocarpine treatment in monkeys with NBM q CA1 and NBMq VDB lesions was greater than the improvement after pilocarpine seen in control monkeys, suggesting that cholinergic agonists are particularly efficacious under conditions of cholinergic dysfunction. The CA1 lesion can be considered to be irrelevant to object discrimination learning since, in itself, it has no effect on the acquisition of this type of task w60x.

4.4. Implications of the transience of the cognitiÕe effects of NBM lesions The transience of the acquisition impairment on object discrimination tasks following NBM lesions, in comparison to the persistence of the impairment following NBM q VDB lesions, cannot be explained by differential neurochemical recovery within the two lesion groups. Histological examination, after the termination of behavioural testing, showed barely detectable AChE staining throughout the target area of the NBM in monkeys with NBM q CA1 lesions, as well as throughout the target areas of the NBM and VDB in monkeys with NBM q VDB lesions. Another suggestion would be that the recovery of function following NBM lesions occurs because the target area of the VDB, namely the hippocampal formation, is able to take over the cognitive analysis usually performed by the neocortex. In contrast, the persistence of the effect of VDB lesion on visuospatial conditional discrimination learning w51,58x suggests that the neocortex is not able to take over the function of the hippocampus. Such asymmetry of compensation could occur if the hippocampus had a ‘complex’ capability which was able to perform ‘simple’ cognitive processing when necessary, but the neocortex had a more ‘simple’ capability which was unable to perform the cognitively more specialised function of the hippocampus. The possible nature of that specialised function of the hippocampal formation is discussed below, in conjunction with a consideration of the role of the hippocampus in humans and monkeys. 4.5. Effects of lesions on retention Monkeys with lesions of the NBM or NBM q VDB were impaired on post-operative retention of a coloured object discrimination first learnt before surgery, with a total retention interval of 1–3 weeks. All unoperated, control monkeys showed substantial positive savings, whereas all except one monkey with an NBM lesion and one with an NBM q VDB lesion showed negative savings on this retention task. This indicates that the lesioned monkeys had both a retention and an acquisition impairment. The results of the current experiment help to clarify our previous findings on the effect of NBM lesions on retention of discriminations learnt prior to surgery. In our first attempt to make immunotoxic lesions of the NBM in monkeys w18x, we did not obtain significant retention deficits in the groups with NBM lesions although a few lesioned monkeys had negative savings or failed to relearn a retention task. In our second experiment w51x, when our technique had become more established, significant retention deficits were found in monkeys with NBM lesions and many monkeys had negative savings or failed to relearn a retention task. Thus immunotoxic lesions of the NBM can result in retention deficits for information acquired before the surgery Žretrograde amnesia. though this effect is idiosyncratic. Although there was a tendency, in all these

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

experiments, for those monkeys with retention deficits to be those with the greatest learning impairments, it was not possible to identify aspects of the lesion or of the training performance which would predict those monkeys with retention deficits. After the CA1 lesions were added to the NBM lesions, the monkeys in Group 2 ŽNBM q CA1. showed negative savings, as well as impaired learning scores relative to control monkeys, on retention of a black object discrimination and a pebble discrimination, but variable savings on retention of a coloured object discrimination. This suggests that these monkeys had both a retention and an acquisition deficit on the black and pebble object discriminations. Control monkeys showed near perfect retention, and monkeys with combined lesions of the NBMq VDB showed positive savings on these tasks. These tasks were, however, retentions of discriminations learnt before stage 2 surgery Žthe CA1 lesions. but after stage 1 surgery ŽNBM, or NBMq VDB lesions.. The results in monkeys with NBM q CA1 lesions demonstrate that removal of hippocampal tissue from those monkeys that had learnt object discriminations without impairment but in the absence of cholinergic innervation to the cortex, results in retrograde amnesia for that information. This contrasts with the effects of CA1 lesions alone that do not impair retention Žor acquisition. of this type of information w60x. CA1 lesions result in impairment on acquisition, and retention, of visuospatial discriminations first learnt prior to surgery, but not for retention of this type of discrimination learnt Žwith impairment. after surgery w60x. The supposition must be that, when the hippocampus was involved in acquisition of information, it is required for its retention, but that when the hippocampus was not involved in acquisition, retention of that information is not affected by removal of the hippocampus. By contrast, the positive savings shown by monkeys with NBM q VDB lesions on discriminations first learnt Žwith impairment. after stage 1 surgery, indicates that these lesions Žlike CA1 lesions. do not cause excessive ‘forgetting’ of information once it has been acquired. 4.6. Information processing in the neocortex and the hippocampal formation Immunotoxic lesions of the NBM and VDB provide a method of ‘switching off’ larger areas of the brain than could be achieved by suction ablation of the target areas while retaining the survival and behavioural competence of the animal. The results of this experiment show that the NBM and VDB, and by implication their target areas, co-operate to sustain mnemonic functions. However, the specific effects of VDB lesions alone, and the different but transient effects of NBM lesions alone, suggest that this co-operation is not merely additive but consists of complementary cognitive functions able to substitute for each other. In order to consider the ways in which the target areas of the NBM and VDB may interact, it is helpful to

135

take a broader view of the functions of these areas, particularly the hippocampus, than is demonstrated by these experiments alone. A distinction can be drawn between the mnemonic functions of the hippocampus and of the neocortex in humans. The hippocampus has been particularly implicated in the episodic memory loss of clinical amnesia w45,46x although the majority of cases of human temporal lobe amnesia have sustained additional damage outside this precise area w64x. The observation, in well documented cases, that fornix transection in man produces amnesia w22x strengthens the argument for a pivotal role for the hippocampus in memory formation, but the occurrence of very dense amnesia in patients with further damage in the temporal lobe implies that damage to areas outside the fornical–hippocampal system contributes to these denser amnesic syndromes. Damage to the medial and ventral cortex of the temporal lobe, in the absence of hippocampal damage, leads to semantic dementia w27,29x andror visual agnosia w37x, but does not cause episodic amnesia. Semantic knowledge can be acquired by people with impaired episodic memory w69x indicating the separability of these two cognitive modules. But the dense amnesia seen in patients with large medial temporal lobe lesions suggests that temporal cortical damage may contribute to clinical anterograde amnesia by damaging the semantic base upon which the formation of episodic memories depends. That ‘comprehension’ should contribute to ‘recollectability’ of events seems reasonable. That the severity of amnesia depends on the size of the hippocampal and temporal cortical lesion may be observationally true Žand confirmed in monkey studies w75x. but this need not imply that these two areas act in an equivalent ‘mass action’ fashion, since each area could sustain its own cognitive ‘processing’ which contributes, in its separate way, to the psychological ‘product’ of memory performance. Descriptions of the different functions of the hippocampal formation and the temporal neocortex in monkeys used to be based on the belief that the hippocampus was crucial to correct performance of the delayed non-matching to sample ŽDNMS. task, whereas it was assumed that lesions confined to other areas within the temporal lobe would not affect performance of this task. Fornix transection, or ablations, confined as far as possible to the hippocampus, led to modest impairments in DNMS performance w20,74x. Lesions which included the hippocampus and surrounding areas produced larger impairments on this task, but this was initially attributed to involvement of the amygdala rather than the temporal cortex w38x. Amnesic patients had difficulty performing the DNMS task was which was therefore thought to tax episodic memory w65x. Impairment on this task by monkeys was taken to indicate that these monkeys were also amnesic w38x. In contrast, lesions of inferotemporal cortex in monkeys led to impairment on acquisition of visual object and pattern discriminations Žfor

136

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

review, see Dean w12x.. These tasks might tax semantic knowledge w48x or visual gnosis w25x. These views of the cognitive requirement for DNMS performance, and of the lesion that most affects that performance, have now changed. The focus of the DNMS impairment has been found to be the perirhinal cortex w36x although lesions of this area also produce impairments on other visual tasks, including concurrent discrimination learning under certain conditions w8,9x. This area is seen as subserving object identification rather than episodic memory w9x. However, while impaired object identification, in the case of perirhinal lesions, may interfere with DNMS performance, object identification is not the only cognitive skill required for performance of the DNMS task. The different effects of fornix transection on slightly different versions of delayed matching and nonmatching tasks suggest that spatial analysis of the test array w21x and the ability to use information about non-reward to predict future reward w23,24x are skills which the hippocampal formation may contribute to performance of DNMS tasks. The demise of DNMS performance in macaques as a model of primate episodic memory permits a fresh exploration of the contribution which temporal lobe areas make to mnemonic function. Tulving and Markowitsch w68x argue that episodic memory is an elaborated form of memory Žassisted by semantic knowledge. but that semantic knowledge can accumulate without the information being processed through episodic memory. This view is consistent with the neuroanatomical structure of the temporal lobe in that it can be envisaged that visual information from the occipital lobe can be processed perceptually, and then semantically, in progressively more anterior areas of the temporal cortex, before being subjected to further processing in the hippocampal formation which allows episodic memories to be formed. These authors also argue that episodic memory is ‘special’ because it is the only form of memory that is specifically about the past. In contrast, semantic memory is about how things are now known to be as a consequence of previous experience, and procedural skills are demonstrable only by current performance. Episodic memory is ‘special’ because it is the only form of memory that is dependent on ‘secondary mental representation’ rather than ‘primary mental representation’ w44x. Primary mental representation comprises perceptions and factual information in conscious working memory and results from ‘single channel’ cognitive processing. Secondary mental representation requires ‘parallel’ cognitive processing and allows recollection of the past, imagination of the future, or construction of hypothetical alternatives to the present, to be processed in an additional ‘on-line’ system that produces the experiential quality of conscious recollection and imagination which is similar to, but different from, the primary representation of the present. It has been argued Že.g., Tulving and Markowitsch w68x. that only humans have episodic memory and that episodic

memory is dependent on the integrity of the hippocampus. What then is the hippocampus for in animals? In rodents there is evidence implicating the hippocampus in the alteration of previously acquired stimulus–response associations w15x, spatial navigation w39,42x and contextual or conditional responding w10x. These apparently very different products of hippocampal activity share the feature that they all require more than one response to be appropriate to a single external stimulus under different conditions. ŽThe spatial domain, as opposed to the objects in it, is a single entity such that more than one spatially determined response requires flexible responding within the one domain.. The tasks which rodents with hippocampal lesions find difficult cannot be done with ‘single channel’ processing but require some form of ‘parallel’ processing. In primates, the separation of ‘parallel’ channels may be more explicit. Tasks involving memory for the location of objects Žwhich monkeys with hippocampal lesions find difficult w43x. require the animal to recollect the previous spatial arrangement of objects Ža secondary representation. without interference from perception of the current spatial layout Ža primary representation. in order to solve the task. Alternation Žwhich monkeys with hippocampal ablations also find difficult w33x. requires the animal to use memory of the place- or stimulus-association with reward on the previous trial Ža secondary representation. to predict nonreward on the current trial, i.e., separately, and without interference, from the motivational value of that reward Ža primary representation.. In humans, the separation of the parallel channels into different time domains would allow the past to be represented as ‘another time and place’ Žepisodic memory.. ŽSuch separation would also allow the future to be regarded as another ‘potential time and place’ and consideration of the possible events in that time and place would permit explicit planning.. On the basis of the current experiments, our earlier work, and comparison with work in macaques, we suggest that monkeys with lesions of the VDBrfornixrhippocampal axis are impaired on tasks which require ‘parallel’ processing Žsuch as the conditional visuospatial task in which each stimulus sometimes is and sometimes is not rewarded depending on its position., but that they are not impaired on tasks which require only ‘single channel’ processing Žsuch as simple discriminations where one stimulus is always rewarded.. Monkeys with lesions of the NBMrcortical axis are initially impaired on tasks requiring ‘single’ processing, but processing by the hippocampus can substitute for this function. Removal of the VDB or hippocampus reinstates impairments on simple discriminations in monkeys with previous lesions of the NBMrcortical system by disrupting this substituted function. Acknowledgements This work was carried out under a Medical Research Council Programme Grant awarded to R.M.R. and H.F.B.

R.M. Ridley et al.r Brain Research 836 (1999) 120–138

This team is a component of the MRC Brain Behaviour and Neuropsychiatry Cooperative Group, in Cambridge. References w1x T.G. Aigner, M. Mishkin, The effects of physostigmine and scopolamine on recognition memory in monkeys, Behav. Neural Biol. 45 Ž1986. 81–87. w2x T.G. Aigner, S.J. Mitchell, J.P. Aggleton, M.R. DeLong, R.G. Struble, D.L. Price, G.L. Wenk, K.D. Pettigrew, M. Mishkin, Transient impairment of recognition memory following ibotenic-acid lesions of the basal forebrain in macaques, Exp. Brain Res. 86 Ž1991. 18–26. w3x T.G. Aigner, D.L. Walker, M. Mishkin, Comparison of the effects of scopolamine administered before and after acquisition in a test of visual recognition memory in monkeys, Behav. Neural Biol. 55 Ž1991. 61–67. w4x H.F. Baker, R.M. Ridley, Use of the common marmoset Ž Callithrix jacchus . in psychopharmacological research. in: M.H. Joseph, L.J. Waddington ŽEds.., Working Methods in Neuropsychopharmacological Research, Manchester Univ. Press, Manchester, 1986, pp. 41–73. w5x M.G. Baxter, D.J. Bucci, T.J. Sobel, M.J. Williams, L.K. Gorman, M. Gallagher, Intact spatial learning following lesions of the basal forebrain cholinergic neurons, NeuroReport 7 Ž1996. 1417–1420. w6x M.G. Baxter, M. Gallagher, Cognitive effects of selective loss of basal forebrain cholinergic neurons: implications for cholinergic therapies of Alzheimer’s disease. in: J.D. Brioni, M.W. Decker ŽEds.., Pharmacological Treatments of Alzheimer’s Disease: Molecular and Neurobiological Foundations, Wiley-Liss, 1997, pp. 87–103. w7x J. Berger-Sweeney, S. Heckers, M.-M. Mesulam, R.G. Wiley, D.A. Lappi, M. Sharma, Differential effects on spatial navigation of immunotoxin-induced cholinergic lesions of the medial septal area and nucleus basalis magnocellularis, J. Neurosci. 14 Ž1994. 4507– 4519. w8x M.J. Buckley, D. Gaffan, Impairment of visual object-discrimination learning after perirhinal cortex ablation, Behav. Neurosci. 111 Ž1997. 467–475. w9x M.J. Buckley, D. Gaffan, Learning and transfer of object–reward associations and the role of the perirhinal cortex, Behav. Neurosci. 112 Ž1998. 15–23. w10x N.J. Cohen, H. Eichenbaum, Memory, Amnesia and the Hippocampus, MIT Press, Cambridge, MA, 1993. w11x T.J. Crow, I.G. Grove-White, An analysis of the learning deficit following hyoscine administration to man, Br. J. Pharmacol. 49 Ž1973. 322–327. w12x P. Dean, Effects of inferotemporal lesions on the behavior of monkeys, Psychol. Bull. 83 Ž1976. 41–71. w13x R. Dias, T.W. Robbins, A.C. Roberts, Primate analogue of the Wisconsin Card Sorting Test: effects of excitotoxic lesions of the prefrontal cortex in the marmoset, Behav. Neurosci. 110 Ž1996. 872–886. w14x W.A. Dornan, A.R. McCampbell, G.P. Tinkler, L.J. Hickman, A.W. Bannon, M.W. Decker, K.L. Gunther, Comparison of site-specific injections into the basal forebrain on water maze and radial arm maze performance in the male rat after immunolesioning with 192 IgG saporin, Behav. Brain Res. 82 Ž1996. 93–101. w15x R.J. Douglas, The hippocampus and behavior, Psychol. Bull. 67 Ž1967. 416–442. w16x S.B. Dunnett, B.J. Everitt, T.W. Robbins, The basal forebrain-cortical cholinergic system: interpreting the functional consequences of excitotoxic lesions, Trends Neurosci. 14 Ž1991. 494–501. w17x H.C. Fibiger, Cholinergic mechanisms in learning, memory and dementia: a review of recent evidence, Trends Neurosci. 14 Ž1991. 220–223. w18x A. Fine, C. Hoyle, C.J. Maclean, T.L. LeVatte, H.F. Baker, R.M. Ridley, Learning impairments following injection of a selective

w19x

w20x

w21x

w22x w23x

w24x

w25x

w26x

w27x

w28x

w29x

w30x

w31x

w32x

w33x w34x

w35x

w36x

w37x w38x

137

cholinergic immunotoxin, ME20.4 IgG-saporin, into the basal nucleus of Meynert in monkeys, Neuroscience 81 Ž1997. 331–343. C.D. Frith, J.T.E. Richardson, M. Samuel, T.J. Crow, P.J. McKenna, The effects of intravenous diazepam and hyoscine upon human memory, Q. J. Exp. Psychol. 36A Ž1984. 133–144. D. Gaffan, Recognition impaired and association intact in the memory of monkeys after transection of the fornix, J. Comp. Physiol. Psychol. 86 Ž1974. 1100–1109. D. Gaffan, The role of the hippocampus–fornix–mammillary system in episodic memory. in: L.R. Squire, N. Butters ŽEds.., Neuropsychology of Memory, Guilford Press, New York, 1992, pp. 336–346. D. Gaffan, E.A. Gaffan, Amnesia in man following transection of the fornix. A review, Brain 114 Ž1991. 2611–2618. D. Gaffan, R.C. Saunders, E.A. Gaffan, S. Harrison, C. Shields, M.J. Owen, Effects of fornix transection upon associative learning in monkeys: role of the hippocampus in learned action, Q. J. Exp. Psychol. 36B Ž1984. 173–222. D. Gaffan, C. Shields, S. Harrison, Delayed matching by fornixtransected monkeys: the sample, the push and the bait, Q. J. Exp. Psychol. 36B Ž1984. 305–317. C.G. Gross, Visual functions of inferotemporal cortex. in: R. Jung ŽEd.., Handbook of Sensory Physiology, Part 3B, Vol. 7, SpringerVerlag, Berlin, 1973, pp. 451–482. J.A. Harder, H.F. Baker, R.M. Ridley, The role of the central cholinergic projections in cognition: implications of the effects of scopolamine on discrimination learning by monkeys, Brain Res. Bull. 45 Ž1998. 319–326. J.R. Hodges, K. Patterson, S. Oxbury, E. Funnell, Semantic dementia: progressive fluent aphasia with temporal lobe atrophy, Brain 115 Ž1992. 1783–1806. E. Irle, H.J. Markowitsch, Basal forebrain-lesioned monkeys are severely impaired in tasks of association and recognition memory, Ann. Neurol. 22 Ž1987. 735–743. N. Kapur, D. Ellison, A.J. Parkin, N.M. Hunkin, E. Burrows, S.A. Sampson, E.A. Morrison, Bilateral temporal lobe pathology with sparing of medial temporal lobe structures: lesion profile and pattern of memory disorder, Neuropsychologia 32 Ž1994. 23–38. G.B. Koelle, The histochemical identification of acetylcholinesterase in cholinergic, adrenergic and sensory neurons, J. Pharmacol. Exp. Ther. 114 Ž1955. 167–184. G. Leanza, O.G. Nilsson, R.G. Wiley, A. Bjorklund, Selective ¨ lesioning of the basal forebrain cholinergic system by intraventricular 192 IgG-saporin: behavioural, biochemical and stereological studies in the rat, Eur. J. Neurosci. 7 Ž1995. 329–343. C.J. Maclean, H.F. Baker, A. Fine, R.M. Ridley, The distribution of p75 neurotrophin receptor-immunoreactive cells in the forebrain of the common marmoset Ž Callithrix jacchus ., Brain Res. Bull. 43 Ž1997. 197–208. H. Mahut, Spatial and object reversal learning in monkeys with partial temporal lobe ablations, Neuropsychologia 9 Ž1971. 409–424. M.-M. Mesulam, E.J. Mufson, A.I. Levey, B.H. Wainer, Cholinergic innervation of cortex by the basal forebrain: cytochemistry and cortical connections of the septal area, diagonal band nuclei, nucleus basalis Žsubstantia innominata. and hypothalamus in the rhesus monkey, J. Comp. Neurol. 214 Ž1983. 170–197. M.-M. Mesulam, E.J. Mufson, A.I. Levey, B.H. Wainer, Atlas of cholinergic neurons in the forebrain and upper brainstem of the macaque based on monoclonal choline acetyltransferase immunohistochemistry and acetylcholinesterase histochemistry, Neuroscience 3 Ž1984. 669–686. M. Meunier, J. Bachevalier, M. Mishkin, E.A. Murray, Effects on visual recognition of combined and separate ablations of the entorhinal and perirhinal cortex in rhesus monkeys, J. Neurosci. 13 Ž1993. 5418–5432. B. Milner, Visual recognition and recall after right temporal-lobe excision in man, Neuropsychologia 6 Ž1968. 191–209. M. Mishkin, Memory severely impaired by combined but not by

138

w39x

w40x

w41x

w42x w43x

w44x w45x

w46x

w47x

w48x w49x

w50x

w51x

w52x

w53x

w54x

w55x

w56x

w57x

R.M. Ridley et al.r Brain Research 836 (1999) 120–138 separate removal of amygdala and hippocampus, Nature 273 Ž1978. 297–298. R.G. Morris, P. Garrud, J.N. Rawlins, J. O’Keefe, Place navigation impaired in rats with hippocampal lesions, Nature 297 Ž1982. 681– 683. J.L. Muir, B.J. Everitt, T.W. Robbins, AMPA-induced excitotoxic lesions of the basal forebrain: a significant role for the cortical cholinergic system in attentional function, J. Neurosci. 14 Ž1994. 2313–2326. O.G. Nilsson, G. Leanza, C. Rosenblad, D.A. Lappi, R.G. Wiley, A. Bjorklund, Spatial learning impairments in rats with selective im¨ munolesion of the forebrain cholinergic system, NeuroReport 3 Ž1992. 1005–1008. J. O’Keefe, L. Nadel, The Hippocampus as a Cognitive Map, Oxford Univ. Press, Oxford, 1978. J.K. Parkinson, E.A. Murray, M. Mishkin, A selective mnemonic role for the hippocampus in monkeys: memory for the location of objects, J. Neurosci. 8 Ž1988. 4159–4167. J. Perner, Understanding the Representational Mind, MIT Press, Cambridge, MA, 1993. G.A. Press, D.G. Amaral, L.R. Squire, Hippocampal abnormalities in amnesic patients revealed by high-resolution magnetic resonance imaging, Nature 341 Ž1989. 54–57. N.L. Rempel-Clower, S.M. Zola, L.R. Squire, D.G. Amaral, Three cases of enduring memory impairment after bilateral damage limited to the hippocampal formation, J. Neurosci. 16 Ž1996. 5233–5255. R.M. Ridley, D.M. Aitken, H.F. Baker, Learning about rules but not about reward is impaired following lesions of the cholinergic projection to the hippocampus, Brain Res. 502 Ž1989. 306–318. R.M. Ridley, H.F. Baker, A critical evaluation of monkey models of amnesia and dementia, Brain Res. Rev. 16 Ž1991. 15–37. R.M. Ridley, H.F. Baker, B. Drewett, J.A. Johnson, Effects of ibotenic acid lesions of the basal forebrain on serial reversal learning in marmosets, Psychopharmacology 86 Ž1985. 438–443. R.M. Ridley, J.A. Baker, H.F. Baker, C.J. Maclean, Restoration of cognitive abilities by cholinergic grafts in cortex of monkeys with lesions of the basal nucleus of Meynert, Neuroscience 63 Ž1994. 653–666. R.M. Ridley, H. Barefoot, C.J. Maclean, P. Pugh, H.F. Baker, Different effects on learning ability following injection of the cholinergic immunotoxin ME20.4IgG-saporin into the diagonal band of Broca, basal nucleus of Meynert, or both in monkeys, Behav. Neurosci. 113 Ž1998. 303–315. R.M. Ridley, P.M. Bowes, H.F. Baker, T.J. Crow, An involvement of acetylcholine in object discrimination learning and memory in the marmoset, Neuropsychologia 22 Ž1984. 253–263. R.M. Ridley, B.A. Clark, L.J. Durnford, H.F. Baker, Stimulus-bound perseveration after frontal ablations in marmosets, Neuroscience 52 Ž1993. 595–604. R.M. Ridley, L.J. Durnford, J.A. Baker, H.F. Baker, Cognitive inflexibility after archicortical and paleocortical prefrontal lesions in marmosets, Brain Res. 628 Ž1993. 56–64. R.M. Ridley, S. Gribble, B. Clark, H.F. Baker, A. Fine, Restoration of learning ability in fornix-transected monkeys after fetal basal forebrain but not fetal hippocampal tissue transplantation, Neuroscience 48 Ž1992. 779–792. R.M. Ridley, T.K. Murray, J.A. Johnson, H.F. Baker, Learning impairment following lesion of the basal nucleus of Meynert in the marmoset: modification by cholinergic drugs, Brain Res. 376 Ž1986. 108–116. R.M. Ridley, C. Pearson, T.R. Kershaw, H. Hodges, C.J. Maclean, C. Hoyle, H.F. Baker, Learning impairment induced by lesion of the CA1 field of the primate hippocampus: attempts to ameliorate the impairment by transplantation of fetal CA1 tissue, Exp. Brain Res. 115 Ž1996. 83–94.

w58x R.M. Ridley, N.A. Samson, H.F. Baker, J.A. Johnson, Visuospatial learning impairment following lesion of the cholinergic projection to the hippocampus, Brain Res. 456 Ž1988. 71–87. w59x R.M. Ridley, H.D. Thornley, H.F. Baker, A. Fine, Cholinergic neural transplants into hippocampus restore learning ability in monkeys with fornix transections, Exp. Brain Res. 83 Ž1991. 533–538. w60x R.M. Ridley, C.J. Timothy, C.J. Maclean, H.F. Baker, Conditional learning and memory impairments following neurotoxic lesion of the CA1 field of the hippocampus, Neuroscience 67 Ž1995. 263–275. w61x T.W. Robbins, B.J. Everitt, H.M. Marston, J. Wilkinson, G.H. Jones, K.L. Page, Comparative effects of ibotenic acid- and quisqualic acid-induced lesions of the substantia innominata on attentional function in the rat: further implications for the role of the cholinergic neurons of the nucleus basalis in cognitive processes, Behav. Brain Res. 35 Ž1989. 221–224. w62x A.C. Roberts, T.W. Robbins, B.J. Everitt, G.H. Jones, T.E. Sirkia, J. Wilkinson, K. Page, The effects of excitotoxic lesions of the basal forebrain on the acquisition, retention and serial reversal of visual discriminations in marmosets, Neuroscience 34 Ž1990. 311–329. w63x A.C. Roberts, T.W. Robbins, B.J. Everitt, J.L. Muir, A specific form of cognitive rigidity following excitotoxic lesions of the basal forebrain in marmosets, Neuroscience 47 Ž1992. 251–264. w64x W.E.B. Scoville, B. Milner, Loss of recent memory after bilateral hippocampal lesions, J. Neurol. Neurosurg. Psychiatry 20 Ž1957. 11–21. w65x L.R. Squire, S. Zola Morgan, K.S. Chen, Human amnesia and animal model of amnesia: performance of amnesic patients on tests designed for the monkey, Behav. Neurosci. 102 Ž1988. 210–221. w66x H. Stephan, G. Baron, W.K. Schwerdtfeger, The Brain of the Common Marmoset Ž Callithrix jacchus .; a Stereotaxic Atlas, Springer, Berlin, 1980. w67x E.M. Torres, T.A. Perry, A. Blokland, L.S. Wilkinson, R.G. Wiley, D.A. Lappi, S.B. Dunnett, Behavioural, histochemical and biochemical consequences of selective immunolesions in discrete regions of the basal forebrain cholinergic system, Neuroscience 63 Ž1994. 59–122. w68x E. Tulving, H.J. Markowitsch, Episodic and declarative memory: role of the hippocampus, Hippocampus 8 Ž1998. 198–204. w69x F. Vargha-Khadem, D.G. Gadian, K.E. Watkins, A. Connelly, W. Van Paesschen, M. Mishkin, Differential effects of early hippocampal pathology on episodic and semantic memory, Science 277 Ž1997. 376–380. w70x N. Vnek, L.F. Kromer, R.G. Wiley, L.A. Rothblat, The basal forebrain cholinergic system and object memory in the rat, Brain Res. 710 Ž1996. 265–270. w71x M.L. Voytko, D.S. Olton, R.T. Richardson, I.K. Gorman, J.T. Tobin, D.L. Price, Basal forebrain lesions in monkeys disrupt attention but not learning and memory, J. Neurosci. 14 Ž1994. 167–186. w72x J.J. Waite, A.D. Chen, M.L. Wardlow, R.G. Wiley, D.A. Lappi, L.J. Thal, 192 Immunoglobulin G-saporin produces graded behavioral and biochemical changes accompanying the loss of cholinergic neurons of the basal forebrain and cerebellar Purkinje cells, Neuroscience 65 Ž1995. 463–476. w73x G.L. Wenk, J.D. Stoehr, G. Quintana, S. Mobley, R.G. Wiley, Behavioral, biochemical, histological, and electrophysiological effects of 192 IgG-saporin injections into the basal forebrain of rats, J. Neurosci. 14 Ž1994. 5986–5995. w74x S. Zola-Morgan, L.R. Squire, Memory impairments in monkeys following lesions limited to the hippocampus, Behav. Neurosci. 100 Ž1986. 155–160. w75x S. Zola-Morgan, L.R. Squire, S.J. Ramus, Severity of memory impairment in monkeys as a function of locus and extent of damage within the medial temporal lobe memory system, Hippocampus 4 Ž1994. 483–495.