Interaction of frontal and perirhinal cortices in visual...

European Journal of Neuroscience, Vol. 10, pp. 3044–3057, 1998 © European Neuroscience Association

Interaction of frontal and perirhinal cortices in visualobject recognition memory in monkeys

Amanda Parker and David GaffanDepartment of Experimental Psychology, Oxford University, South Parks Road, Oxford OX1 3UD, UK

Keywords: amygdala, fornix, frontal cortex, MD thalamus, perirhinal cortex, primates, visual object recognition memory

Abstract

Monkeys were trained preoperatively in visual object recognition memory. The task was delayed matching-to-sample with lists of trial-unique randomly generated visual stimuli in an automated apparatus, and the stimuliwere 2D visual objects made from randomly generated coloured shapes. We then examined the effect of either:(i) disconnecting the frontal cortex in one hemisphere from the perirhinal cortex in the contralateral hemisphereby crossed unilateral ablations; (ii) disconnecting the magnocellular portion of the mediodorsal (MDmc) thalamicnucleus in one hemisphere from the perirhinal cortex in the contralateral hemisphere; or (iii) bilaterally ablatingfirst the amygdala, then adding fornix transection, then finally perirhinal cortex ablation. We found that bothfrontal/perirhinal and MDmc/perirhinal disconnection had a large effect on visual object recognition memory,whereas both amygdalectomy and the addition of fornix transection had only a mild effect. We conclude that thefrontal lobe needs to interact with the perirhinal cortex within the same hemisphere for visual object recognitionmemory, but that routes through the amygdala and hippocampus are not of primary importance.

Introduction

The structures necessary for visual object recognition memory per-formance in primates, and the routes by which they interact, are stillthe subject of vigorous debate. Recent research attention has beenfocused on the rhinal cortex (particularly the perirhinal cortex) andits role in visual object memory (Meunieret al., 1993; Eacottet al.,1994; Buckley & Gaffan, 1997a,b). Much human and non-humanprimate evidence also points to the importance of the frontal cortexboth in recognition memory (Bachevalier & Mishkin, 1986; Buckner& Petersen, 1996; Meunieret al., 1997) and visual association learning(Gaffan & Murray, 1990).

One important question about the role of perirhinal and frontalcortices in visual object recognition memory concerns the routes bywhich these two cortical areas interact. One possible route whichdoes not appear to be vital for recognition memory performance isthe cortico-cortical connections between frontal and temporal cortices,as bilateral transection of this pathway does not have an effect ondelayed matching-to-sample (Gaffan & Eacott, 1995). A furtherpossible route of interaction is via limbic structures. It has beenproposed that the hippocampus and amygdala relay visual objectinformation via the anterior and medial thalamus to the frontal cortex(Aggleton & Mishkin, 1983a; Bachevalieret al., 1985). A morerecent view is that neither the amygdala (Murrayet al., 1996) northe hippocampus (Murray & Mishkin, 1996) are involved in thisform of memory. If this alternative view is accepted, other possibleroutes of frontal/perirhinal interaction must be sought. To explorethese competing positions, the present experiments examined theperformance of three groups of monkeys. In the first experiment, thefirst group had crossed unilateral lesions of perirhinal cortex and

Correspondence: A. Parker, as above. E-mail: [email protected]

Received 5 January 1998, revised 22 April 1998, accepted 24 April 1998

frontal lobe, and the second group had crossed unilateral lesions ofperirhinal cortex and magnocellular portion of the mediodorsalthalamus, in both cases combined with partial forebrain commissuro-tomy (involving sectioning the body and splenium of the corpuscallosum and anterior commissure) to prevent transfer of visualinformation between the hemispheres (Noble, 1973). In the secondexperiment, a third group of monkeys had bilateral removal of theamygdala, then fornix transection, then finally had bilateral perirhinalcortex ablation. All animals were retested after each surgery.

One issue to be investigated here was whether it is necessary invisual object recognition for the frontal cortex to interact withthe perirhinal cortex intrahemispherically, i.e. exchange detailedinformation, or if alternatively the frontal cortex is involved in amore general, diffuse form of control or executive processing.The wide possible range of outcomes for the frontal/perirhinaldisconnection group is suggested by recent experiments which haveshown that whilst conditional learning and strategy–object learning arecritically dependent on intrahemispheric frontal/temporal interaction,simple visual object reward association learning is unaffected bydisconnection of the inferior temporal cortex from the frontal cortex,even though bilateral lesions of either the temporal cortex or frontalcortex produce a severe impairment (Parker & Gaffan, 1997, 1998).

The question being asked of the second experimental group,who had disconnection of the entire magnocellular portion of themediodorsal thalamus (MDmc) from the perirhinal cortex, waswhether this would produce the same degree of impairment as frontal/perirhinal disconnection. Earlier studies have concentrated on onlythe medial portion of MDmc, because of its position as a relay

Recognition memory and frontal/perirhinal interaction 3045

FIG. 1. (A) Lateral (top), medial (centre) and ventral (bottom) views of the intended frontal ablation. (B) Ventral view of the intended perirhinal ablation. (C) Aseries of sections from animal FL/P1 to illustrate the unilateral frontal and perirhinal cortex ablations. The sections run from anterior at the top to posterior atthe bottom. Sections from the other animals were similar to these. See text for description of ablations.

© 1998 European Neuroscience Association,European Journal of Neuroscience, 10, 3044–3057

3046 A. Parker and D. Gaffan

FIG. 2. Reconstructions of the lesions in animals FL/P1 (top row), FL/P2 (middle row) and FL/P3 (bottom row). Lesions have been mapped onto standarddrawings for ease of comparison, and some are consequently reversed. See text for description of ablations.

between limbic regions, particularly the amygdala, and orbital frontalcortex (Aggleton & Mishkin, 1983b; Zola-Morgan & Squire, 1985;Parkeret al., 1997). The possibility considered here is that the entiremagnocellular division of the mediodorsal thalamus (MDmc) plays acrucial role in recognition memory performance, possibly because ofits function in the interaction of afferents and efferents of differentfrontal cortical regions, or because of the importance of other afferents,e.g. from the basal forebrain (Russchenet al., 1987), which projectto a wider area of the magnocellular division. This hypothesis canbe tested by comparison of the effects of larger mediodorsal thalamiclesions with those which have been limited to the projection targetsof limbic structures. A larger effect of the present disconnectionwould suggest that MDmc has a different role in visual objectrecognition than that of relay between amygdala and frontal cortex.Within the present group, a control was made for fornix participationby transecting the fornix unilaterally; in two animals this was donein the contralateral hemisphere to the thalamic lesion, in the thirdipsilaterally. According to the limbic view, an ipsilateral transection


should produce a larger impairment. According to the other routeposition there should be no difference. Figure 7 gives a schematicview of the anatomical structures and their interactions.

Experiment 1

Materials and methods

Subjects

These were six male rhesus macaques (Macaca mulatta). At the timeof their first surgery they weighed on average 5.5 kg. Before takingpart in this experiment, four were naive animals and two hadpreviously served as unoperated control animals in studies whichused the same experimental equipment as the present study, but didnot involve recognition memory performance. One of the moreexperienced monkeys was placed in each of the experimental groups.

The first group (FL/P), was three rhesus. Each animal was operated


FIG. 3. Columns 1–3 illustrate the extent of the mediodorsal thalamic lesion in animals MD/P2, MD/P1 and MD/P3, from left to right. In the rightmost columnare a series of sections from MD/P2. See Abbreviations section for definitions of labels.

on twice, once to remove the perirhinal cortex in one hemisphere,and once to ablate the cortex of the contralateral frontal lobe and totransect the forebrain commissures. Two of the three animals hadfrontal ablation and commissurotomy in their first surgery (animalFL/P1 in the right hemisphere and FL/P3 in the left). Animal FL/P2had right perirhinal ablation as the first surgery.


The second group (MD/P) was three rhesus. Each animal wasoperated on twice, once to remove the perirhinal cortex in onehemisphere, and once, in the contralateral hemisphere, to ablate themedial part of the mediodorsal nucleus of the thalamus unilaterally,and to partially section the forebrain commissures. Animals MD/P1and MD/P2 had mediodorsal thalamic ablation and commissurotomy


in their first surgery (MD/P1 in the right hemisphere and MD/P2 inthe left hemisphere). Animal MD/P3 had a left perirhinal ablation asthe first surgery.

Surgery

On the completion of preoperative training or adaptation from theirprevious training experience, the monkeys had a first operation, asdetailed above and described below, then a period of rest, thenresumed daily testing. After a second surgery, then a further periodof rest, they were retested. All operations were performed in sterileconditions with barbiturate anaesthesia and with the aid of an operatingmicroscope. When the lesion was complete, overlying tissue wassutured in anatomical layers. Following each surgery, before theresumption of training, 10–14 days were allowed for postoperativerecovery.

Perirhinal cortex ablation

The arch of the zygoma was removed and the temporal muscledetached from the cranium and retracted. For the lesion in eachhemisphere, a bone flap was raised over the frontal and temporallobe. The medial and posterior limits of the flap were in a crescentshape extending from within 5 mm of the midline at the brow to theposterior insertion of the zygomatic arch. The anterior limit of theflap was the brow and the orbit. Ventrally the flap extended from theposterior insertion of the zygomatic arch to the level of the superiortemporal sulcus in the lateral wall of the temporal fossa anteriorly.The ventral anterior part of this bone opening was then extendedwith a rongeur ventrally into the wall of the temporal fossa to reachthe base of the temporal fossa. The dura mater was cut to expose thedorsolateral frontal and lateral temporal lobes. The most anterior partof the rhinal sulcus was visualized by retracting the frontal lobe fromthe orbit with a brain spoon. The dorsal limit of the removal on theanterior face of the temporal pole wasµ 2 mm ventral to the lateralsulcus. A line of pia mater was cauterized, and the underlying cortexwas removed by aspiration in the lateral bank of the anterior part ofthe rhinal sulcus and in the adjacent 2 mm of cortex on the thirdtemporal convolution. The monkey’s head was then tilted to an angleof 120° from vertical, and the base of the temporal lobe was retractedfrom the floor of the temporal fossa with a brain spoon. The posteriortip of the first part of the ablation was identified visually and thenextended in the lateral bank of the rhinal sulcus to the posterior tipof the sulcus, again removing 2 mm of laterally adjacent tissue. Thedura mater was sewn and the wound closed in layers. The intendedlesion is shown in Fig. 1B.

Frontal cortical ablation and forebrain commissurotomy

A D-shaped bone flap was cut over the left hemisphere and removed.The dura was cut along the curved side of ‘D’ and reflected to exposepart of the hemisphere and the midline. The hemispheres wereseparated and the veins obscuring access to the corpus callosum werecauterized. The hemisphere was retracted from the falx to enableaccess to the interhemispheric fissure. A glass aspirator was used tosection the corpus callosum in the midline from the posterior limitof the splenium to the genu. At the level of the interventricularforamen, anterior to the thalamus, one descending column of the

FIG. 4. On the left, four sections from animals A/FX1 (bottom four sections) and from A/FX2 (top four sections) to illustrate the method of amygdalectomy,fornix transection and perirhinal cortex ablation in this group. The sections run from anterior at the top left to posterior at the bottom right. Enlarged views ofthe sections are in the right column.


fornix was retracted in order to gain access to the foramen. Theanterior commissure was visualized and sectioned in the midline byelectrocautery and aspiration with a 23-gauge metal aspirator whichwas insulated to the tip. The frontal cortex was removed in threeparts, and in each case the cortical tissue and pia mater bounded byeach continuous line of cautery were then removed by aspiration.The first area of cortex to be removed was the lateral surface. Thecontinuous line of cautery extended firstly through the centre of theprecentral dimple from the most dorsal point on the lateral surfaceof the frontal lobe to the ventral surface, following a line approximatelyparallel to the central sulcus and ventrally 1–2 mm posterior to thedescending limb of the arcuate sulcus. The line then extended aroundthe boundary of the visible lateral surface of the frontal lobe, alongthe margin of the orbital surface and the lip of the interhemisphericfissure, until it re-joined the point at which it started. The second lineof cautery began from the ventral-most point of the previous line,and extended along the lip of the lateral sulcus on the orbital surfaceof the frontal lobe to the base of the olfactory tract and beyond it, tothe boundary of the medial surface. It then extended along the orbitalsurface anteriorly to the frontal pole, joining the line of cautery fromthe first area of cortex which was removed. The final line of cauteryextended from the origin of the first line of cautery ventrally to thecorpus callosum, then followed the line of the corpus callosumanteriorly and ventrally, then followed the boundary of the corticalsurface until meeting the line of cautery made on the orbital surface.In all cases, the grey matter within the sulci was removed, and thewhite matter surrounding the striatum, and the striatum itself was leftintact. The intended lesion is shown in Fig. 1A.

Mediodorsal thalamic ablation, body and splenium of the corpuscallosum transection and anterior commissure section

A D-shaped bone flap was cut over the left hemisphere and removed.The dura was cut along the curved side of ‘D’ and reflected to exposepart of the hemisphere and the midline. The hemispheres wereseparated and the veins obscuring access to the corpus callosum werecauterized. The hemisphere was retracted from the falx to enableaccess to the interhemispheric fissure. A glass aspirator was used tosection the corpus callosum near the midline from the posterior limitof the splenium to the level of the interventricular foramen. Thedescending column of the fornix was gently retracted with a narrowbrain spoon to enter the third ventricle. The anterior commissure wasvisualized and sectioned in the midline by electrocautery and aspirationwith a 23-gauge metal aspirator which was insulated to the tip. Thesplenium section (described above) was carried out 1–2 mm lateralfrom the midline, in order to open the lateral ventricle between thefimbria-fornix and the splenium. When the posterior limit of thefimbria and, anteriorly, the lateral limit of the fornix had both beenexposed, the fimbria-fornix was cut unilaterally in a line betweenthese two limits. The membrane covering the thalamus was cut inthe midline by applying cautery in order to expose the posteriorcommissure, the third ventricle and the posterior 5 mm of the midlinethalamus. Next, the posterior 5 mm of the massa intermedia of thethalamus was cut in the midline with a glass aspirator. Approximately3.2 mm of tissue adjacent to the cut massa intermedia was removedunilaterally by aspiration with a metal aspirator.


TABLE 1. Delayed matching-to-sample performance in experiment 1

Delays Lists Percent correct at list position

0 5 2 3 5 1 2 3 mean

Experiment 1

Group FL/P monkey T (E) T (E) T (E) T (E) T (E)

Pre-op FL/P1 777 (217) 548 (48) 709 (109) 1594 (214) 769 (69) 95.7 93.6 90.0 93.1FL/P2 346 (46) 673 (73) 3877 (677) 4143 (643) 5823 (1123) 90.9 88.6 82.3 87.3FL/P3 760 396 1069 (269) 3270 (545) 2098 (348) 2261 (561) 96.8 87.6 73.6 86.0mean 628 (220) 763 (130) 2619 (444) 2612 (403) 2951 (584) 94.5 89.9 82.0 88.8

Post-op1 FL /P1 153 (28) 106 (6) 107 (7) 448 (48) 95.2 88.7 88.7 90.7FL/P2 116 (16) 610 (110) 2573 (413) 1782 (282) 91.8 80.4 80.6 84.2FL /P3 184 (34) 218 (18) 437 (37) 1194 (194) 94.8 79.1 77.8 83.9mean 151 (36) 245 (45) 1039 (152) 1141 (175) 93.9 82.7 82.4 86.3

Post-discon FL /P1 222 (22) 2245 (345) 1725 (525)F 75.6 53.7 64.7FL /P2 516 (176) 1049 (249) 1786 (586)F 73.1 58.5 65.8FL /P3 731 (131) 1339 (239) 1550 (450)F 79.1 59.9 69.5mean 297 (110) 1544 (278) 1687 (520) 75.9 57.4 66.7

Group MD/P

Pre-op MD/P1 792 (192) 331 (31) 1072 (212) 2261 (341) 3854 (674) 93.7 88.6 86.1 89.5MD/P2 720 (120) 683 (83) 2892 (672) 4586 (786) 3554 (554) 95.8 87.9 84.5 89.4MD/P3 432 (157) 740 (140) 170 (20) 698 (98) 2250 (355) 96.9 91.6 82.8 90.4mean pre-op 648 (156) 585 (85) 1378 (301) 2515 (408) 3219 (528) 95.5 89.4 84.5 89.8

Post-op1 MD /P1 107 (7) 215 (15) 1241 (181) 1818 (318) 95.7 80.8 80.4 85.6MD /P2 104 (4) 306 (6) 105 (5) 1548 (248) 97.7 77.6 74.5 83.3MD/P3 103 (3) 209 (9) 107 (7) 1379 (179) 96.2 82.1 80.2 86.2mean post-op1 105 (5) 243 (10) 484 (64) 1582 (248) 96.5 80.2 78.4 85.0

Post-discon MD /P1 101 (1) 334 (34) 1532 (372)F 87.9 62.5 75.2MD /P2 107 (7) 209 (9) 1318 (318)F 88.0 64.2 76.1MD /P3 1073 (173) 1173 (73) 1677 (502)F 79.3 62.0 70.7mean post-disc 427 (60) 572 (39) 1509 (397) 85.1 62.9 74.0

Scores are trials T and errors (E) to criterion in all stages of the experiment. Per cent correct at list position is performance measured for the first three testpositions at the last level of criterion attained in each stage of the experiment.F 5 failed to reach criterion. Bold font in animal identification indicates surgerycompleted at that stage.

Histology

At the conclusion of the behavioural experiment the animals weredeeply anaesthetized, then perfused through the heart with salinefollowed by formol–saline solution. The brains were blocked in thecoronal stereotaxic plane posterior to the lunate sulcus, removed fromthe skull, and allowed to sink in sucrose–formalin solution. Thebrains were cut into 50-mm sections on a freezing microtome. Everyfifth section was retained and stained with cresyl violet.

Figure 2 shows the extent of the cortical ablations in each of theanimals. Figures 1c and 4 show representative sections illustratingthe perirhinal lesion. In the majority of cases the ablation was asintended, in that in all cases the ablation included the entire anterior–posterior extent of the lateral bank of the rhinal sulcus. The cortexin the medial bank of the rhinal sulcus, and the cortex medial to thesulcus, were substantially intact. The lateral limit of the ablation wasin every case on the inferior temporal gyrus, lateral to the rhinalsulcus, leaving the cortex in the anterior middle temporal sulcusintact. The posterior third of the ablation extended slightly morelaterally in case FL/P1 than intended, but by less than 1 mm. In allcases, degeneration was visible in the white matter adjacent to thecortical removal. Although coronal sections are most appropriate forvisualizing the main part of the ablation, as shown in Fig. 1, the mostanterior part of the ablation is difficult to interpret in coronal sectionsthrough the temporal pole, however, inspection of the temporal polebefore the brains were sectioned confirmed that the ablation extended


anteriorly and dorsally on the anterior face of the temporal pole towithin 2 mm of the lateral sulcus, in the lateral bank of the mostanterior part of the rhinal sulcus and in the laterally adjacent cortex,sparing the tissue medial to the rhinal sulcus.

Figure 1C shows representative sections revealing the unilateralfrontal cortical ablation in animal FL/P1. In two cases there was acomplete removal of the frontal cortex as intended, sparing theprimary motor area. In case FL/P1 there was slight sparing of theposterior part (µ 5 mm anterior to posterior and 1.5 mm ventral todorsal) of the ventromedial cortex. As can be seen in Fig. 1c, thestriatum and internal capsule are partially degenerated. In all threeof these animals, this striatal damage was due to degeneration, ratherthan unintended damage during surgery. In all three animals, theforebrain commissures were completely transected, as intended.

Figure 3A shows the extent of the mediodorsal thalamic lesion ineach of the animals in group MD/P, and shows an illustrative seriesof sections from animal MD/P2. The sections from the brains werematched to a separately prepared set of sections of brains with anormal thalamus, in order to identify by cytoarchitecture the limitsof the lesions. The method used in the present experiment is thatwhich was fully described by Gaffan & Murray (1990), whereillustrations of the normal material on which the analysis was basedcan be found. In all three cases there was substantial unilateralremoval of MDmc, with minor sparing of MDmc and some damageto other structures. Examination of the histological material indicated


FIG. 5. Mean errors made in list positions 1 and 2 combined by groups FL/Pand MD/P in each stage of the experiment. PRE, final preoperativeperformance. POST 1, after first unilateral ablation. POST 2, after second,contralateral ablation which completed disconnection procedure. Symbolsrepresent individual animals.

that in all three cases, the initial removal of the posterior 5 mm ofthe massa intermedia of the thalamus in the midline had been largelyas intended, excepting in subject MD/P1, in whom slight unilateraldamage to the anterior medial nucleus was sustained. In all threecases some damage to the reuniens nucleus and in one case (MD/P2)central medial nucleus and parafascicular nucleus damage was seenin the midline. The second phase of the ablation, where tissue adjacentto the cut massa intermedia was removed unilaterally, was also largelysuccessful in all three animals. Of the three animals, two showedvery little sparing of MDmc, whereas the third (MD/P1) showedgreater ventral sparing. There was also some unilateral damage tothe parvocellular region of MD in all three cases, but this was verysmall in extent in two animals (MD/P1 and MD/P2). In MD/P3, thedamage extended into the paracentral nucleus and the posterior partof the ventral lateral nucleus. The three animals with mediodorsalthalamic lesions all had unilateral fornix damage, and some gliosiswas observed in the fornix and fimbria. In MD/P1 and MD/P2 thefornix was deliberately transected ipsilateral to the MD lesion. InMD/P3 there was unintentional damage to the fornix contralateral tothe MD lesion, sustained during the anterior commissure transection.In all three animals, the anterior commissure was transected, and thecorpus callosum was transected from the posterior limit of thesplenium to the level of the interventricular foramen, as intended.

Apparatus

The apparatus consisted of a computer-controlled touch-sensitivescreen on which coloured stimuli were displayed. The monkey sat in


FIG. 6. Percent correct by list position in delayed matching-to-sampleperformance before surgery and after either disconnection (i.e. either frontallobectomy plus contralateral perirhinal cortex ablation, or mediodorsal thalamicablation plus contralateral perirhinal cortex ablation) or bilateralamygdalectomy plus fornix transection. Symbols represent: open circles,dotted lines, preoperative mean for all eight monkeys. Filled diamond,amygdala plus fornix transection (A/FX). Filled triangle, mediodorsal thalamic/perirhinal disconnection (MD/P). Filled square, frontal/perirhinal disconnection(FL/P).

a transport cage in front of the screen and made choices betweenstimuli by reaching out from the cage to touch them. Touches wereregistered by the computer and 190 mg food pellets were automaticallydelivered as rewards for correct responses. The stimuli consisted oftwo coloured typographic characters, abutted one next to the other.Each stimulus wasµ 4 3 4 cm. The two typographic characters andeach of their colours were chosen at random from a large set ofavailable characters and colours, and therefore it was possible togenerate a very large number of different stimuli (µ 500 million).There were three places on the screen at which stimuli could appear:a central position and a position to the left and right of the centralposition, the centre to centre separation between the left and rightpositions beingµ 15 cm. The screen was 380 mm wide and 280 mmhigh. A closed-circuit television system allowed the experimenter inanother room to watch the monkey. The small food rewards (pellets,190 mg) were delivered into a hopper placed centrally underneaththe monitor screen, and a single large food reward was delivered atthe end of each training session by opening a box which was set toone side of the centrally placed hopper. The box contained peanuts,raisins, proprietary monkey food, fruit and seeds. The amount of thislarge reward was adjusted for individual animals in order to avoidobesity. The small and large rewards dispensed in the training


apparatus provided the whole daily diet of the monkeys on days witha training session. Opening of the box with the large food reward,like all other aspects of the stimulus display and the experimentalcontingencies during any session of training, was under computercontrol.

Training procedure

The animals were trained in matching-to-sample before and aftersurgery. In initial training, on each trial a sample appeared in thecentre of the screen and the animal touched it to proceed. Touchingthe sample caused the sample to disappear 1 s later, and thenimmediately produced two choice stimuli in the left and right positionson the screen, one of which was identical to the sample and the othernot, and the animal was rewarded with a food pellet for touching thechoice stimulus which matched the sample. Both stimuli disappearedfrom the screen immediately that one of them was touched. Theintertrial interval began immediately the stimulus or stimuli dis-appeared from the screen. For the final correct choice in the session,the final reward pellet was followed by a large food reward (seeApparatus) and the animal was given at least 10 min to eat some ofthis food and take the remainder into the cheek pouches, before beingreturned to the home cage. Subsequent preoperative training proceededthrough increasing delays of 1, 2 then 5 s between the sampledisappearing from the screen and the choice stimuli appearing.Subjects remained on each of these stages until a criterion of 90%correct choices were made in a session which continued until 100rewards were earned. On attaining criterion on delays with a listlength of one, the list length was then increased to 2, 3 then 5.Presentation of lists of choice stimuli was in the reverse order to thepresentation of the sample stimuli. For lists, the time intervals wereas follows: after a sample presentation there was an interval of 5 s,and after a choice trial there was an interval of 10 s. For the listlength of 2, the criterion was set at 90% in a session of 100 rewards,as before. For the list lengths of 3 and 5, the criterion was twosuccessive days at 90% in sessions of 100 rewards. On attaining thiscriterion preoperatively, the animal then underwent surgery, followedby a period of rest. Animals resumed training on lists of 1. Postopera-tive criterions were attained in the same progression of increasingdelays and list lengths as preoperatively, with failure determined bynot attaining the 90% criterion after 10 daily sessions of training,with each session requiring the animal to earn 100 rewards. Themaximum list length tested postoperatively was 3. Performance wasmeasured as percentage correct at each of the first three list positions,or two positions if a list length of two was all that was attained.

Results

Table 1 and Figs 5 and 6 show the pre- and postoperative results forgroups FL/P and MD/P. We compared the memory performance foreach of these two groups, in all the phases of the experiment, usingthe measure of percentage correct in each of the first three listpositions (if a criterion of 3 or above was reached), or two listpositions if this was the final stage the animals reached. In eachphase this was for the last 10 (or less if the criterion was reached)sessions each monkey attained final criterion or failed in (see above,procedure). In all of these sessions, each animal earned 100 rewardsper session. A three-wayANOVA for the three phases of the experiment(preoperative, post-operation 1 (i.e. after first unilateral lesion) andpost-disconnection (i.e. after second, contralateral, lesion), see Fig. 7)by list position 1 and 2 by group showed a very significant effect ofphase (F2,85 134.83, P , 0.0001), a significant effect of group


FIG. 7. (A) Illustrates the hypothesis that in normal performance of delayedmatching-to-sample the frontal cortex and perirhinal cortex interact viamultisynaptic cortico-cortical connections, involving other cortical areas. (B)Shows that this interaction is blocked in a hemisphere with perirhinal cortexablation (indicated by dark grey shading), and (C) shows the same thing forfrontal cortex ablation. (D) Illustrates our hypothesis that ablation of themagnocellular part of the mediodorsal thalamic nucleus induces a consequentfunctional disorder in the ipsilateral frontal lobe (indicated by pale greyshading), thus also blocking perirhinal–frontal interaction. Interhemisphericcortico-cortical connections, not shown in the figure, were cut by forebraincommissurotomy in our experiment.


(F1,45 7.23,P , 0.05), and a very significant effect of list position(F1,45 140.50,P , 0.0001). There was a very significant interactionbetween phase and list position (F2,85 35.63,P , 0.0001), indicatingthat list position 2 was more affected by phase than list position 1was, and the interaction between group and phase approachedsignificance (F2,85 4.42,P 5 0.051). A planned comparison of phase(using the pooled error term, see Snedecor & Cochran, 1967, pp.268–271) between preoperative and post-operation 1 performanceshowed a significant decrement in performance (t(8) 5 2.77,P , 0.05), which was equivalent in both operated groups (t , 1). Afurther planned comparison between post-operation 1 and post-operation 2 performance showed a highly significant difference(t(8) 5 12.64,P , 0.001), with a significant interaction between theoperated groups (t(8) 5 2.61,P , 0.05). This interaction shows thatboth these disconnections have an effect on memory performance,but that this effect was more pronounced following frontal/perirhinaldisconnection than following MDmc thalamus/perirhinal disconnec-tion. This is illustrated in Fig. 5, where it can be seen that performanceof both groups is very similar in phases 1 and 2, but in the finalphase, after disconnection, the frontal/perirhinal disconnected groupis achieving a lower level of correct choices than the MDmc thalamus/perirhinal disconnected group are. Further analysis of the scores forfirst and second list positions revealed that, for the difference inperformance between preoperative and post-operation 1 performance,there was no effect of phase upon the memory of items in the firstlist position (t , 1). In the second list position, there was a significantdecrement in performance (t(8) 5 6.46, P , 0.001). There was aneffect of phase on both first and second list positions when thedifference in scores after first and second surgeries was compared(first position:t(8) 5 11.50,P , 0.001; second position:t(8) 5 16.89,P , 0.0001).

Discussion

A large deficit in visual object recognition memory was seen afterthe removal of the frontal cortex in one hemisphere and the perirhinalcortex in the contralateral hemisphere, or after the disconnection ofthe magnocellular division of the mediodorsal thalamus from theperirhinal cortex. The impairment was particularly marked in thememory of items in the last position in lists of 2, with performance inthe frontal/perirhinal disconnection group being particularly impaired.Although both groups were slightly impaired after their first unilaterallesion, they were far more impaired following disconnection. In thefinal stage of the experiment, the group which had the frontalcortex disconnected from the perirhinal cortex was significantly moreimpaired than the group which had a disconnection of the mediodorsalthalamus from the perirhinal cortex.

Experiment 2

This experiment further investigates the proposed limbic route forthe interaction of perirhinal and frontal cortex, through the amygdala,which projects via mediodorsal thalamus to the frontal lobe, and/orthe hippocampal system, which projects via the fornix and anteriorthalamus to the frontal lobe (Goldman-Rakic & Porrino, 1985). Alarge effect of combined transection of the fornix and amygdalofugalpathway on delayed matching-to-sample performance has been found(Bachevalieret al., 1985). Bachevalier and colleagues did not findany effect of separate transections, however, suggesting that theselimbic pathways may have some overlap in their function. The presentexperiment is a partial replication of their experiment, using the same


experimental procedure as in experiment 1, to allow close comparisonwith the two disconnection groups we presented there.

Materials and methods

Subjects

These were two male cynomolgus monkeys (Macaca fascicularis).At the time of their first surgery they weighed on average 5.3 kg.Before taking part in this experiment they had previously served asunoperated control animals in a study which used the same experi-mental equipment as the present study, but did not involve recognitionmemory performance. Each animal had three surgeries, in the sameorder, which were firstly amygdalectomy, secondly fornix transectionand finally perirhinal cortex ablation. All of these were bilateral. Themethod for perirhinal ablation was as before.

Surgery

On the completion of the adaptation from their previous trainingexperience and preoperative training, the monkeys had a bilateralamygdalectomy, then a period of rest, then resumed daily testing.After a second surgery to transect the fornix bilaterally, then a furtherperiod of rest and then retraining, they were given bilateral perirhinalcortex lesions in a third and final surgery. All operations wereperformed in sterile conditions with barbiturate anaesthesia and withthe aid of an operating microscope. When the lesion was complete,overlying tissue was sutured in anatomical layers. Following eachsurgery, before the resumption of training, 10–14 days were allowedfor postoperative recovery.

Amygdalectomy

The surgical method used for making amygdala lesions was identicalto that described by Gaffan & Harrison (1987). The medial temporallobe was exposed by retracting the frontal lobe over the orbit. Thepia mater was cauterized in an areaµ 5 mm in diameter on themedial surface of the temporal lobe medial and superior to the rhinalsulcus. The amygdala and periamygdaloid cortex medial to theamygdala were then ablated by aspiration through the defect in thepia mater. White matter of the temporal stem, and the lateral ventricleand anterior surface of the hippocampus, were visible boundaries ofthe ablation posteriorly, laterally and inferiorly. The dura was replacedover the cortex and sewn.

Bilateral fornix transection

The galea was cut and the left temporal muscle retracted. A D-shapedbone flap was cut over the left hemisphere and removed. The durawas cut along the curved side of ‘D’ and reflected to expose part ofthe hemisphere and the midline. The hemispheres were separated andthe veins obscuring access to the corpus callosum were cauterized.A midline slit µ 5–10 mm in length was made through the corpuscallosum using a glass sucker. The fornix was exposed just posteriorto its descent along the superior boundary of the third ventricleanterior to the massa intermedia of the thalamus. The fornix wastransected using a 20-gauge metal sucker with cautery. Using suction,the fornix was lifted slightly before being cauterized. The entirecircumference of the remaining trunks could then be visualized inorder to verify the completeness of the transection. The dura wasreplaced over the cortex but not sewn. The bone flap was replacedand secured with loose silk sutures.


TABLE 2. Delayed matching-to-sample performance in experiment 2

Delays Lists Percent correct at list position

0 5 2 3 5 1 2 3 mean

Experiment 2

monkey T (E) T (E) T (E) T (E) T (E)

Pre-op A/FX1 197 (47) 331 (31) 1220 (220) 1226 (166) 2348 (348) 98.7 94.4 93.3 95.5A/FX2 718 (168) 662 (62) 466 (66) 1997 (297) 1129 (129) 98.7 91.7 87.3 92.6mean 449 (108) 497 (47) 483 (143) 1622 (232) 1739 (239) 98.7 93.1 90.3 94.0

Post-op1 A/FX1 1881 (381) 2314 (414) 1301 (201) 1873 (373) 89.9 84.2 83.0 85.7A/FX2 1100 (150) 479 (79) 1559 (259) 1654 (254) 96.2 86.0 83.4 88.5mean 1491 (266) 1396 (247) 1430 (230) 1763 (314) 93.1 85.1 83.2 87.1

Post-op2 A/FX1 101 (1) 457 (57) 1412 (212) 2348 (348) 95.7 89.7 85.4 90.3A/FX2 401 (51) 1824 (324) 1151 (251) 2610 (610) 93.2 79.5 68.4 80.4mean 223 (26) 1140 (191) 1282 (232) 1348 (479) 94.5 84.6 76.9 85.4

Post-op3 A/FX/P1 2053 (583) 70.2A/FX/P2 1957 (457) 75.8mean 2005 (520) 73.0

As in Table 1.

Bilateral perirhinal cortex ablation

The method was the same as in experiment 1.

Histology

The brains were prepared as before. Figure 4 shows representativesections revealing the bilateral amygdalectomy, fornix transection andperirhinal cortex ablation in animals A/FX1 and 2. In both cases,there was a complete bilateral removal of the amygdala as intended.In both animals, white matter laterally adjacent to the amygdalaappeared intact. The bilateral fornix transection in both animals wasalso as intended. The perirhinal lesions in A/FX2 were as intended,in that in both hemispheres the ablation included the entire anterior–posterior extent of the lateral bank of the rhinal sulcus. Posterior tothe amygdalectomy the cortex medial to the perirhinal cortex wasintact. The lateral limit of the ablation was in every case on theinferior temporal gyrus, lateral to the rhinal sulcus, leaving the cortexin the anterior middle temporal sulcus intact. The posterior third ofthe ablation extended slightly more laterally in the case of A/FX1than intended, but by less than 1 mm. In all cases degeneration wasvisible in the white matter adjacent to the cortical removal. In bothanimals there was some slight postmortem damage to the inferiortemporal gyrus which was caused by adhesion to the dura mater.

Training procedure

As in experiment 1.

Results

As before, memory performance was calculated for each of these twogroups, in all the phases of the experiment, using the measure ofpercentage correct in each of the first 3 list positions (if a criterionof 3 or above was reached), or 1 or 2 list positions if this was thefinal stage the animals reached. In each phase this was for the last10 sessions each monkey attained final criterion or failed in (seeabove, procedure), in all of these sessions each animal earned 100rewards per session. Table 2 and Fig. 6 show the results for groupA/FX1 in each phase of the experiment. Both of the animals reattainedcriterion on lists of 5 stimuli after each of their first two surgeries,


but both appeared to show an increase in error rate after amygdal-ectomy. Post-fornix transection, one animal showed a further increasein error rate, and the other showed an improvement in performance.

A comparison of the post-operation 2 results of the animals fromthe present experiment with the two groups from experiment 1, usingone-wayANOVA of the mean performance of list positions 1 and 2,showed a significant difference between the groups (F2,55 34.53,P , 0.005). Planned comparisons (method as above) showed thatthe performance of the amygdala and fornix-lesioned group wassignificantly less affected than either group FL/P (t(5) 5 8.28,P , 0.001) or group MD/P (t(5) 5 5.58,P , 0.01).

After the addition of perirhinal cortex ablation, the animals in thepresent experiment were unable to reach criterion on lists of one. Acomparison of their post-operation 3 performance with the first listposition data of the two disconnected groups from the previousexperiment, using one-wayANOVA, showed a significant differencebetween the groups (F2,55 6.25, P , 0.05). Planned comparisonsrevealed that there was no significant difference between group FL/Pand group A/FX/P (t , 1), but there was a difference between groupMD/P and group A/FX/P (t(5) 5 3.23,P , 0.05).

Discussion

In contrast to both groups of animals in experiment 1, the monkeyswho received bilateral amygdalectomy showed only a mild impairmentin delayed matching-to-sample. Furthermore, their level of perform-ance was not further decremented by the addition of fornix transection.The earlier finding that combined transection of the fornix andamygdalofugal pathway produces a more severe effect on visualobject recognition performance than either of these lesions alone(Bachevalieret al., 1985) was therefore only partially supported bythe present data, as we found that following amygdalectomy alonethe monkeys were only mildly impaired, and the addition of fornixtransection led to an improvement in one animal and a decrement inthe other. The dramatic overall impairment in performance seen inBachevalieret al.’s amygdalofugal and fornix transected animals didnot have a counterpart here. However, direct comparison of thepresent results with the results of Bachevalieret al. (1985) is difficult,


due to differences in procedure, e.g. the monkeys in the presentexperiment received many more trials postoperatively, and extensivepreoperative training on delayed matching-to-sample with lists. Thefurther addition of perirhinal cortex ablation in the two monkeys withamygdalectomy and fornix transection led to them failing to reachcriterion with single items lists in delayed matching-to-sample per-formance, equivalent to the deficit seen at position 1 after frontal/perirhinal disconnection.

General discussion

The results of the two experiments presented here indicate that alarge deficit in visual object recognition memory results from theremoval of the frontal cortex in one hemisphere and the perirhinalcortex in the contralateral hemisphere, or from the disconnection ofthe magnocellular division of the mediodorsal thalamus from theperirhinal cortex. Although both of the disconnection groups wereimpaired, the group who had the frontal cortex disconnected fromthe perirhinal cortex were more impaired than the group who had adisconnection of mediodorsal thalamus from perirhinal cortex. Incontrast to the severe effects of disconnection, the animals fromexperiment 2 who received bilateral amygdalectomy and fornixtransection were not seriously impaired in DMS performance. Thecombined results from these two experiments show that visualrecognition memory performance is dependent on information flowbetween the perirhinal cortex and frontal cortex, and that thisinformation flow is not dependent on the amygdala and fornix.

Because we obtained a large impairment by disconnecting theperirhinal cortex from the mediodorsal thalamus, at first sight ourresults may seem to be consistent with the proposal by Aggleton &Mishkin (1983a,b) and Bachevalieret al. (1985) that visual recognitionmemory depends on information flow from the temporal cortexto the frontal cortex via the thalamus. However, several furtherconsiderations argue against this interpretation. First, the route ofinformation flow from the temporal cortex to the thalamus, asenvisaged by these authors, was via the fornix and amygdala, andwe have directly shown that with preoperative training in delayedmatching-to-sample, aspiration amygdalectomy combined with fornixtransection need not produce an impairment. Second, as Gouletet al.(1998) recently showed, although there is a small direct projectionfrom the perirhinal cortex to the mediodorsal thalamus, independentof synaptic relays in the hippocampus or amygdala, this directprojection passes through or very near to the amygdala, and istherefore interrupted by aspiration amygdalectomy carried out in thesame way as in the present experiment. Third, bilateral lesionsrestricted to the medial portion of the magnocellular part of themediodorsal nucleus, which is the portion that receives limbic andperirhinal afferents, had only a very mild effect on delayed matching-to-sample when assessed in the same way as in the present experiment(Parkeret al., 1997). For these reasons the impairment we have seenafter disconnecting the perirhinal cortex from the whole of themagnocellular part of the mediodorsal nucleus cannot be explainedby the direct or limbic interactions between these two structures.Instead, we propose that because of the extensive efferent and afferentconnections of the magnocellular part of the mediodorsal nucleuswith the frontal cortex, a unilateral lesion of the whole magnocellularmediodorsal nucleus is sufficient to produce widespread dysfunctionof the ipsilateral prefrontal cortex. Thus, when crossed with a unilateralperirhinal cortex ablation, a unilateral lesion of the magnocellularmediodorsal nucleus produces a functional effect similar to the effectof a large unilateral prefrontal cortex lesion, although somewhatmilder (Fig. 6).


This conclusion leaves open the question by what route the frontaland temporal cortex interact with each other in visual recognitionmemory. In principle, one possible route of interaction between theperirhinal cortex and the frontal lobe is via the corpus striatum.However, Wanget al. (1990) reported in abstract that bilateral lesionsof the ventral putamen and tail of caudate, aimed at destroying allthe area of the corpus striatum that receives afferents from thetemporal cortex, had no effect on visual recognition memory. Althoughevaluation of this evidence must await its full publication, the reportby Wang et al. gives no support to the idea of interaction viathe corpus striatum in visual recognition memory. Furthermore,electrophysiological recording from cells within the ventrocaudalneostriatum found that only a small proportion of cells show task-specific changes related to the reward contingencies of visual stimuli(Brown et al., 1995).

The most obvious route by which the perirhinal and frontalcortex might interact is by the direct and reciprocal cortico-corticalprojections which connect them within each hemisphere. Ungerleideret al. (1989) showed that the uncinate fascicle carries the directcortico-cortical projections between the prefrontal cortex and thecortex of the middle and inferior temporal gyri, and that theseconnections are cut off by surgical section of the uncinate fascicle.Gaffan & Eacott (1995) found that uncinate fascicle section had noeffect on delayed matching-to-sample, and concluded that directcortico-cortical projections between the temporal lobe and frontallobe were not important for this task. However, Meunieret al. (1997)pointed out that the study by Ungerleideret al. (1989) did notspecifically examine the connections of the perirhinal cortex, and thattherefore it is logically possible that some of the direct projectionsbetween the perirhinal cortex and frontal lobe might travel by someroute other than the uncinate fascicle, and might mediate visualrecognition memory in spite of Gaffan & Eacott’s (1995) findings.To evaluate this possibility we have examined in one monkey theeffect of crossed unilateral lesions of the perirhinal cortex in onehemisphere, and the frontal target of perirhinal cortex projections inthe other hemisphere. The unilateral frontal lesion in this animalremoved all the cortex on the orbital surface between the medial andlateral orbital sulci, including in the removal the lateral bank of thelateral orbital sulcus and the medial bank of the medial orbital sulcus.This ablation was designed to remove all the frontal cortex whichreceives a direct projection from the perirhinal cortex (Carmichael &Price, 1995). Our monkey was trained in delayed matching-to-samplewith reverse-order tests of lists of five trial-unique samples, exactlyas described in the present experiments. After preoperative trainingfollowed by unilateral ablation of perirhinal cortex alone, the monkeyperformed this task at 87.9% correct on average in 10 sessions oftesting; after the addition of the unilateral orbital lesion in thecontralateral hemisphere the monkey performed the task at 86.6%correct on average in a further 10 sessions. This observation stronglysupports Gaffan and Eacott’s conclusion that direct cortico-corticalprojections between the frontal lobe and temporal lobe are notnecessary for visual recognition memory.

However, these negative findings bear exclusively on the hypothesisthat recognition memory depends on frontal-temporal cortico-corticalinteraction by a direct, monosynaptic route. A separate possibility isthat interaction between perirhinal cortex and frontal cortex in visualrecognition memory takes place via indirect, multisynaptic cortico-cortical interactions, many of which must be left intact by uncinatefascicle section. In their 1989 study, Ungerleideret al. specificallyshowed that the frontal connections of the prestriate cortex wereunaffected by uncinate fascicle section, and it is reasonable to supposethat the same is true of parietal cortex. Furthermore, the perirhinal


cortex can have widespread influence on other cortical areas via itsstrong back-projection to area TE (Saleem & Tanaka, 1996). Thishypothesis of indirect cortico-cortical interaction is illustrated inFig. 7.

The present evidence needs to be considered alongside evidencefrom temporal-frontal disconnection effects in other tasks. Parker &Gaffan (1997, 1998) showed that frontal cortex ablation in onehemisphere combined with ablation of both the perirhinal cortex andarea TE in the opposite hemisphere did not disrupt the learning ofvisual object–reward associations in an object discrimination learningtask. However, this combination of lesions did produce a severeimpairment in the performance and learning of visual conditionaldiscriminations (Parker & Gaffan, 1998) and of a visual strategyimplementation task (Parker & Gaffan, 1997). Thus, recognitionmemory is like conditional learning or strategy implementation,and unlike object–reward associative learning, in requiring frontal-temporal interaction. There are some similarities between conditionallearning and recognition memory. In delayed matching-to-sample, theanimal must learn to use a particular strategy for the obtaining offood reward, i.e. the choice of familiar visual objects, and this strategyis quite different from the animal’s usual strategy for obtaining foodwhich is to approach objects that are associated in memory withfood. This latter strategy applies not only to the animal’s extra-experimental experience of food, but also to some aspects of thebehaviour required in the experimental sessions, e.g. approaching thefood hopper when the pellet dispenser operates, or the lunchbox whenit opens. The animal must therefore maintain two different strategiesfor obtaining food, each appropriate to a specific set of circumstances,and this is similar to a conditional discrimination.

In conditional discriminations and in delayed matching-to-sample,the animal’s choice between visual objects is dictated not only bythe memories directly attached to an object itself but also by thememories retrieved by other sources of information, often non-visualand in any case extrinsic to the object chosen. Extrinsic, non-objectinformation indicates to the animal that this is the appropriate contextfor choosing this particular object, in a conditional discrimination, orfor choosing a familiar but as yet unrewarded object, in delayedmatching-to-sample. In an object discrimination learning task, bycontrast, the animal’s strategy of approaching objects which areassociated in memory with food reward is not so context specific,and may be simply the default strategy for obtaining food—thenormal strategy which is adopted in the absence of any contextualinformation to indicate that it is temporarily to be eschewed in favourof some other way of getting food. If so, choices in the objectdiscrimination learning task may be controlled solely by object–reward associations in memory, while in delayed matching-to-sampleand conditional learning, choices are dictated by the combination ofcontextual information with object memories. The requirement forfrontal interaction with visual object memories retrieved in thetemporal lobe varies from task to task, according to this view, as thetasks demand more or less integration of object memories with otherkinds of information. Certainly, the widespread and diffuse sourcesof information which flow into the frontal lobe make it an appropriatesite for the integration of disparate sources of information. This viewis similar to that which was put forward by Gaffan & Harrison (1991)in discussing temporal-frontal disconnection effects in an auditory-visual conditional task.

Acknowledgements

This research was supported by the UK Medical Research Council. We thankJudi Wakeley and Colin Akerman for help in training the monkeys.


Abbreviations

AD anterodorsal nucleusAM anteromedial nucleusAV anteroventral nucleusCa caudate nucleusCeM central medial nucleusCL central lateral nucleusCM central median nucleusCSL central superior lateral nucleusFx fornixHl lateral dorsal nucleusHm medial habenular nucleusHpt habenulopeduncular tractLD lateral dorsal nucleusLi limitans nucleusLP lateral posterior nucleusMD mediodorsal nucleusMDdc mediodorsal nucleus, densocellular portionMDmc mediodorsal nucleus, magnocellular portionMDmf mediodorsal nucleus, multiformis portionMDpc mediodorsal nucleus, parvocellular portionMG medial geniculate complexMtt mamillothalamic tractPa paraventricular nucleusPc paracentral nucleusPf parafascicular nucleusPla anterior pulvinar nucleusPo posterior nucleusR reticular nucleusRe reuniens nucleusRh rhomboid nucleusSm stria medullarisSN substantia nigraSt stria terminalisSub subthalamic nucleusVA ventral anterior nucleusVLa ventral lateral nucleus, anterior portionVLp ventral lateral nucleus, posterior portionVMb ventral medial nucleus, basal portionVMp ventral medial nucleus, principal portionVPI ventral posterior inferior nucleusVPL ventral posterior lateral nucleusVPM ventral posterior medial nucleusZi zona incerta

References

Aggleton, J.P. & Mishkin, M. (1983a) Visual recognition impairment followingmedial thalamic lesions in monkeys.Neuropsychologia, 21, 189–197.

Aggleton, J.P. & Mishkin, M. (1983b) Memory impairments followingrestricted medial thalamic lesions in monkeys.Exp. Brain Res., 52, 199–209.

Bachevalier, J. & Mishkin, M. (1986) Visual recognition impairment followsventromedial but not dorsolateral prefrontal lesions in monkeys.Behav.Brain Res., 20, 249–261.

Bachevalier, J., Parkinson, J.K. & Mishkin, M. (1985) Visual recognition inmonkeys: effects of separate vs. combined transection of fornix andamygdalofugal pathways.Exp. Brain Res., 57, 554–561.

Brown, V.J., Desimone, R. & Mishkin, M. (1995) Responses of cells in thetail of the caudate nucleus during visual discrimination learning.J.Neurophys, 74, 1083–1095.

Buckley, M.J. & Gaffan, D. (1997a) Impairment of visual object discriminationlearning after perirhinal cortex ablation.Behav. Neurosci., 111, 467–475.

Buckley, M.J. & Gaffan, D. (1997b) Learning and transfer of object-rewardassociations and the role of the perirhinal cortex.Behav. Neurosci., 112,15–23.

Buckner, R.L. & Peterson, S.E. (1996) What does neuroimaging tell us aboutthe role of prefrontal cortex in memory retrieval?Sem. Neurosci., 8, 47–55.

Carmichael, S.T. & Price, J.L. (1995) Limbic connections of the orbital andmedial prefrontal cortex in macaque monkeys.J. Comparative Neurol.,363, 615–641.

Eacott, M.J., Gaffan, D. & Murray, E.A. (1994) Preserved recognition memoryfor small sets, and impaired stimulus identification for large sets, followingrhinal cortex ablation in monkeys.Eur. J. Neuroscience, 6, 1466–1478.


Gaffan, D. & Eacott, M.J. (1995) Uncinate fascicle section leaves delayedmatching-to-sample intact, both with large and small stimulus sets.Exp.Brain Res., 105, 175–180.

Gaffan, D. & Harrison, S. (1987) Amygdalectomy and disconnection in visuallearning for auditory secondary reinforcement by monkeys.J. Neuroscience,7, 2285–2292.

Gaffan, D. & Harrison, S. (1991) Auditory–visual associations, hemisphericspecialization and temporal–frontal interaction in the rhesus monkey.Brain,114, 2133–2144.

Gaffan, D. & Murray, E.A. (1990) Amygdalar interaction with the mediodorsalnucleus of the thalamus and the ventromedial prefrontal cortex in stimulus-reward associative learning in the monkey.J. Neuroscience, 10, 3479–3493.

Goldman-Rakic, P.S. & Porrino, L.J. (1985) The primate mediodorsal (MD)nucleus and its projections to the frontal lobe.J. Comp. Neurol., 242,535–560.

Goulet, S., Dore´, F.Y. & Murray, E.A. (1998) Aspiration lesions of theamygdala disrupt the rhinal corticothalamic projection system in rhesusmonkeys.Exp. Brain Res., 119, 131–140.

Meunier, M., Bachevalier, J. & Mishkin, M. (1997) Effects of orbital frontaland anterior cingulate lesions on object and spatial memory in rhesusmonkeys.Neuropsychologia, 35, 999–1015.

Meunier, M., Bachevalier, J., Mishkin, M. & Murray, E.A. (1993) Effects onvisual recognition of combined and separate ablations if the entorhinal andperirhinal cortex in rhesus monkeys.J. Neuroscience, 13, 5418–5432.

Murray, E.A., Gaffan, E.A. & Flint, R.W. (1996) Anterior rhinal cortex andamygdala: dissociation of their contributions to memory and food preferencein rhesus monkeys.Behav. Neurosci., 110, 30–42.


Murray, E.A. & Mishkin, M. (1996) 40-Minute visual recognition memory inrhesus monkeys with hippocampal lesions.Soc. Neurosci. Abs, 22, 281.

Noble, J. (1973) Interocular transfer in the monkey: rostral corpus callosummediates transfer of object learning set but not of single problem learning.Brain Res., 50, 147–162.

Parker, A., Eacott, M.J. & Gaffan, D. (1997) The recognition memory deficitcaused by mediodorsal thalamic lesion in non-human primates: a comparisonwith rhinal cortex lesion.Eur. J. Neurosci., 9, 2423–2431.

Parker, A. & Gaffan, D. (1997) Frontal/temporal disconnection in monkeys:memory for strategies and memory for visual objects.Soc. Neurosci. Abs,23, 11.

Parker, A. & Gaffan, D. (1998) Memory after frontal/temporal disconnectionin monkeys: conditional and non-conditional tasks, unilateral and bilateralfrontal lesions.Neuropsychologia, 36, 259–271.

Russchen, F.T., Amaral, D.G. & Price, J.L. (1987) The afferent input to themagnocellular division of the mediodorsal thalamic nucleus in the monkey,macaca fascicularis.J. Comp. Neurol., 256, 175–210.

Saleem, K.S. & Tanaka, K. (1996) Divergent projections from the anteriorinferotemporal area TE to the perirhinal and entorhinal cortices in themacaque monkey.J. Neuroscience, 16, 4757–4775.

Snedecor, G.W. & Cochran, W.G. (1967)Statistical Methods. Iowa StateUniversity Press, Iowa.

Ungerleider, L.G., Gaffan, D. & Pelak, V.S. (1989) Projections from inferiortemporal cortex to prefrontal cortex via the uncinate fascicle in Rhesusmonkeys.Exp. Brain Res., 76, 473–484.

Wang, J., Aigner, T. & Mishkin, M. (1990) Effects of neostriatal lesions onvisual habit formation in rhesus monkeys.Soc. Neurosci. Abs, 16, 617.

Zola-Morgan, S. & Squire, L.R. (1985) Amnesia in monkeys after lesions ofthe mediodorsal nucleus of the thalamus.Ann. Neurol., 17, 558–564.

Interaction of frontal and perirhinal cortices in visual...

Documents

Transcript of Interaction of frontal and perirhinal cortices in visual...