Brain Research, 190 (1980) 347-368 © Elsevier/North-HollandBiomedicalPress 347 CORTICAL AND SUBCORTICAL AFFERENTS TO THE A M Y G D A L A OF THE RHESUS MONKEY ( M A C A C A M U L A T T A ) J. P. AGGLETON, M. J. BURTON and R. E. PASSINGHAM Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD (U.K.) (Accepted October 25th, 1979) Key words: amygdala -- monkey -- horseradish peroxidase SUMMARY The afferent projections to the primate amygdala were studied using horseradish peroxidase. The potential advantages of this technique are discussed compared with those previously used to determine amygdaloid afferents. The findings indicate that certain agranular or dysgranular cortical regions may project directly to the amygdala: in particular, the orbital frontal cortex, anterior cingulate gyrus, subcallosal gyrus, temporal pole and anterior insula. These projections probably terminate predominantly in either the lateral or accessory basal nuclei. Other cortical projections from the inferotemporal and superior temporal gyri are described. Evidence was found for a heavy projection from the superior temporal sulcus to the lateral nucleus. Subcortical afferents were found from the hypothalamus, substantia innominata, diagonal band, thalamus, periaqueductal central gray, peripeduncular nucleus and from a band of cells extending medially from the peripeduncular nucleus to the midline, just ventral to the thalamus. In the thalamus, labelled cells were restricted to the non-specific nuclei, and were common in the rostral midline nuclei. No projection was observed from the dorsomedial nucleus of the thalamus. We discuss the implications of these results for interpreting the functions of the amygdala. INTRODUCTION It has frequently been supposed that the amygdala plays a role in the control of emotion and motivation12,1L One clue to the function of the amygdala is provided by knowledge of its anatomical connections, and by inference the classes of information it receives. Since lesions of this area are made in the treatment of epileptic, hyperkinetic and hyperaggressive disorders it is clearly of crucial importance that our understanding of these connections be as accurate and refined as possible. ....... ill, lit A i ...... 1 349 There are, however, problems in interpreting some of the relevant anatomical evidence. Neuronographic studies a4,45 indicated that the amygdala was directly interconnected with the orbital frontal, subcallosal and cingulate cortex and with the anterior insula and the temporal pole. Evidence that some fibre pathways do not respond to strychnine 11,57 or can show transsynaptic synchronous conduction 57 leaves these results inconclusive. More reliable information concerning afferents to the amygdala has been obtained from anterograde degeneration studies 15,~9. Although such studies have been successful in looking at cortical projections, subcortical projections have been more difficult to elucidate by this technique. Interpretation is difficult because fibres of passage may be damaged, as well as afferent axons projecting from the area of the lesion. Thus Nauta a° made lesions of the dorsomedial nucleus of the thalamus and reported terminal degeneration in the amygdala; but it is unclear if this resulted from damage to fibres traversing the lesion site and not arising from it. This technique is also inefficient if the aim is to delineate the precise boundaries of the areas which project into a structure, or to produce an exhaustive list of its afferents. These objections can be largely overcome by utilizing the retrograde axonal transport of horseradish peroxidase (HRP) 31. Although there is evidence that H R P can be taken up by severed fibres of passage 6°, the damage to such fibres may be assumed to be far less than in a comparable lesion degeneration study. In the present study, horseradish peroxidase was injected into the amygdala in a series of rhesus monkeys and the brains then studied for evidence of retrograde cell labelling. Injections were made into subareas of the amygdala, as defined cytoarchitectonically9. There is evidence that the different amygdaloid areas may subserve different functions 23, and by making localized injections we hoped to discover not only which areas project to the amygdala but also in which nuclei they terminate. METHODS Subjects Nine rhesus monkeys (Macaca mulatta) were used. Of these, 5 had previously been used for behavioural studies and had bilateral lesions. Two (nos. 182 and 113) had lesions confined to the sulcus principalis in the prefrontal cortex. Two (nos. 103 and 105) had lesions of the dorsolateral prefrontal cortex, anterior to the arcuate sulcus and including sulcus principalis and the cortex of the dorsal convexity. One had the superior colliculi removed (no. 231). The extent of the cortical lesions is shown in Figs. 3, 4 and 5. Fig. 1. Diagram of the sites of the HRP injections, plotted on coronal sections taken through the extent of the amygdala. The cross-hatched area indicates the extent of the HRP reaction product and the blackened region marks the centre of the injection (see the section 'Analysis'). The large numerals identify the animals. The small numerals refer to the number of microlitres of HRP (50 %) injected. Abbreviations : A, anterior area of amygdala; Ab, accessory basal nucleus; AC, anterior commissure; Bla, basolateral nuclei; C, central nucleus; CA, caudate; CL, claustrum; Co, cortical nucleus; GP, globus pallidus; La, lateral nucleus; Lb, lateral basal nucleus; Mb, medial basal nucleus; Me, medial nucleus; OC, optic chiasm; OT, optic tract; P, putamen; Pam, periamygdaloid nucleus; Pyr, prepiriform cortex; RS, rhinal sulcus; SF, sylvian fissure; STS, superior temporal sulcus. 020 65 0-20 Fig. 2. Diagram of the site o f the H R P injections. Conventions as in Fig. I. 351 Injections Operations were carried out under intravenous pentothal anaesthesia. A 1 #1 Hamilton syringe was used to inject 0.1-0.85 #1 of 50 ~ horseradish peroxidase (Sigma Type IV) dissolved in a 0.9 ~o physiological saline solution. A single unilateral injection was hydraulically delivered over a period of 10 rain. Where previously operated animals were used, care was taken to match their injections with those made into unoperated animals. It was not possible to rely on the stereotaxic co-ordinates given by the atlas of Snider and Lee 5°, since the animals varied considerably in size (5.0-8.3 kg) and the stereotaxic co-ordinate of any structure differs with the size of the animal 55. In previous investigations in the rhesus monkey we found that the amygdala maintains a constant position with respect to the posterior margin of the sphenoid bone, despite variations in brain and skull size. The sphenoid bone was visualised using X-ray photographs taken during the course of the operation (typically 90 kV at 30 mA for 0.5 sec using Kodak X-ray plates). The anteroposterior and dorsoventral coordinates were taken relative to the posterior tip of the sphenoid, whilst the lateral co-ordinate, which is the least variable between animals 55 was calculated on the basis of previous experiments. The animals were sacrificed 24 h after the injection. Those animals which had been previously operated upon and had lesions were kept under urethane anaesthesia during the intervening period, to comply with U.K. Home Office regulations. All of the animals were then given a lethal intravenous injection of barbiturate anaesthetic (Nembutal) and perfused with a mixture of 2 . 5 ~ paraformaldehyde and 1.5~ glutaraldehyde in 0. I M phosphate buffer. The brains were blocked in the stereotaxic vertical plane, as shown in Figs. 3, 4 and 5. The brains were kept in 30 ~ sucrose buffer for the next 48-72 h at pH 7.2 at 4 °C. They were then cut in 50 #m coronal sections on a freezing microtome and every section was retained. Two l-in-10 series of adjacent sections were saved and these were reacted with diaminobenzidine (BDH) 14 and 0.4 9/00 o-dianisidine (Sigma) 7 respectively. The sections were mounted and the dianisidine series was then counterstained with cresyl violet and the diaminobenzidine (DAB) series with thionine. This material was studied microscopically with both bright- and dark-field illumination. In one animal (no. 105) only a diaminobenzidine series was prepared. The reaction proved to be significantly more sensitive with dianisidine than DAB, both in the area of the injection site and in the degree of retrograde labelling. The majority of the results were therefore derived from the dianisidine series, of which every section was studied. The overall picture from the DAB series was similar in the location of the cells found, and only differed quantitatively. Analysis The position and extent of the HRP injections were plotted on to a standardized series of coronal sections, taken throughout the extent of the amygdala as shown in Figs. 1 and 2. That part of the reaction product which was so dense that no underlying cell structures could be discerned microscopically is shown in black. The remaining extent of the reaction product is shown by the hatched lines. This is not meant to imply 352 that the area of HRP uptake is equivalent to the area shown in black, although there is good evidence that the area of active HRP uptake is less than the full extent of the reaction product 60. Some indication of the extent of uptake could be inferred by reference to previous studies of amygdala afferents 15, and by comparing the results of overlapping injections in the present study. Despite partially overlapping regions of reaction product, evidence of discrete zones of HRP uptake is supplied by the consistently different distributions of labelled cells (nos. 113, 182 and 355). In animal no. 355 the injection (Fig. 2), although centred on the lateral basal nucleus, extended into several other nuclei; but the striking lack of labelled cells in the temporal lobe (Fig. 4) in this animal suggests that the area of active HRP uptake was confined to the lateral basal nucleus 15. In addition it was possible to assess the contribution of uptake from areas outside the amygdala by comparing the largest with the more discrete injections. Animals nos. 105 and 103 both showed the greatest spread of HRP from the injection site into adjoining structures, in particular the putamen, claustrum and globus pallidus (Fig. 1). In both animals label was found in caudal inferotemporal cortex, area 6, posterior cingulate cortex, somatosensory area I1, and thalamic nuclei including nucleus ventralis posteralis (VPM and VPLo) and nucleus ventralis lateralis (VLm and VLo). These areas were not labelled with more restricted injections. This indicates that the occasional striatal HRP reaction product seen in those animals with more restricted injections did not contribute to the overall pattern of labelling. Thus active uptake probably did not occur at the striatal periphery of the injection site in these animals. It was of some concern that HRP might be taken up by damaged fibres of passage, especially as we often damaged the caudal end of the anterior commissure in our approach to the amygdala. But the label seen in the contralateral hemisphere did not correspond to the projections of the anterior commissure 62 except in animals 103 and 105. Also, animal 355 showed virtually no label in the temporal lobe, although there was damage to the caudal part of the anterior commissure and the spread from the injection extended up as far as the commissure and into the lateral white matter. We therefore think it unlikely that, with the probable exceptions of animals 103 and 105, there was significant uptake of HRP by the fibres of the anterior commissure. RESULTS The position and extent of the cortical label in the ipsilateral hemisphere was plotted on standardized diagrams (Figs. 3, 4 and 5). The sylvian fissure (Figs. 3, 4 and 5) and the superior temporal sulcus (STS) (Fig. 6) were opened up to allow the representation of labelled cells found within them. Each dot in the diagram represents one labelled cell. The concentration of label in other sulci is not represented on the surface view, though its extent is shown by dots lying alongside the sulci (Figs. 3, 4 and 5). The vertical dashed lines mark the caudalmost sections that were saved and studied. Where no labelled cells are reported, none were found. Though the contralateral hemisphere was scanned, the results reported here are almost exclusively for the , I : '~ ;li I tt, i Fig. 3. Diagram of cortical labelled cells. Each dot represents one labelled cell exceptfor certain sulci, as explained in the section on 'Analysis". The blackened areas indicate the extent of removal of prefrontal cortex. The sylvian fissure has been opened up to expose the insula. Sections were saved and studied anterior to the vertical dashed line. The position and extent of the respective injections is shown on the right. Abbreviations: CC, corpus callosum; CIN, cingulate sulcus; CS, central sulcus; INS, insula; IP, intraparietal sulcus; IT, inferotemporal sulcus; OS, orbital sulcus; OT, occipitotemporal sulcus; RS, rhinal sulcus; SF, sylvian fissure; SP, sulcus principalis; STS, superior temporal sulcus. , J Fig. 4. Conventions as for Fig. 3. / 355 ],,; tl i t,, l# ,..,. ', ,~ "-'- Fig. 5. C o n v e n t i o n s as for Fig. 3. ~ - ml ib UP Qm m CD f 357 ipsilateral hemisphere. The cortical nomenclature is based on Von Bonin and Bailey's study 54, except for the prefrontal cortex where the more detailed description provided by Walker s8 is used. Temporal lobe Evidence for temporal lobe afferents was first obtained after injections aimed at filling the amygdala (nos. 103 and 105). In these animals, the basolateral nuclei were comprehensively covered (Fig. 1), although some corticomedial areas were not involved. As there was spread of H R P into adjoining structures we subsequently made more discrete injections. In this way we could check that the projection was indeed to the amygdala and provide more detailed evidence concerning its site of termination. Injections involving the corticomedial areas (nos. 113, 231 and 65) provided a more complete picture of temporal-amygdaloid afferents. (a) lnferotemporal cortex. Injections centred on the lateral nucleus (nos. 104 and 182) produced the pattern of labelled cells shown diagrammatically in Fig. 4. A low concentration of label was found in the anterior inferotemporal area (TE), which continued ventrally into areas T H and TF. HRP-positive cells were found as far posterior as the anterior portion of the occipitotemporal sulcus. The labelled cells in the inferotemporal cortex were predominantly pyramidal, being confined to layers III and V. On the ventral surface of the temporal lobe (TH and TF) the majority were found in layer V, but as one moved more dorsally towards the superior temporal sulcus (TE), an increasing proportion were in layer III. Inferotemporal label was not seen after any other localized amygdaloid injections. (b) Superior temporal sulcus. Extensive injections which included the basolateral nuclei (nos. 103 and 105) (Fig. 1) resulted in high concentrations of labelled cells in both banks and fundus of the superior temporal sulcus (STS) which in animal 103 extended posteriorly to a level with the caudal extent of the sylvian fissure (Fig. 3). A similar distribution of label was found after injections centred on the lateral nucleus (nos. 104 and 182) (Figs. 4 and 6). In animal 104, labelled cells were found in all subareas of the rostral half of STS, but no cells were found in the caudal half of the sulcus, including most of OAa 4s. This sulcal area accounted for approximately one fifth of all the label found in animals 104 and 182. Evidence for similar but lighter contralateral labelling was found in those animals (nos. 104 and 182) with lateral amygdala injections. The ipsilateral label was mainly restricted to pyramidal cells. These were in layer III, especially in the deeper portion, and a few in layer V. The contralateral label was solely in layer III. Evidence for afferents to the medial nuclei of the amygdala were derived from only one animal (no. 65). After an injection of the medial and central nuclei with involvement of the substantia innominata an occasional cell could be seen in the banks of the STS (Fig. 3). Fig. 6. Diagram of projection from superior temporal sulcus and temporal lobe to the amygdala. Each dot represents one labelled cell. The sulcus has been opened up and shown in isolation to allow visualization of the depths of the sulcus. Sections were saved and studied anterior to the vertical line. Conventions as for Figs. 1 and 3. 358 (c) Superior temporal gyrus. The rostral half of the superior temporal gyrus contained only a few labelled cells after the largest injections, which included all of the basolateral nuclei (nos. 103 and 105) (Fig. 3). In both animals with injections centred on the lateral nucleus (nos 104 and 182) a few HRP-positive cells were located in the rostral portion of the superior temporal gyrus which merges with the temporal pole (TG) (Fig. 4). This label was found in pyramidal cells in layers llI and V. (d) Temporalpole (TG). Dense aggregations of labelled cells were observed after injections centred in either the lateral (nos. 104 and 182) or the accessory basal nuclei (no. 113) (Figs. 4 and 5). In animal 113 almost half of the labelled cells found outside the amygdala were located in the temporal pole, and in animals 104 and 182 the proportion was a fifth. In one animal (no. 355) where the injection was centred on the lateral basal nucleus the labelled cells in TG were very much reduced as compared with the more lateral and medial injections described above. Very light label was seen after corticomedial injections, although this may have been due to involvement of the substantia innominata (no. 65) (Fig. 3). In all animals the label was mainly found in the pyramidal cells of layer III, especially the deeper portions, with smaller numbers present in layer V. (e) Entorhinal andperirhinal areas. Few labelled cells were found in the entorhinal cortex after large amygdaloid injections (nos. 103 and 105) (Fig. 3), and even less label was seen after more restricted lateral injections (nos. 104 and 182) (Fig. 4). Injections filling the basolateral amygdala (nos. 103 and 105) produced a consistent pattern of labelled cells along both banks of the rhinal sulcus, but with more restricted lateral (nos. 104 and 182) and accessory basal nuclei (no. 113) injections this labelling was reduced to a sparse scattering of cells (Figs. 4 and 5). In one animal (no. 65), with an injection involving the central and medial nuclei and the area of the substantia innominata, dense labelling was observed from the rostral rhinal area. Although the perirhinal label was mainly found in subarea 35a, it was also present in Pr2 and 35b. This label derived from pyramidal cells in layer V, with a smaller number present in layer III. (f) Insula. The anterior two-thirds of the insula was labelled after injections into the basolateral nuclei. The caudal-third of the insula was only labelled following those larger amygdaloid injections (nos. 103 and 105) which spread to surrounding structures (Fig. 3). In animals with more restricted injections the insula and parainsula were labelled after injections into the lateral (nos. 104 and 182) and lateral basal nuclei (no. 355) (Fig. 4), while label was only found in the parainsular area following injections into the accessory basal nuclear area (nos. 113 and 403) (Fig. 5). A few scattered cells were observed in the insula after injections of the central nucleus and the substantia innominata (no. 65). The labelled cells were pyramidal cells in layers 1II and V. Some label was also observed in the contralateral rostral parainsula area in animal 104. The frontal lobe The considerations concerning injection size and placement, mentioned in the preceding section on the temporal cortex, also apply to the frontal lobe. In addition, 359 when considering prefrontal cortex, only the results from the 5 animals without prefrontal lesions are discussed. (a) Orbital prefrontal cortex. An injection centred on the lateral nucleus (no. 104) produced labelled cells in area 13, especially in the more posterior portion, the orbito-insula transition zone and in the orbital and frontomarginal sulci (Fig. 4). Injections into the lateral or accessory basal regions produced small amounts of label in area 14 (nos. 104 and 403) (Figs. 4 and 5). This orbitofrontal label arose from pyramidal cells in layers III and V. Many labelled cells were observed from all of area 13 and the caudal part of area 11 after an injection into the medial nucleus and the adjacent region of the substantia innominata (no. 65) (Fig. 3). (b) Dorsalprefrontal cortex. Injections into the basolateral nuclei (nos. 104, 355 and 403) (Figs. 4 and 5) failed to show any evidence of a projection from this region to the amygdala apart from a few cells in sulcus principalis (no. 104). However, a significant quantity of label was found in sulcus principalis and the superior prefrontal dimple 53, after an injection into the medial nucleus and substantia innominata region (no. 65) (Fig. 3). (c) Medialfrontal cortex. No label was observed in areas 10, 25 and 9 apart from a few cells found in the animal (no. 65) with a medial nucleus and substantia innominata injection. The anterior cingulate, above the genu of the corpus callosum, was significantly labelled but only after injections into the lateral nucleus (no. 104) (Fig. 4). This label was restricted to pyramidal cells in layer III. The subcallosal (FL) gyrus, however, was labelled after both basolateral (nos. 104 and 403) and medial amygdaloid injections (nos. 65 and 231) (Figs. 3, 4 and 5). It is of interest that, even though animal 113 had a lesion of sulcus principalis, it had the heaviest labelling seen in the subcallosal gyrus. This was after an injection centred in the accessory basal nucleus. The injection into the lateral nucleus (no. 104) labelled primarily large pyramidal cells in layer V. However, only small circular cells in the same layer were labelled after injections into the accessory basal nucleus (no. 113). As the precise boundary between areas 24 and 25 is unclear, some label in the most rostral cingulate or subcallosal cortex may have belonged to posterior 25. (d) Precentral cortex. The labelled cells found in precentral cortex in animals 103 and 105 almost certainly resulted from spread of H R P into the putamen (Fig. 3). It is unclear how to interpret the cluster of labelled cells found around the most posterior extent of the arcuate sulcus (no. 182) (Fig. 4). Parietal cortex The brains were blocked in such a way that in only one animal (no. 104) could the full extent of parietal cortex be examined. No labelled cells were found in this animal (Fig. 4). But in one of the animals with restricted injections (no. 182) a few cells were found on either bank in the depths of the intraparietal sulcus (Fig. 4). The difference between the results for animals 182 and 104 is difficult to explain given the general similarity between their injections. Subcortical structures (a) The amygdala. In some animals the spread of label from the injection site 360 precluded analysis of intra-amygdaloid projections. However, with the more discrete injections very dense label was present in the lateral portion of the lateral nucleus and claustral area of the amygdala after an injection into the accessory basal region (nos. 113 and 403). Conversely, the magnocellular part of the accessory basal nucleus was labelled following injections centred on the lateral nucleus (nos. 104 and 182). Both animals (nos. 65 and 231) injected in the dorsal amygdaloid region showed a heavy labelling in the caudal, lateral basal, and magnocellular accessory basal nuclei. These labelled cells within the amygdala were the most numerous single source of afferents found in the brains (nos. 113,403, 65 and 231) and it is highly probable that the spread of HRP from the injection site obscured other intra-amygdaloid connections. The interpretation of results for the amygdala is further clouded by the possible uptake of HRP by severed axons which arise from areas of the amygdala other than the injection site. (b) Substantia innominata (SI). Every animal examined had labelled cells in the SI area, which were very numerous in some cases. The SI was the major source of subcortical label in several animals (nos. 104, 182, 231 and 355). The greatest densities of cells appeared after injections into the lateral basal nucleus (no. 355). In animals with injections in the central or medial nucleus (nos. 65 and 231) the possibility that labelled cells resulted from the passive spread of HRP from the injection site could not be excluded. The horizontal limb of the diagonal band of Broca showed a similar, if less dense, pattern of label. Only a few cells were observed in the vertical limb and no labelled cells were found in the septal areas. No evidence of a topographic projection to the amygdala was found. (c) Hypothalamus. The hypothalamic area contained only a few scattered labelled cells and there was no evidence that any of these groups of cells was associated with any particular nuclear division of the amygdala. In animals with corticomedial involvement (nos. 113, 231 and 65) the hypothalamic labelling appeared to be heavier than with injections elsewhere and labelled cells could be found in the lateral, ventromedial, and dorsomedial hypothalamic nuclei. (d) Basal ganglia. In all of the animals cells were occasionally found in the caudate and globus pallidus. Only in the putamen was it possible to find evidence of a dense labelling, and this was in the two animals with extensive injections (nos. 103 and 105) which spread beyond the amygdala. This suggests that the label in the putamen was not derived from the amygdala. The claustrum was usually lightly labelled, independently of the injection site, though injections into the lateral basal (no. 355) or accessory basal nuclei (no. 113) resulted in the greatest amount of label, especially in the ventral claustrum lateral to the amygdala. (e) Thalamus (nuclei after Olszewski41). Thalamic label was found almost exclusively in non-specific nuclei with the exception of the magnocellular part of the ventral anterior (VAmc) nucleus which was labelled after the injections centred on the lateral, and possibly the lateral basal nuclei. The majority of the remaining labelled cells were found in the following midline and intralaminar thalamic nuclei: nucleus paraventricularis (Pa, Pac), nucleus centralis latocellutaris (Clc), nucleus centralis densocellularis (Cdc), nucleus centralis superior (Cs), nucleus centralis inferior (Cif), nucleus centralis 361 intermedialis (Cim), nucleus rotundis (Ro), nucleus reuniens (Re), nucleus paracentralis (Pcn), nucleus subfascicularis (Sf.pc and Sf.mc) and nucleus centrum medianum (Cn.Md). All these nuclei appeared to be labelled after amygdaloid injections and no evidence of specific topographic injections was observed. The lateral (nos. 104 and 182) and lateral basal (no. 355) nuclei injections, for example, produced label in all of these thalamic nuclei, the lateral basal nucleus producing the greatest number. Injections into the accessory basal area (nos. 113 and 403), however, only resulted in labelled cells in the nucleus paraventricularis (Pa). Very light label was present in nucleus medianum (Cn.Md) and the habenula following injections centred in the lateral nuclei (nos. 104 and 182). This is consistent with the similar but denser label seen in these same nuclei in those animals with the largest injections (nos. 103 and 105). The medial pulvinar was lightly labelled after injections centred on the lateral nuclei (nos. 104 and 182). None of the other thalamic nuclei, including the dorsomedial nucleus, were labelled. The midbrain, caudal diencephalon and brain stem afferents (a) The substantia nigra and peripeduncular nuclei. A few labelled cells were observed in the substantia nigra after injections to the dorsal amygdala and substantia innominata region (nos. 231 and 65). A few cells were also observable after lateral nucleus injections but the problems of endogenous B1 label make interpretation difficult. The majority of animals revealed a narrow band of spindle-shaped labelled cells running rostrally and diagonally just beneath the thalamus, between the peripeduncular nucleus and the nucleus subfascicularis parvocellularis. The labelled cells, which were quite numerous, occurred after injections into the basolateral nuclei and especially the lateral basal nucleus (no. 355). Labelled cells were occasionally found in the tegmental area stretching ventromedially from the peripeduncular nucleus and merging caudally with the nucleus reticularis tegmenti. (b) Central gray. The midbrain was studied up to the caudal extent of the periaqueductal central gray. The position of midbrain label was compared with the stereotaxic atlas of Snider and Lee 50; thus the nomenclature is taken from their study. All the animals examined showed a consistent but sparse scattering of labelled cells within the midbrain. This occurred in an area of large cells, nucleus medialis annuli aqueductus, ventromedial to the third ventricle in the central gray, and from a group of smaller cells which course ventrally around the boundary of the trochlear nucleus and interdigitate with the fibres of the fascicularis longitudinalis medialis and spread laterally around the ventral margin of the fibre bundle. It was also possible to find labelled cells bilaterally, running more caudally along the margin of the mesencephalic trigeminal nucleus, but these appeared to derive from endogenous label 61. Midbrain label was seen in all animals, but those with injections into the accessory basal nucleus (nos. 113 and 403) produced label only in the more ventral central gray. DISCUSSION The demonstration of labelled cells does not necessarily prove a projection to the 362 amygdala. HRP may be picked up and transported by damaged axons ,59, and thus great care is needed when interpreting results obtained by applying HRP to subcortical nuclei. Fibres from the piriform cortex are known to traverse the amygdala of the rat 8, although the routes of such connections in the primate are not known in detail. However, Van Hoesen and Pandya 5z have shown that fibres from the anterior piriform cortex pass medial to the basolateral nuclei. We do not think it likely that these fibres were extensively damaged in this study, and take as direct evidence the very low density of labelled cells found in the piriform cortex. It is more difficult to determine the precise zones of active HRP uptake and so dismiss the possibility of uptake by structures outside the amygdala. Evidence concerning this point has been discussed in the 'Analysis' section. It is likely that in animals 103 and 105 there was significant extra-amygdaloid involvement, and that this is reflected in the more widespread pattern of labelling. Furthermore, the apparent projections from sulcus principalis and the rhinal fissure in animal 65 point to some uptake of HRP from an area outside the amygdala, probably the substantia innominata. However, in general the projections suggested by this study are consistent with the findings of investigations using anterograde methodslS,lS,19,21, 4°. The anatomical literature also provides information on the areas which project to structures adjacent to the amygdata. Where we failed to find labelled cells in these areas we have supposed that there was no significant uptake of HRP by these neighbouring structures. We cannot make strong claims on the pattern of projections to specific amygdaloid nuclei. The injections of HRP were not sufficiently discrete and they might have involved fibres passing to adjacent nuclei within the amygdala. This problem has been discussed in the 'Analysis' section. We are reassured, however, by the similarity of the results reported here for the temporal lobe, and those reported by Herzog and Van Hoesen 15 using silver degeneration and autoradiography. Two generalizations are suggested by this study. The first is that the amygdala is closely connected with areas thought to be involved in the regulation of emotion and motivation. The second is that the projections from association cortex are most dense from the temporal lobe, and particularly so from the superior temporal sulcus (Fig. 6). The caudal orbital frontal cortex, temporal pole, anterior insula and the anterior cingulate and subcallosal gyri, all contain HRP-positive cells following injections into the amygdala. These results confirm earlier neuronographic studies 44,45. This label was most dense after injections involving the basolateral amygdala, in particular the lateral or accessory basal nuclei areas. In all of these areas the cortex is dysgranular or agranular a4. These cortical regions all appear to play some role in autonomic modulation. Thus Kaada 22 showed that in anaesthetized rhesus monkeys stimulation in any of these agranular areas could result in inhibition of respiration, pyloric contraction, changes in brood pressure and dilation of the pupils. Similar changes could not be found after stimulation of other cortical areas. The effects of amygdala stimulation seen in this and other studies were grossly similar to those of stimulating agranular cortex, suggesting a functional relationship. Other regions of consistent label were the substantia innominata and the hypothalamus. The hypothalamic label appeared relatively sparse in this study, although 363 evidence of a larger projection has been found in the rat s,a2 and squirrel monkey 19 using different techniques. The hypothalamus has traditionally been ascribed a central role in controlling emotion and motivation, and recent recording studies3, 3s have shown that the substantia innominata, like the lateral hypothalamus, has neurons which specifically respond to the motivational properties of stimuli. The much denser label in the substantia innominata may indicate that in the rhesus monkey this area may be a more important final link in modulating the activity of the amygdala. Neither the hypothalamus nor the substantia innominata appear to project solely to specific amygdala nuclei, although label in the hypothalamus was more frequent after medial or dorsal amygdala injections. Evidence was found of a connection from the periaqueductal central gray area of the midbrain to the amygdala. The results indicate a sparse but consistent projection to most of the amygdaloid nuclei. The central gray is thought to receive information related to pain24, 34. It is therefore of interest that the amygdala has been associated with responses to pain1,13, 5s and this projection may be important in mediating these effects. Furthermore, the highest concentrations of opiate receptors in the rhesus monkey brain 80 are in the amygdala and periaqueductal central gray. The thalamic afferents to the amygdala indicated by this study were confined to the non-specific nuclei and the magnocellular part of nucleus ventralis anterior (VAmc), which is often regarded as a non-specific nucleus. It is unlikely that this label was caused by the uptake of HRP outside the amygdala, as every animal showed some degree of thalamic labelling. These areas are thought to be rostral extensions of the reticular formation and their activity is believed to be associated with arousal and pain 1°,24,3s. Like the amygdala and periaqueductal gray the medial thalamus is rich in opiate receptors 30. We were unable to find any evidence of a projection from the dorsomedial nucleus (DMN) to the amygdala as described by Nauta 4°. Our failure to find this projection agrees with a previous report 37. It is possible that this pathway is insensitive to the HRP technique 49, but when Nauta lesioned the D M N and demonstrated degeneration in the amygdala he also damaged the non-specific midline nuclei lying medial to the DMN. In our study these nuclei appear to project to the amygdala, and so Nauta's finding could derive from damage to these midline structures, and not to the DMN. The pattern of amygdaloid degeneration he reports agrees with the sites of termination of midline thalamic afferents as demonstrated by HRP in the present study. The existence of an apparent amygdaloid projection from an area stretching from the peripeduncular nucleus rostrally and medially to the nucleus subfascicularis agrees with a previous brief report 37. The importance of this connection is as yet unknown. There is a close anatomical association between the agranular cortical areas, in particular the orbital frontal cortex, temporal pole and the amygdala. Experiments studying the effects of lesioning these 3 areas suggest that they are all crucial in maintaining appropriate social responses 27. Furthermore, some aspects of the KliiverBucy syndrome occur after removal of any of these structures 4,16,27,5s and the effect is greater if more than one of these areas is damaged27, 48. The syndrome es,Ss consists of: 364 (l) abnormal tameness and approach behaviour; (2) 'visual agnosia'; (3) dietary changes including meat eating and coprophagia; (4) orality; (5) hypermetamorphosis, or the abnormally rapid switching of attention between stimuli: and (6) hypersexuality. It has been proposed that these areas may represent part of a neural system essential for normal social behaviour 27. The anatomical results reported here support the view that these structures may be part of a functional system. The orbital frontal cortex and temporal pole both appear to project directly to the amygdala, especially to the lateral and accessory basal nuclei. These connections appear to be reciprocal 39 as are those between the temporal pole and the orbital frontal cortex21, 42. There is evidence of an important indirect pathway from the amygdala to the orbital frontal cortex via the magnocellular part of DMNaa, 40 and it has been supposed that this was reciprocal 4°. Our results indicate that if there is an indirect thalamic route from the orbital frontal cortex to the amygdala, it is via VAmc and the midline nuclei33, 39 and not the DMN. In view of the anatomical and functional interconnections of the orbitofrontal cortex, temporal pole and amygdala, it is of interest whether they share similar afferents. The rostral inferotemporal and rostral superior temporal gyri both project to the amygdala, temporal pole ~1 and orbital frontal cortex 6. Whilst the amygdala and orbitofrontal cortex share other similar projections such as midline thalamic nuclei 2°,26, VAmc 5,47 and substantia innominata25, the afferents to the temporal pole have not been studied in sufficient detail to enable a comparison to be made. The second main finding of this study was that the temporal lobe projects more densely to the amygdala than do other association areas. There have been several previous studies of temporal association cortex projections to the amygdala 15,21,59. With the use of HRP we have been able to confirm these connections and accurately delineate their sites of origin. Thus both anterior inferotemporal and anterior superior temporal gyri project to the amygdala, both apparently to the lateral nucleus. This site of termination agrees with previous studies, though we have been unable to show inferotemporal afferents to the more dorsal portions of the amygdala 15. The extensive label in the superior temporal sulcus suggests a projection which has not been previously described. It is of particular interest due to its size and extent. Both superior and inferior banks and the fundus of the sulcus were heavily labelled after injections into the lateral nuclear area (animals 104 and 182) (Fig. 6). There is no evidence that particular portions of the sulcus contribute to this projection, which extends to the level of the caudal limit of the sylvian fissure. There also appears to be a lighter contralateral projection, of similar extent, restricted to the lateral nucleus (animals 104, 182). This contralateral projection may comprise part of the anterior commissural projection which terminates in the lateral nucleus of the amygdala51. This sulcus supplies approximately one-fifth of all the label found in animals 104 and 182, making it numerically one of the most important projections to the lateral nucleus. Studies of the connections of primary sensory areas (somesthetic, visual and auditory) have shown that the superior temporal sulcus is one of the first areas of convergence of these respective association areas21, 48. Electrophysiological recording studies have confirmed the existence of polysensory cells in the rostral-half of the superior temporal 365 sulcus 2, although more detailed anatomical work 4s has indicated that the different association areas mainly project to different sites within the sulcus. That much of this sulcus projects to one amygdala area suggests a further degree of convergence, which is borne out by the finding of polysensory units within the amygdalaZL Machne and Segundo 35, recording from cells in the amygdala of anaesthetized cats, remark that 'the convergence of different sensory modalities upon single units was outstanding'. Most of the polysensory units they located were in the basolateral nuclei. Thus the superior temporal sulcus could be the main source of polysensory information to the amygdala. The existence of direct projections to the amygdala from visual (inferotemporal) and auditory (superior temporal) association areas suggests that the other cortical association areas may project directly to this structure. However, afferents were not conclusively demonstrated from somatosensory area II or parietal association cortex. A small group of labelled cells was found in the depths of the intraparietal sulcus in one animal (182) but not in others. Because of the levels at which the brains were blocked the full posterior extent of parietal cortex was only studied in one animal (no. 104). The lateral frontal cortex was only occasionally labelled in those 5 animals with intact frontal lobes. Injections to more dorsal and medial nuclei suggest dense intraamygdaloid connections. These findings parallel some of those of Krettek and Price 29 in the rat. The role of these intra-amygdaloid projections is unknown but they are clearly of great importance when considering functional subdivisions within this structure 23. These connections were more numerous than those found from any other one area in nos. 113,403, 231 and 65, and since the spread of HRP around the injection site may have obscured other intra-amygdaloid projections our results probably underestimate the size and extent of these connections. The density of these projections shows that the normal functioning of a particular nucleus may be partly dependent on the integrity of another nucleus with which it is interconnected. This consideration should be borne in mind when interpreting the effects of selective amygdaloid lesions. These anatomical results suggest that two types of information converge in the amygdala: external polysensory information, via temporal lobe association cortex; and internal motivational and visceral information from various regions including the substantia innominata, the hypothalamus and the dysgranular cortical areas. Our findings are compatible with the proposal that the amygdala plays a role in integrating information about the sensory aspects of stimuli and their motivational and emotional significance17. It is true that Sanghera et al. a6 were unable to find cells in the amygdala which modified their firing according to changes in the motivational significance of objects with which the monkeys were presented. But the proposal is more directly tested by analyzing the varied effects of amygdala lesions on behaviour: and for these effects it still provides the most parsimonious explanation. 366 ACKNOWLEDGEMENTS T h i s r e s e a r c h was s u p p o r t e d by M . R . C . G r a n t 971/1/397/B. W e are g r a t e f u l to Dr. V. H. Perry f o r helpful c o m m e n t s on the m a n u s c r i p t , and to I v o r H u g h e s a n d M a r y W a l k e r f o r the h i s t o l o g i c a l p r o c e s s i n g o f the brains. REFERENCES I Bagshaw, M. H. and Pribram, J. D., Effect of amygdalectomy on stimulus threshold of the monkey, Exp. NeuroL, 20 (1968) 197-202. 2 Bruce, C., Desimone, R. and Gross, C., Large visual receptive fields in a polysensory area in the superior temporal sulcus of the macaque, Neurosci. Abstr., 3 (1977). 3 Burton, M. J., Rolls, E. T. and Mora, F., Effects of hunger on the responses of neurons in the lateral hypothalamus to the sight and taste of food, Exp. Neurol., 51 (1976) 668-677. 4 Butter, C. M. and Snyder, D. R., Alterations in aversive and aggressive behaviour following orbital frontal lesions in rhesus monkeys, Arch. Neurobiol., 32 (1972) 525-565. 5 Carmel, P. W., Efferent projections of the ventral anterior nucleus of the thalamus in the monkey, Amer. J. Anat., 128 (1970) 159-184. 6 Chavis, D. A. and Pandya, D. N., Further observations on corticofrontal connections in the rhesus monkey, Brain Research, 117 (1976) 369-386. 7 Colman, D. R., Scalia, F. and Cabrales, E., Light and electron microscopic observations on the anterograde transport of horseradish peroxidase in the optic pathway in the mouse and rat, Brain Research, 102 (1976) 156-163. 8 Cowan, W. M., Raisman, G. and Powell, T. P. S., The connections of the amygdala, J. Neurol. Neurosurg. Psyehiat., 28 (1965) 137-151. 9 Crosby, E. C. and Humphrey, T., Studies of the vertebrate telencephaton. II. The nuclear pattern of the anterior olfactory nucleus, tuberculum olfactorium and the amygdaloid complex in adult man, J. comp. NeuroL, 74 (1941) 309-352. 10 Dong, K. W., Ryu, H. and Wagman, I. H., Nociceptive responses of neurons in medial thalamus and their relationship to spinothalamic pathways, J. Neurophysiol., 41 (1979) 1592-1613. 11 Frankenhauser, B., Limitations of method of strychnine neuronography, J. Neurophysiol., 14 (1951) 73-79. 12 Gloor, P., Amygdala. In J. Field (Ed.), Handbook of Physiology, Vol. 2, Neurophysiology, American Physiological Society, Washington, D.C., 1960, pp. 1395-1420. 13 Goddard, G. V., Functions of the amygdala, Psyehol. Bull., 62 (1964) 89-109. 14 Graham, R. C. and Karnovsky, M. J., The early stages of absorption of horseradish peroxidase in the proximal tubu[es of the mouse kidney: ultrastructural cytochemistry by a new technique, J. Histochem. Cytochem., 14 (1966) 291-302. 15 Herzog, A. G. and Van Hoesen, G. W., Temporal neocortical afferent connections to the amygdala in the rhesus monkey, Brain Research, 115 (1976) 57-69. 16 Horel, J. A., Keating, E. G. and Misantone, L. J., Partial Kffiver-Bucy syndrome produced by destroying temporal neocortex or amygdala, Brain Research, 94 (1975) 347-359. 17 Jones, B. and Mishkin, M., Limbic lesions and the problem of stimulus-reinforcement associations, Exp. Neurol., 36 (1972) 362-377. 18 Jones, E. G. and Burton, H., A projection from the medial pulvinar to the amygdala in primates, Brahz Research, 104 (1976) 142-147. 19 Jones, E. G., Burton, H., Saper, B. and Swanson, L. W., Midbrain, diencephalic and cortical relationships of the basal nucleus of Maynert and associated structures in primates, J. comp. Neurol., 167 (1976) 385-420. 20 Jones, E. G. and Leavitt, R. Y., Retrograde axonal transport and the demonstration of nonspecific projections to the cerebral cortex and striatum from the thalamic intralaminar nuclei in the rat, cat and monkey, J. comp. Neurol., 154 (1974) 349-378. 21 Jones, E. G. and Powell, T. P. S., An anatomical study of converging sensory pathways within the cerebral cortex of monkey, Brain Research, 93 (1970) 793-820. 22 Kaada, B. R., Somato-motor, autonomic and electrocorticographic responses to electrical stimulation of 'rhinencephalic' and other structures in primates, Acta physiol, scand., 24 (1951) suppl. 83, 1-260. 367 23 Kaada, B. R., Stimulation and regional ablation of the amygdaloid complex with reference to functional representations. In B. E. Eleftheriou (Ed.), The Neurobiology of the Amygdala, Plenum Press, New York, 1972, pp. 205-281. 24 Kerr, D. I. B., Haugen, F. P. and Melzack, R., Responses evoked by the brain stem by tooth stimulation, Amer. d. Physiol., 183 (1955) 253-258. 25 Kievit, J. and Kuypers, H. G. J. M., Subcortical afferents in the frontal lobe in the rhesus monkey shown by means of retrograde horseradish peroxidase transport, Brain Research, 85 (1975) 261-266. 26 Kievit, J. and Kuypers, H. G. J. M., Organization of the thalamo-cortical connexions to the frontal lobe in the rhesus monkey, Exp. Brain Res., 29 (1977) 299-322. 27 Kling, A. and Steklis, H. D., A neural substrate for affiliative behaviour in nonhuman primates, Brain Behav. Evol., 13 (1976) 216-238. 28 Kliiver, H. and Bucy, P. C., Preliminary analysis of functions of the temporal lobe in monkeys, Arch. Neurol. Psychiat., 42 (1939) 979-1000. 29 Krettek, J. E. and Price, J. L., A description of the amygdaloid complex in the rat and cat with observations on intra-amygdaloid axonal connections, J. comp. NeuroL, 178 (1978) 255-280. 30 Kuhar, M. J., Pert, C. B. and Snyder, S. H., Regional distribution of opiate receptor binding in monkey and human brain, Nature (Lond.), 245 (1973) 447-450: 31 LaVail, J. H. and LaVail, M. M., The retrograde intraaxonal transport of horseradish peroxidase in the chick visual system: a light and electron microscopic study, J. comp. Neurol., 187 (1974) 303-357. 32 Lammers, H. J., The neural connections of the amygdaloid complex in mammals. In B. E. Eleftheriou (Ed.), The Neurobiology of the Amygdala, Plenum Press, New York, 1972, pp. 123-144. 33 Leichnetz, G. R. and Astruc, J. Efferent connections of the orbitofrontal cortex in the marmoset (Saguinus oedipus), Brain Research, 84 (1975) 169-180. 34 Liebeskind, J. C. and Paul, L. A., Psychological and physiological mechanisms of pain, Ann. Rev. Psychol., 28 (1977) 41-60. 35 Machne, X. and Segundo, J. P., Unitary responses to afferent volleys in amygdaloid complex, d. Neurophysiol., 19 (1956) 232-240. 36 Mark, V. H., Ervin, F. R. and Yakolev, P. I., Stereotaxic thalamotomy. III. The verification of anatomical lesion sites in the human thalamus, Arch. NeuroL, 8 (1963) 228-238. 37 Mehler, W. R., Pretorius, J. and Pretorius, J. A., Affeent connections of the amygdala in monkey, Anat. Rec., 190 (1978) 477-478. 38 Mora, F., Rolls, E. T. and Burton, M. J., Modulation during learning of the responses of neurons in the lateral hypothalamus to the sight of food, Exp. NeuroL, 53 (1976) 508-519. 39 Nauta, W. J. H. Fibre degeneration following lesions of the amygdaloid complex in the monkey, J. Anat. (Land.), 95 (1961) 515-530. 40 Nauta, W. J. H., Neural associations of the amygdaloid complex in the monkey, Brain, 85 (1962) 505-520. 41 Olszewski, J., The Thalamus of the Macaca mulatta: an Atlas for Use with the Stereotaxic Instrument, Karger, Basel, 1952. 42 Pandya, D. N. and Kuypers, H. G. J. M., Cortico-cortical connections in the rhesus monkey, Brain Research, 13 (1969) 13-36. 43 Pribram, K. H. and Bagshaw, M., Further analysis of the temporal lobe syndrome utilizing fronto-temporal ablations, J. comp. Neurol., 99 (1953) 347-375. 44 Pribram, K. H., Lennox, M. A. and Dunsmore, R. H., Some connections of the orbito-frontaltemporal, limbic and hippocampal areas of Macaca mulatta, J. Neurophysiol., 13 (1950) 127-135. 45 Pribram, K. H. and Maclean, P. D., Neuronographic analysis of medial and basal cerebral cortex. II. Monkey, d. Neurophysiol., 16 (1953) 324-340. 46 Sanghera, M. K., Roils, E. T. and Roper-Hall, A., Visual responses of neurons in the dorsolateral amygdala in the alert monkey, Exp. Neurol., 63 (1979) 610-626. 47 Scheibel, M. E. and Scheibel, A. B., The organization of the ventral anterior nucleus of the thalamus. A golgi study, Brain Research, 1 (1966) 250-268. 48 Seltzer, B. and Pandya, D. N., Afferent cortical connections and architectonics of the superior temporal sulcus and surrounding cortex in the rhesus monkey, Brain Research, 149 (1978) 1-24. 49 Siegel, A., Fukushima, T., Meibach, R., Burke, L., Edinger, H. and Weiner, S., The origin of the afferent supply to the mediodorsal thalamic nuclei: enhancement of HRP transport by selective lesions, Brain Research, 135 (1977) 11-23. 368 50 Snider, R. S. and Lee, J. C., A Stereotaxic Atlas o[" the Monkey Brain (Maeaca mulatta), The University of Chicago Press, Chicago, 111., 1961. 51 Turner, B. H., Mishkin, M. and Knapp, M. G. E°, Distribution of the anterior commissure to the amygdaloid complex in the monkey, Brain Research, 162 (1979) 331-337. 52 Van Hoesen, G. W. and Pandya, D. N., Some connections of the entorhinal (area 28) and perirhinal (area 35) cortices of the rhesus monkey. 1. Temporal lobe afferents, Brain Research, 95 (1975) 1-24. 53 Vanegas, H., Hollander, H. and Distel, H., Early stages of uptake and transport of horseradish peroxidase by cortical structures, and its use for the study of local neurons and their processes, J. comp. Neurol., 177 (1978) 193-212. 54 Von Bonin, G. and Bailey, P., The Neocortex of Macaca mulatta, University of Illinois Press, Urbana, II1., 1947. 55 Wagman, I. H., Loeffler, J. R. and McMillan, J. A., Relationship between growth of brain and skull of Macaca mulatta and its importance for the stereotaxic technique, Brain Behav. EvoL, 12 (1975) 116-134. 56 Walker, A. E., A cytoarchitectural study of the prefrontal area of the macaque monkey, J. comp. NeuraL, 73 (1940) 59-86. 57 Wall, P. D. and Horwitz, N. H., Observations on the physiological action of strychnine, J. Neurophysiol., 14 (1951) 257-263. 58 Weiskrantz, L., Behavioural changes associated with ablations of the amygdaloid complex in monkeys, J. camp. physioL Psychol., 49 (1956) 381-391. 59 Whitlock, D. G. and Nauta, W. J. H., Subcortical projections from the temporal neocortex in Macaca mulatta, J. comp. NeuroL, 106 (1956) 183-212. 60 Winer, J. A., A review of the status of the horseradish peroxidase method in neuroanatomy, Biobehav. Rev., 1 (1977)45-54. 61 Wong-Riley, M. T. T., Endogenous peroxidase activity in brain stem neurons as demonstrated by their staining with diaminobenzidine in normal squirrel monkeys, Brain Research, 108 (1976) 257-277. 62 Zeki, S. M., Comparison of the cortical degeneration in the visual regions of the temporal lobe of the monkey following section of the anterior commissure and the splenium, J. comp. Neurol., 148 (1973) 167-176.