cortical connections of the auditory cortex in marmoset monkeys: core and medial belt regions

Cortical Connections of the AuditoryCortex in Marmoset Monkeys: Core and

Medial Belt Regions

LISA A. DE LA MOTHE,1 SUZANNE BLUMELL,2 YOSHINAO KAJIKAWA,1

AND TROY A. HACKETT1,2*1Department of Psychology, Vanderbilt University, Nashville, Tennessee 37203

2Department of Hearing and Speech Sciences, Vanderbilt University School of Medicine,Nashville, Tennessee 37203

ABSTRACTThe auditory cortex of primates contains a core region of three primary areas surrounded

by a belt region of secondary areas. Recent neurophysiological studies suggest that the beltareas medial to the core have unique functional roles, including multisensory properties, butlittle is known about their connections. In this study and its companion, the cortical andsubcortical connections of the core and medial belt regions of marmoset monkeys werecompared to account for functional differences between areas and refine our working modelof the primate auditory cortex. Anatomical tracer injections targeted two core areas (A1 andR) and two medial belt areas (rostromedial [RM] and caudomedial [CM]). RM and CM hadtopographically weighted connections with all other areas of the auditory cortex ipsilaterally,but these were less widespread contralaterally. CM was densely connected with caudalauditory fields, the retroinsular (Ri) area of the somatosensory cortex, the superior temporalsulcus (STS), and the posterior parietal and entorhinal cortex. The connections of RM favoredrostral auditory areas, with no clear somatosensory inputs. RM also projected to the lateralnucleus of the amygdala and tail of the caudate nucleus. A1 and R had topographicallyweighted connections with medial and lateral belt regions, infragranular inputs from theparabelt, and weak connections with fields outside the auditory cortex. The results indicatedthat RM and CM are distinct areas of the medial belt region with direct inputs from the core.CM also has somatosensory input and may correspond to an area on the posteromedialtransverse gyrus of humans and the anterior auditory field of other mammals. J. Comp.Neurol. 496:27–71, 2006. © 2006 Wiley-Liss, Inc.

Indexing terms: multisensory; somatosensory; entorhinal; amygdala; striatum; homology; primate;

interhemispheric

In recent years we have adopted a model of auditorycortical organization in primates based on findings com-piled from both Old and New World primates (for reviews,see Pandya, 1995; Kaas and Hackett, 1998; Rauschecker,1998; Kaas et al., 1999; Kaas and Hackett, 2000; Hackett,2002; Jones, 2003). The model defines the auditory cortexas the corpus of cortical areas that are the preferentialtargets of neurons in either the ventral (MGv) or dorsal(MGd) divisions of the medial geniculate complex (MGC).By this definition, three regions of the superior temporalcortex comprise the auditory cortex in primates: core, belt,and parabelt (Fig. 1). Numerous cortical regions outsidethe boundaries of auditory cortex also process auditoryinformation. These include areas in the rostral superiortemporal gyrus (STGr), temporal pole, superior temporal

sulcus (STS), and posterior parietal and prefrontal cortex.Because these areas generally do not receive significantinputs from the MGC, and auditory activation is largelydependent on corticocortical inputs from some portion of

Grant sponsor: National Institutes of Health/National Institute on Deaf-ness and other Communications Disorders; Grant number: R01 04318 (toT.A.H.).

*Correspondence to: Troy A. Hackett, Vanderbilt University, 301 WilsonHall 111 21st Avenue South, Nashville, TN 37203.E-mail: [email protected]

Received 10 July 2005; Revised 11 September 2005; Accepted 4 Novem-ber 2005

DOI 10.1002/cne.20923Published online in Wiley InterScience (www.interscience.wiley.com).

THE JOURNAL OF COMPARATIVE NEUROLOGY 496:27–71 (2006)

© 2006 WILEY-LISS, INC.

auditory cortex, they are referred to as auditory-relatedcortex.

One major feature of the model is that the core, belt, andparabelt regions represent successive stages in the pro-cessing of auditory information in cortex. This hierarchywas introduced to account for patterns of connections be-tween areas and related physiological observations in pri-mates (Kaas and Hackett, 1998; Rauschecker, 1998). Keyanatomical support for a hierarchy is that that the coreregion projects to the belt, but not the parabelt, region(Morel et al., 1993; Morel and Kaas, 1992; Hackett et al.,1998a). Physiological evidence of progressive spectral andtemporal integration in the belt areas (Rauschecker et al.,1995; Recanzone, 2000a; Rauschecker, 2004; Kajikawa etal., 2005), as well as evidence that neuronal activity in CMis at least partly dependent on intact inputs from A1(Rauschecker et al., 1997).

A second feature of the model is that each of the threemajor auditory cortical regions consists of two or moreareas, or subdivisions (e.g., AI-R-RT, AL-ML-CL), inwhich thalamic and cortical inputs are processed in par-allel. Because the establishment of individual subdivi-sions depends on the identification of subsets of uniqueanatomical and physiological features, this component ofthe model is in the greatest need of refinement and vali-dation. The areas within the core region, especially A1,have been intensively studied, whereas several other sub-divisions were established from minimal anatomical orphysiological data. This is especially true of the belt re-gion, where as many as seven distinct areas have beenproposed (Fig. 1) but only a few studied in much detail, asdescribed below.

The current model groups all the belt areas togetherwithin a single region representing the second stage ofauditory cortical processing. This probably represents an

oversimplification, as several lines of evidence suggestthat the belt is structurally and functionally heteroge-neous. First, the architectonic features of the belt regionare not uniform. The parainsular (medial belt) cortex,positioned between the core and insula, has always beenconsidered architectonically distinct from the cortex lat-eral to the core in humans (Brodmann, 1909; Vogt andVogt, 1919; von Economo and Koskinas, 1925; Beck, 1928;Hopf, 1954; Galaburda and Sanides, 1980) and nonhumanprimates (Sanides and Krishnamurti, 1967; Pandya andSanides, 1973; Jones and Burton, 1976; Imig et al., 1977;Galaburda and Pandya, 1983; Morel and Kaas, 1992; Mo-rel et al., 1993; Jones et al., 1995). Pandya and Sanides(1973) distinguished a medial belt of areas (i.e., root areas)from those lateral to the core (i.e., belt areas) in macaquemonkeys, a distinction that was maintained in subsequentrevisions of that scheme (Galaburda and Pandya, 1983;Cipolloni and Pandya, 1989).

Second, the thalamic inputs to the belt differ betweenareas. The main source of input is the dorsal division(MGd) of the medial geniculate complex (MGC), which hasanterior (MGad) and posterior (MGpd) subdivisions. Thebelt areas appear to differ with respect to the balance ofinputs from these divisions, as well as other nuclei in theposterior thalamus (Burton and Jones, 1976; Molinari etal., 1995; Rauschecker et al., 1997; Hackett et al., 1998b;Jones, 2003; de la Mothe et al., 2006). Third, the rostraland caudal belt areas are connected with distinct areas ofprefrontal and posterior parietal cortex (Romanski et al.,1999a,b; Lewis and Van Essen, 2000). Fourth, area CMand several of the lateral belt areas have been distin-guished by auditory response properties, including rever-sals in tonotopic organization, FM rate preferences, andpreferences for spatial and nonspatial stimuli (Merzenichand Brugge, 1973; Imig et al., 1977; Rauschecker et al.,

Abbreviations

A1 auditory area 1 (core)A11 secondary auditory cortexA1L auditory area 1, lateral divisionA1M auditory area 1, medial divisionAAF anterior auditory fieldAChE acetylcholinesteraseAL anterolateral area (belt)A-m anteromedial auditory fieldAS arcuate sulcusBA basal amygdalaBDA biotinylated dextran amine (tracer)Caud caudate nucleusCis circular sulcusCL caudolateral area (belt)Cla claustrumCM caudomedial area (belt)CO cytochrome oxidaseCPB caudal parabelt area (parabelt)CS central sulcusCTB cholera toxin, subunit B (tracer)Ent entorhinal cortexFB Fast Blue (tracer)FE Fluoro Emerald (tracer)FR Fluoro Ruby (tracer)Hip hippocampusIns insulaIPS intraparietal sulcusITG inferior temporal gyrusLA lateral amygdalaLS lateral sulcusLuS lunate sulcus

MF myelinated fibersMGad medial geniculate complex, anterodorsal divisionMGC medial geniculate complexMGd medial geniculate complex, dorsal divisionMGm medial geniculate complex, magnocellular divisionMGpd medial geniculate complex, posterodorsal divisionMGv medial geniculate complex, ventral divisionML middle lateral area (belt)MT middle temporal areaPa posterior auditory areapaAc caudal parakoniocortex areaPAF posterior auditory fieldP-m posteromedial auditory fieldPro proisocortical areaproA prokoniocortex areaPS principal sulcusPut putamenPV parietoventral areaR rostral area (core)Ri retroinsular areaRM rostromedial area (belt)RPB rostral parabelt area (parabelt)RT rostrotemporal area (core)RTL rostrotemporal lateral area (belt)RTM rostrotemporal medial area (belt)S2 somatosensory area 2STG superior temporal gyrusSTS superior temporal sulcusTpt temporal parietotemporal areaVPAF ventroposterior auditory fieldWM white matter

28 L.A. DE LA MOTHE ET AL.

1995; Kosaki et al., 1997; Romanski et al., 1999b; Raus-checker and Tian, 2000, 2004; Recanzone et al., 2000a;Tian et al., 2001; Tian and Rauschecker, 2004; Kajikawaet al., 2005). Fifth, recent studies indicate that neurons inthe caudomedial belt area, CM, are responsive to bothauditory and somatosensory stimulation (Schroeder et al.,2001; Fu et al., 2003), confirming limited observations inearlier studies of somatosensory areas in the caudal lat-eral sulcus (Robinson and Burton, 1980ab–c). Known con-nectivity suggests that the belt areas are likely to differwith respect to multisensory activity. Thus, the areas thatcomprise the belt region appear to be both structurallyand functionally heterogeneous in ways that are graduallybeing revealed but are not yet firmly established.

The most poorly studied areas of the belt region arethose medial to the core, hereafter referred to as themedial belt. At least three areas comprise this region:caudomedial [CM], rostromedial [RM], and rostrotempo-ral medial [RTM] (Fig. 1). The lack of data is partly aconsequence of their location deep within the lateral sul-cus of all primates. Thus, their connections are knownmostly from tracer injections of more accessible auditoryand auditory-related cortical areas. Physiological proper-ties, recordings have mainly focused on CM.

In the present study and its companion (de la Mothe etal., 2006), the cortical and thalamic connections of RM andCM were compared with adjacent core areas, R and A1,following tracer injections into these target areas. The

Fig. 1. Schematic models of macaque (A) and marmoset(B,C) monkey auditory cortex. The lateral sulcus (LS) of the lefthemisphere was graphically opened (cut) to reveal the locations ofauditory cortical areas on its lower bank. The circular sulcus (CiS)was flattened to show the position of the rostromedial (RM) androstrotemporal medial (RTM) areas that occupy its lateral wall. Theupper bank of the LS was partly opened (cut) to show the locations ofthe retroinsular area (Ri) in the fundus, second (S2) and parietoven-tral (PV) somatosensory areas on the upper bank, and insula (Ins).The three areas that comprise the core region of the auditory cortex(dark shading) are located on the lower bank (A1, auditory area 1; R,rostral; RT, rostrotemporal). The core is surrounded by seven or eightareas that belong to the belt region (light shading) (CM, caudomedial;CL, caudolateral; ML, middle lateral; RM, rostromedial; AL, antero-lateral; RTM, rostrotemporal medial; RTL, rostrotemporal lateral).

The proisocortex area (Pro) is a putative addition to the medial belt.The core and lateral belt regions are mostly contained within thelateral sulcus in macaques but extend onto the superior temporalgyrus (STG) in the marmoset. On the surface of the STG are two areasthat make up the parabelt region (medium shading; RPB and CPB,rostral and caudal parabelt). The rostral part of the STG (STGr)extends to the temporal pole. The temporal parietotemporal area(Tpt) occupies the caudal end of the STG and extends onto the supra-temporal plane within the LS. Tonotopic gradients within areas areindicated by H (high frequency) and L (low frequency). Other sulcishown include the arcuate sulcus (AS), central sulcus (CS), intrapa-rietal sulcus (IPS), and superior temporal sulcus (STS). D: Photo-graphic image of the marmoset left hemisphere and schematic show-ing the plane of section (diagonal lines) used in the present study forhistological processing. Scale bar � 10 mm in A,B,D; 5 mm in C.

29MARMOSET AUDITORY CORTEX CONNECTIONS

main goal of these experiments was to refine and extendour working model of the primate auditory cortex, withspecial emphasis on the organization of the medial beltregion. More specifically, the following predictions of themodel were tested: 1) RM and CM are anatomically dis-tinct areas of the auditory cortex and of the medial beltregion; 2) RM and CM receive direct projections from thecore, consistent with their position in the processing hiear-chy; 3) the auditory cortical connections of the medial beltareas are distinct from those of the core; and 4) RM andCM receive inputs from the somatosensory cortex.

A secondary goal of these experiments was to begin todefine the anatomical organization of the marmoset audi-tory cortex and determine how closely it approximatesthat of other primates. Our knowledge of the anatomicalorganization of the auditory cortex in this species is lim-ited to connections of the core (Aitkin et al., 1988; Luethkeet al., 1989), yet marmosets have increasingly become animportant neurophysiological model for the study of theprimate auditory cortex (see Discussion). Establishmentof the anatomical features of the auditory cortex in themarmoset would facilitate ongoing and future studies ofaudition in this vocal primate species and also reveal theextent to which auditory cortex organization may be con-served across taxa. A preliminary report of these findingsappeared in abstract form (de la Mothe et al., 2002).

MATERIALS AND METHODS

Animal subjects

All experimental procedures were conducted in marmo-set monkeys (Callithrix jacchus jacchus) in accordancewith NIH Guidelines for the Use of Laboratory Animalsunder a protocol approved by the Vanderbilt UniversityInstitutional Animal Care and Use Committee. Eightadult marmosets served as animal subjects in the presentstudy. The experimental history of each animal is in-cluded in Table 1.

General surgical procedures

Aseptic techniques were employed during all surgicalprocedures. Animals were premedicated with cefazolin (25mg/kg), dexamethasone (2 mg/kg), cimetidine HCl (5 mg/

kg), and Robinul (A.H. Robins Co., Inc., Richmond, VA)(0.015 mg/kg). Anesthesia was induced by intramuscularinjection of ketamine hydrochloride (10 mg/kg) and thenmaintained by intravenous administration of ketaminehydrochloride (10 mg/kg) supplemented by intramuscularinjections of xylazine (0.4 mg/kg) or by isoflurane inhala-tion (2–3%). Body temperature was kept at 37°C with awater circulating heating pad. Heart rate, expiratory CO2,and O2 saturation were continuously monitored through-out the surgery and used to adjust anesthetic depth. Ox-ygen was delivered passively through an endotrachealtube at a rate of 1 liter/min.

The head was held by hollow ear bars affixed to a ste-reotaxic frame (David Kopf Instruments, Tujunga, CA). Amidline incision was made exposing the skull, followed byretraction of the temporal muscle. A craniotomy was per-formed exposing the left dorsal superior temporal gyrus,lateral fissure, and overlying parietal cortex. After retrac-tion of the dura, warm silicone was applied to the brain toprevent desiccation of the cortex. Photographs of the ex-posed cortical surface were taken for recording the loca-tions of electrode penetrations in relation to blood vesselsand the lateral sulcus.

Retraction of the parietal operculum andneuroanatomical tracer injections

Tracer injections were made into target areas through apulled glass pipette affixed to a 1-�l Hamilton syringe.The pipette was advanced into the cortex under stereomicroscopic observation to a depth of 1,000 �m by using astereotaxic micromanipulator. After manual pressure in-jection of the tracer volume (Table 1), the syringe was heldin place for 10 minutes under continuous observation tomaximize uptake and minimize leakage. Injections of thecore areas (A1, R) were made directly into the lateralsurface of the superior temporal gyrus (STG) after re-moval of the dura (see Fig. 1B,C). Injections of medial belttargets within the lateral fissure were achieved in one oftwo ways. In cases 1, 3, and 7, BDA or CTB were injectedinto RM or CM by passing the syringe through the over-lying parietal cortex. Depth was controlled by stereotaxicmeasurements and verified by recordings made with atungsten microelectrode affixed to the syringe.

In all other cases, access to injection targets within thelateral fissure was achieved by retraction of the banks ofthe lateral fissure, as recently described (Hackett et al.,2005). This was done to gain direct access to target areaswithout tissue resection, as connections with somatosen-sory areas would have been lost. Briefly, after microdis-section of the arachnoid membrane around blood vesselsat the edge of the lateral sulcus, the upper bank wasgently retracted by using a stereotaxic arm and bluntdissection of arachnoid within the sulcus. Once the de-sired opening was achieved, tracer injections were madedirectly into target areas relative to gross anatomicallandmarks and blood vessel patterns.

Auditory stimulation and recordings

For most of the cases included in this report, detailedrecordings were obtained 7 days after tracer injectionsduring a terminal experiment that averaged 24 hours induration. The recording sites were concentrated in A1 andCM using a battery of stimuli, including tones, broad-andnoise, frequency-odulated tones, and marmoset vocaliza-tions. The tonotopic maps derived from these recordings

TABLE 1. Experimental History of Animal Subjects1

Case SexAreas

injected Tracer %Volume

(�l)

1 (01–37) M RM BDA 10 0.4R FR 10 0.3R FE 10 0.4AL/ML2 FB 10 0.25

2 (01–118) M RM BDA 10 0.4R FR 10 0.3CL2 FB 10 0.2

3 (02–17) M CM CTB 1 0.4PP BDA 10 0.4

4 (02–51) M A1 CTB 1 0.4R FR 10 0.3

5 (02–60) M A1 FR 10 0.3CPB2 FB 10 0.2

6 (04–51) M CM CTB 1 0.4AL2 FR 10 0.3

7 (01–89)3 M CM CTB 1 0.48 (04-40)3 M CM BDA 10 0.4

1Areas of tracer injections. Neuroanatomical tracers, aqueous concentration, and vol-ume injected are listed for each tracer. For abbreviations, see list.2Tracer injection was not analyzed for inclusion in the present study.3Cell plot reconstructions were not illustrated for this case.


were marked by electrolytic lesions and aided the recon-structions of architecture and connections, primarily atthe borders of A1 and CM. The physiological results ofthese experiments and methodological details are re-ported elsewhere (Kajikawa et al., 2005; Kajikawa andHackett, 2005). In one case (case 1) the left hemispherewas mapped prior to tracer injections. However, becauseneuronal responses could be abolished or otherwise al-tered by nearby tracer injections, postinjection recordingswere generally confined to the opposite hemisphere.

Perfusion and histology

At the end of the terminal recording experiment, a le-thal dose of pentobarbital was administered intrave-nously. Just after cardiac arrest, the animal was perfusedthrough the heart with cold (4°C) saline, followed by cold(4°C) 4% paraformaldehyde dissolved in 0.1 M phosphatebuffer (pH 7.4). Immediately following perfusion thebrains were removed and photographed. The cerebralhemispheres were separated from the thalamus andbrainstem, blocked, and placed in 30% sucrose for 1–3days. The cerebral hemispheres were cut perpendicular tothe lateral sulcus in the caudal to rostral direction at 40�m, as shown in Figure 1D. In each series of sectionsevery sixth section was processed for the following set ofhistochemical markers: 1) fluorescent microscopy; 2) bio-tinylated dextran amine (BDA) or cholera toxin subunit B(CTB); 3) myelinated fibers (MF; Gallyas, 1979); 4) acetyl-cholinesterase (AChE; Geneser-Jensen and Blackstad,1971); 5) staining for Nissl substance with thionin; 6)cytochrome oxidase (CO; (Wong-Riley, 1979); or 7) parval-bumin (PV) immunohistochemistry.

Architectonic identification of cortical areas

A full architectonic analysis was necessary because acomplete parcellation of the marmoset auditory cortex hasnot been previously published. The architectonic criteriaused to identify areas of auditory cortex in other primateswere used to guide identification of corresponding areas inmarmosets (Imig et al., 1977; Galaburda and Pandya,1983; Luethke et al., 1989; Morel and Kaas, 1992; Morel etal., 1993; Hackett et al., 1998a; Hackett et al., 2001). Thecombined use of multiple markers improved the reliabilityof border identification and was especially useful whenborders between areas were ambiguous in one or anotherstain. Of particular importance in this context was theplane of section. A standard coronal plane was not idealfor visualization of the auditory cortex because cell andfiber columns were then cross-cut at an angle of about 30degrees relative to the orientation of the lateral sulcus. Tominimize these distortions, all brains were cut perpendic-ular to the long axis of the lateral sulcus after removal ofthe thalamus and brainstem, as shown in the inset ofFigure 1D. Columnar orientation also has implications forthe approach angle chosen for microelectrode recordings,as radial orientation varied between cortical areas, andcell columns were often curved.

Digital images were acquired by using a NikonDXM1200F digital camera and Nikon E800S microscope.These images were cropped and adjusted for brightnessand contrast using Adobe (San Jose, CA) Photoshop 7.0software but were otherwise unaltered. Final figures con-taining images and line drawings were made by usingCanvas 8.0 software (Deneba Systems, Miami, FL) andAdobe Illustrator 10.0.

Analysis and reconstruction of connections

The x-y locations of cell somata labeled by retrogradeaxonal transport of each tracer were plotted by using aNeurolucida system (MicroBright Field, Williston VT).Auditory areas were identified in sections stained for thehistochemical markers listed above, according to architec-tonic features described in the Results. For each histo-chemical marker, the borders of individual areas andpatches of anterograde terminal labeling were drawn ontoplots of labeled cells by alignment of blood vessels andcommon architectonic features by using a drawing tubeaffixed to a Zeiss Axioscope. These drawings were used tocreate the reconstructions (e.g., Fig. 9). In most figures,every other section was chosen for illustration. For eachtracer injection, the percent of total labeled cells was de-rived for each by dividing total cell counts for each area bythe total number of cells in the auditory cortex labeled bythat injection. Labeled cells in areas outside of the audi-tory cortex were counted separately and not factored intothe percent total calculations. Values were tabulated sep-arately for ipsilateral and contralateral hemispheres.

Multiple tracers were used in this study to maximizethe information gained from each experiment. Becausesensitivity varies between tracers, greater numbers of la-beled cells were observed in some cases. CTB was the mostsensitive retrograde tracer used and labeled the most cellsper case. BDA and FR typically labeled fewer cells percase, but BDA was the most sensitive anterograde tracer.In the analysis and interpretation of results, it was as-sumed that the proportion of labeled cells found in eacharea was maintained, whereas absolute cell counts dif-fered between tracers. Therefore, the percent of total la-beled cells was used to reflect connection magnitude,rather than total cell number. A second important factoraffected cell counts in the diffusion zone in and aroundinjection sites. Due to high tracer density and tissue dam-age in these zones, cell counts for one or more tracers werelower than normal. The potential error was reflected inthe histograms by using a white bar in the column asso-ciated with the injected area.

RESULTS

Architectonic identification of auditoryareas

The architecture of the marmoset auditory cortex wasillustrated at different levels of magnification over severalfigures to show the relative locations of individual areas(e.g., Figs. 2, 3), and the structural details of each (Figs.4–8). The auditory areas occupied most of the superiortemporal lobe between the fundus of the lateral sulcus(LS) and upper bank of the superior temporal sulcus(STS), as shown in Figures 1 and 2. The greatest source ofvariation between animals stemmed from the depth of theSTS, which ranged from about 2.5 mm (Figs. 2A, 3) to notmore than a mild depression (Fig. 2B, 5) and typicallyreached maximum depth in the caudal half of the tempo-ral lobe. In animals with a prominent STS, the parabeltregion usually occupied the lateral portion of its upperbank, bordering a weakly myelinated zone in the fundus(Fig. 3F,G). In animals with a shallow fissure, the weaklymyelinated zone usually straddled the banks of the STS,shifting the ventral border of the parabelt onto the surfaceof the STG (Fig. 5). Variations in the gross anatomical


configuration of the superior temporal lobe varied betweenanimals, as can be seen in figures thoughout this reportand may relate to the variability previously observed inthis species (Aitkin et al., 1986).

Cytoarchitecture of the core region. The core regionwas easily identified in all cases and typically straddledthe edge of the lateral sulcus along most of its length fromcaudal to rostral, with roughly two-thirds of its area on thesurface of the STG (Fig. 1B,C). Its location could be ap-proximated at low magnification due to dense expressionof cytochrome oxidase in the middle cortical layers (Fig. 2).The cytoarchitecture of the core areas was koniocellular,as typified by a cell-sparse layer V, broad granular layerIV, and high density of small pyramidal cells in layer III(Figs. 5A, 6A–D). No clear differences were noted betweenA1M and A1L other than an increase in cortical thicknesswhere A1M wrapped over the edge of the lower bank. Keydistinctions between A1 and the rostral core areas (R andRT) were a reduction in cortical thickness, most obvious inlayers III and IV, and an increase in layer V cell density.In comparing R and RT, granular cell density in layersIII–IV was slightly greater in R, consistent with greaterfiber density in R, as described below.

Myeloarchitecture of the core region. Myelin den-sity was higher in the core compared with neighboring

belt areas (Figs. 3F–H, 4, 5B).The main exception wasarea CM, which was also myelin-dense across laminae.Within A1 a division between its medial (A1M) and lateral(A1L) halves was consistently noted. Whereas the myeli-nation pattern in A1M was astriate, due to high densityacross layers III–VI, layer IV could be more clearly re-solved in A1L due to a reduction of myelin density in layersIII and Va (Figs. 5B, 6E,F). This pattern has also beenobserved in macaques, chimpanzees, and humans (Pan-dya and Sanides, 1973; Hackett et al., 2001, 1998a) andtherefore appears to be conserved across taxa. We high-light the distinction here because the connections of themedial belt areas varied with respect to the lateral andmedial halves of the core. Whereas myelin density wasgreatest in A1, myelin density decreased rostrally in thecore, reaching a minimum in RT (Fig. 3E–H, 6E–H). Thisdensity shift mainly reflected a reduction of horizontalaxons in layers III–V of R and RT (Fig. 6E–H). Accord-ingly, R and RT had a stronger radial appearance com-pared with A1, where horizontal and radial fibers formeda dense astriate matrix. Compared with the lateral beltareas, however, the inner and outer horizontal striae inlayers IV and Vb were not prominent in any of the coreareas.

Fig. 2. Marmoset monkey left hemisphere. Series of sections fromthe left hemisphere of two animals (A,B) stained for cytochromeoxidase (CO) to show the gross anatomy of the temporal lobe. Thelocation of the core region is outlined by dashed lines, which encom-

pass the band of dense CO staining centered on layer IV of the core.Rostral is at the bottom left of each panel. Solid arrows, lateral sulcus.Open arrows, superior temporal sulcus. Scale bar � 4 mm in B(applies to A,B).


Chemoarchitecture of the core region. Within thecore region, the metabolic enzyme cytochrome oxidase(CO) was densely expressed in a horizontal band involvinglayer IV and the lower half of layer III (Figs. 2, 3I–L, 5D).This band was slightly narrower in R and RT but promi-

nent throughout the core by comparison with the belt andparabelt areas (Fig. 2). One exception to this pattern con-cerned area CM, in which the density of the layer III/IVband was comparable to that of A1M (Figs. 3J, 5D). Ace-tylcholinesterase (AChE) was most densely expressed in

Fig. 3. Architecture of the marmoset auditory cortex at low mag-nification. A–D: Thionin stain for Nissl. E–H: Myelin stain.I–L: Cytochrome oxidase. M–P: Acetylcholinesterase. Columns arearranged from caudal (left) to rostral (right). Dashed black lines markboundaries between areas. Dashed white line denotes subareal border

between the medial (A1M) and lateral (A1L) divisions of A1. Filledarrowhead denotes the lateral sulcus (LS). Open arrowhead marksthe superior temporal sulcus (STS). For abbreviations, see list. Scalebar � 1 mm in P (applies to A–P).


the layer III/IV band and layer Vb, corresponding to themost prominent inner and outer bands of myelinated fi-bers. Whereas AChE expression was slightly more intensein the core, its distribution across all three regions wasrather uniform (Fig. 3M–P). This result was unexpectedbecause dense AChE expression in the layer III/IV bandhas been a key marker of the core in other primates. It wasnot clear whether the present results reflected a speciesdifference in AChE expression or histological incompati-bility. The latter seemed more likely, given the densecoexpression of CO and parvalbumin in the layer III/IVband (Fig. 5C,D). In any event, the observation was reli-able using the same protocol in 12 cases over 4 years withvariable incubation times.

Cytoarchitecture of the lateral belt and parabelt re-

gions. A key cytoarchitectonic feature of the lateral beltand parabelt areas was the prominent radial orientationof generously spaced cell columns, especially in the gran-ular and supragranular layers (Fig. 7A–F). This con-trasted with dense columns of smaller cells in the core.Other features included narrowing of layer IV and theappearance of larger pyramidal cells in lower layer III andlayer Va. Area CL, located caudal to A1 and lateral to CM,had a broad layer III and dense columns of granule cells inlayer IV. This contrasted with the adjacent belt area ML,which had a relatively narrow layer III and broader co-

lumnar spacing in layer V. The cytoarchitecture of ALresembled ML, but columnar spacing was slightly moregenerous in layer III and overall cortical thickness wasreduced. This trend continued rostrally into RTL, wherelayers IV and VI became less distinct. The cytoarchitec-tonic border between the lateral belt and parabelt regionswas generally not robust, except for the following features.Layer III was typically broader in the parabelt and char-acterized by prominent radial alignment of granular andpyramidal cells. The orderly spacing and orientation ofcells in these columns extended across most of the corticalmantle from layers VI through III, giving the CPB andRPB a striking radial appearance (Fig. 7E,F).

Myeloarchitecture of the lateral belt and parabelt

regions. Compared with the core, myelin density in thelateral belt and parabelt regions was relatively weak inlayers III and Va, revealing a prominent band of horizon-tal fibers in layer IV and a weaker band in layer Vb (Figs.3E–H, 4, 7G–L). Like the core and medial belt regions,myelin density decreased from caudal to rostral in boththe lateral belt (Fig. 7G–J) and parabelt (Fig. 7K,L). InCL, the outer horizontal stria in layer IV was prominentagainst dense radial fibers that extended well into layerIII. A secondary horizontal band in the Vb was also ap-parent but was less prominent due to a dense network offibers in the infragranular layers. In ML, the density ofmyelinated fibers was reduced overall, but the layer IVband remained prominent. In AL, horizontal fiber densitywas greatly reduced compared with CL and ML, alongwith greater spacing between radial fascicles. This reduc-tion continued into RTL, which had very weak fiber orga-nization in layer III. In the parabelt, the radial appear-ance noted in the cytoarchitecture was matched by thestrong radial orientation of myelinated fibers that ex-tended from layer VI through layer III and into layer II(Fig. 7K,L). Horizontal fibers formed clear bands in layerIV of the CPB, compared with weak horizontal organiza-tion in RPB.

Chemoarchitecture of the lateral belt and parabelt

regions. The main finding in the lateral belt and para-belt areas was a dramatic reduction in the expression ofCO and parvalbumin in the layer III/IV band comparedwith the adjacent core areas (Figs. 2, 3–J, 5C,D). Theexpression of CO in this band diminished rostrally in bothregions. Parvalbumin expression was weaker in the para-belt than the lateral belt (Fig. 5). Otherwise, there were noclear differences between or within the lateral belt andparabelt regions with respect to these markers.

Cytoarchitecture of the medial belt region. The cy-toarchitecture of the medial belt region and adjoiningfields varied reliably between areas. The most distinctivearea was CM, which bordered A1 medially and caudome-dially, occupying most of the superior temporal plane cau-dal to A1 (Fig. 1B,C). CM was characterized by a popula-tion of medium-sized pyramidal cells with poor radialalignment concentrated in the lower half of layer III (Fig.8A). Layer III was broad compared with layers IV throughVI, but cortical thickness was reduced relative to A1M.Layer IV was narrow compared with A1 and was popu-lated by stacks of granule cells arranged in broad columns.Layer V was densely populated by small to medium-sizedcells. The cytoarchitecture of the rostral medial belt areaswas more like the lateral belt areas than CM (Fig. 8B,C).Area RM was located medial to area R and rostral to thenarrow extension of CM medial to A1, whereas RTM was

Fig. 4. Myeloarchitecture of marmoset auditory cortex. Myelinstain through A1(case 8) to show medial (A1M) and lateral (A1L)subdivisions of A1 in relation to CM and ML. Dense myelinationacross laminae in A1M is reduced in layers III and Va of A1L. Corticallayers are indicated by Roman numerals I–VI. For abbreviations, seelist. Scale bar � 500 �m.


medial to RT (Fig. 1C,D). Layer IV was reduced in widthcompared with the neighboring core areas, highlighted bythin strings of granule cells arranged in broadly spacedcolumns. Layer III was populated by orderly columns ofsmall to medium-sized pyramidal cells that contrastedsharply with their disorganized counterparts in CM.

The broader columnar spacing in RM and RTM was akey feature in their distinction from R and RT. Cell spac-ing in RTM was slightly greater than that in RM. Medialto CM was the retroinsular area (Ri), which occupied thefundus of the lateral sulcus. It extended onto the lowerbank of the medial LS for about 0.5 mm and bordered the

Fig. 5. Architecture of marmoset auditory cortex through the ros-tral A1. A: Thionin stain for Nissl. B: Myelin stain. C: Parvalbuminimmunohistochemistry. D: Cytochrome oxidase stain. Note correspon-dence of dense myelin, parvalbumin, and cytochrome oxidase in thelayer III/IV band of A1. Layer III/IV expression was moderately dense

in CM and less dense in ML and CPB. Parvalbumin expression wasweakest in CPB. Solid arrowheads mark lateral sulcus. Open arrow-heads mark the location of the STS, which was very shallow in thisanimal (compare with Fig. 3). For abbreviations, see list. Scale bar �1 mm in D (applies to A–D).


Figure 6

second somatosensory area, S2, on the upper bank (Fig.1C,D, 3A,B). In the mapping experiments involving thissame group of animals, we often observed robust re-sponses to cutaneous somatosensory and vibratory(Pacinian-like) stimulation in the Ri but no clear auditoryresponses (Kajikawa et al., 2005). The cytoarchitecture ofRi was characterized by a dramatic reduction in corticalwidth reflecting compression across laminae (Fig. 8E).Columnar spacing in Ri was greater than in CM. Layer IIIcontained a uniform distribution of pyramidal cells, and anarrow layer IV contained broad stacks of granule cells.Layer V was somewhat cell-sparse, and layer VI containednumerous cells with horizontally oriented dendrites. RTMmay wrap around the rostral end of RT to join RTL on thelateral surface of the STG, but at this level the architec-ture of the temporal polar region was quite uniform, andso this observation remains tentative.

Medial to RM and RTM was another area that sepa-rated these areas from the insular cortex (Fig 1D, 3C,D).We adopted the name “Pro,” because it appeared to corre-spond to a similar field identified in macaque monkeys(Galaburda and Pandya, 1983). Compared with RM andRTM, cell density was reduced in Pro overall. Layers IVand V became very narrow in their transition toward theinsula (Fig. 8D). Pro was consistently labeled by injectionsof core and medial belt areas in this study and may there-fore comprise part of the auditory cortex.

Myeloarchitecture of the medial belt region. Themyeloarchitecture of CM complemented its cytoarchitec-ture. Thick radial fascicles ran between cell columns,crossed by a dense plexus of horizontal fibers from layer VIto the middle of layer III (Figs. 4, 8F). The density of thisnearly astriate pattern was only slightly reduced com-pared with A1. However, reduced density in the upperpart of layer III, reduced cortical thickness, and coarseappearance of the broadly spaced radial fibers allowed forreliable identification of CM. Medial to CM, area Ri wasdominated by horizontal fibers, especially in layers IV andVI (Fig. 8J). These bands were crossed by broadly spacedthick radial fascicles that extended into layer III. Rostralto CM, myelin density in the medial belt was greatlydiminished (Fig. 3E–H). In RM and RTM, radial fiberswere broadly spaced, and horizontal fiber organizationwas greatly reduced compared with CM and R or RT (Fig.8G,H). Myelin density was weakest in RTM due to broadcolumnar spacing and sparse horizontal fibers. These fea-tures were even weaker in Pro (Figs. 3G,H, 8I).

Chemoarchitecture of the medial belt region. Asbriefly noted above, CO expression was comparable in thelayer III/IV band of both A1M and CM (Figs. 2, 3I,J, 5D);thus reliable identification of CM depended mainly onanalysis of the cyto- and myeloarchitecture. Parvalbuminexpression was also moderately dense in CM (Fig. 5C).Rostrally, in RM and RTM, CO expression was relativelyweak, comparable to that of the lateral belt areas. AChEand parvalbumin expression in the layer III/IV band was

also similar to the lateral belt and parabelt areas. Bycomparison with RM and RTM, therefore, CM was ratherprimary-like due to dense myelination and expression ofCO and parvalbumin.

Ipsilateral connections of CM

In case 3 the CTB injection was made into the part ofCM that caps A1 caudally (Fig. 9). In the most caudalsections (#165–177), labeled cells and terminals were con-centrated in the supragranular and infragranular layersof CM, whereas in Ri, there was no clear evidence ofanterograde projections to any layer. Labeling in CL wasconcentrated in layer II. The distribution of labeling didnot extend beyond Ri dorsally into S2 but stopped cleanlyat the border. With the emergence of A1 (#189–201), la-beled cells and terminals spanned supragranular and in-fragranular layers of Ri, CM, ML, and A1. Further ros-trally (#213–255), connections with A1 weakenedsignificantly, especially with A1M, whereas connectionswith CM, Ri, ML, and CPB remained strong. In ML andCPB, labeling was concentrated in columnar patcheswhere strong anterograde projections overlapped clustersof retrogradely labeled somata. A column of terminals andcells also overlapped in A1L over part of this range. Theabsence of connections with A1M formed a prominent gapbetween the lateral and medial belt regions, as viewed inthe coronal plane (Fig. 10).

Variants of this pattern characterized all medial beltinjections in this study, as described below. Near the bor-der of A1 and R (#255), labeling in ML and CPB weakenedand was concentrated in the infragranular layers. Labeledcells were dense in Ri, whereas only scattered cells re-mained in A1 and CM. Rostrally, at the level of R and RT(#267–327), the core and lateral belt areas were mostlydevoid of labeled elements. Labeling in RM did not persistbeyond A1/R border region. Labeled cells were consis-tently observed in Pro over this range and became increas-ingly infragranular. In RPB, labeled cells were also con-fined to the infragranular layers over this entire range.Outside of the auditory cortex, cells were found ventral toRPB in the STS. Labeled cells in the lower layers of theentorrhinal cortex (Fig. 11A) were distributed in a contin-uous band in sections rostral to the A1/R border (#267–327). Additional connections were revealed with the pari-etal cortex caudal to the lateral sulcus. Labeled cells andterminals from a BDA injection not illustrated in thereconstructions of this case were concentrated in supra-granular CM, as well as in Ri and CL (Fig. 11B). Label wasalso found distributed lightly throughout the posteriorbelt and parabelt areas in this case, suggesting that thecaudal auditory cortex has significant connections withposterior parietal areas (Lewis and Van Essen, 2000).

In cases 6, 7, and 8, the injection was made along thenarrow extension of CM that borders A1 medially. In case6, the injection was made directly into medial CM justcaudal to its border with RM (#128) whereas the upperbank was retracted (Fig. 12). The injection was fairly wellconfined to CM on the lower bank between Ri and A1M.Unfortunately, a small amount of tracer appears to havediffused into the upper bank after the sulcus was allowedto close, leading to some labeling in the part of S2 opposingthe injection site. Therefore, we cannot be certain whetherthe labeling observed in S2 in this case was due to the CMinjection, diffusion across the sulcus, or both. In the mostcaudal sections (#212–200) labeled cells and terminals

Fig. 6. Cytoarchitecture and myeloarchitecture of marmoset audi-tory cortex, core region. Top, thionin stain for Nissl. Bottom, corre-sponding section, myelin stain. A,E: Area A1M. B,F: Area A1L. C,G:Area R. D,H: Area RT. Cortical layers are indicated by Roman nu-merals I–VI. wm, white matter. Scale bar � 250 �m in H (applies toA–H).


Figure 7

were concentrated in CL and CM. Distinct bands of denseanterograde labeling overlapped the labeled cells in layersII, III, V, and VI but avoided layer IV and the lower part

of III. Cells were also found in Ri and dorsally onto thesurface of the posterior parietal cortex. CL was borderedventrally by a region characterized by dense astriate my-elination, possibly corresponding to the middle temporalarea (MT). Although some labeling extended into thisregion, it did not appear to coincide with CPB.

At the level of A1 (#188–152), dense labeling was foundin supragranular and infragranular layers of A1, CM, andML. Labeled cells in Ri and CPB were less numerouscaudally but increased in numbers rostrally over thisrange closer to the injection site. In this range (#152–122),

Fig. 8. Cytoarchitecture and myeloarchitecture of marmoset auditory cortex, medial belt region andadjoining areas. Top, thionin stain for Nissl. Bottom, corresponding section, myelin stain. A,F: Area CM.B,G: Area RM. C,H: Area RTM. D,I: Area Pro. E,J: Aea Ri. Cortical layers are indicated by Romannumerals I–VI. wm, white matter. Scale bar � 250 �m in J (applies to A–J).

Fig. 7. Cytoarchitecture and myeloarchitecture of marmoset audi-tory cortex, lateral belt and parabelt regions. Top, thionin stain forNissl. Bottom, corresponding section, myelin stain. A,G: Area CL.B,H: Area ML. C,I: Area AL. D,J: Area RTL. E,K: Area CPB. F,L:Area RPB. Cortical layers are indicated by Roman numerals I–VI.wm, white matter. Scale bar � 250 �m in L (applies to A–L).


Fig. 9. Ipsilateral cortical connections of area CM, case 3. Series of serial sections are arranged fromcaudal (upper left) to rostral (lower right). Labeled cells (filled circles) and terminals (shading) are drawnonto each section, showing borders between areas identified by architectonic criteria. Inset: schematic ofmarmoset auditory cortex showing location of CTB injection in caudal CM. For abbreviations, see list.


the greatest concentration of labeled cells and terminalswas centered on CM and extended into Ri and A1M, withsecondary labeling in the CPB and ventrally in the STS.By contrast, labeling in ML and especially A1L weregreatly diminished, forming a notable gap between A1Mand CPB (Fig. 5). With the emergence of R (#110–74),dense labeling continued into RM and extended mediallyinto Pro. Labeling in R was moderate medially, but thelateral portion of R was nearly devoid of label. Althoughthis is reminiscent of the division of A1M and A1L, anarchitectonic division of R was not obvious in most sec-tions. In AL, labeled cells and terminals overlapped insupragranular and infragranular layers for most of its

range, forming a gap comprising the lateral part of R.Labeling in RPB was mostly infragranular, as was label-ing in the STS.

Rostrally, labeling rapidly diminished in RTM and la-beled cells in Pro were mostly infragranular. Labeling wasmoderate in RTL and RPB, favoring the lower layers.Weak labeling in RT (#62–50) maintained the gap be-tween medial and lateral belts. Rostral to known areas ofthe auditory cortex (not illustrated), scattered labeledcells persisted in the lower layers along the lateral STGr.Also outside of the auditory cortex, labeled cells werebroadly distributed in the lower layers of the entorhinalcortex (#176–98), as in case 3 and with all CM injections.

Figure 9 (Continued)


In cases 7 and 8, injections of CM medial to A1 (re-constructions not illustrated) produced similar patternsto case 6. In case 7, the pattern of labeling was almostidentical to case 6 (Fig. 10C). However, the injectionwas made through the overlying parietal cortex andlabeled even larger numbers of cells in S2 and otherposterior parietal areas on the lateral surface of thebrain. In case 8, BDA was injected directly into CMafter sulcus retraction. The injection was confined to thelower bank of the lateral sulcus with no diffusion intothe upper bank. Despite a shortened survival time (3days) in this case, labeled cells and terminals weredistributed broadly throughout the auditory cortex inpatterns that also matched case 6 (Fig. 10D). By con-trast, there were almost no labeled cells beyond theborders of Ri or S2 in this case, suggesting that thelabeling in S2 in cases 6 and 7 was the result of diffusionof CTB into the upper bank.

Interhemispheric connections of CM

In case 3 the CTB injection into caudal CM was mir-rored in the opposite hemisphere in the most caudal sec-tions (#195–171), where a dense focus of label was cen-tered in layer III of CM, with secondary labeling in theadjacent areas, Ri and CL (Fig. 13). With the emergence ofA1 (#171–159) rostrally, labeled cells were found in A1and ML, but remained concentrated in CM and Ri. Fur-ther rostral (#147–117), labeled cells were distributedwidely across A1L, ML, and CPB, but this number dropped

Fig. 10. Columns of labeled cells and terminals overlap in themedial and lateral belt regions after BDA injections of RM and CM.Labeling in the intervening core region is absent or greatly attenu-ated, forming a continuous gap that spans the rostrocaudal axis ofauditory cortex. A: CTB labeling in RM and AL rostral to the RMinjection. Note dense focus of anterograde label in layer IV of RM.Case 2, #263. B: CTB labeling in the RM and AL just rostral to RMinjection. Dense anterograde projections to layer IV are visible in RMand AL. Case 1, #69. C: CTB labeling in CM, Ri, and ML after largeCM injection. Label in the A1 and CPB is mostly infragranular. Case7, #129. D: BDA labeling in CM and CPB after CM injection. Weaklabeling is visible in the A1L. Case 8, #176. For abbreviations, see list.Scale bar � 1 mm in D (applies to A–D).

Fig. 11. CM connections outside auditory cortex. A: CTB-labeledcells (between arrows) in the lower layers of entorhinal cortex fromCM injection. Cells were distributed in this region along most of therostrocaudal axis of the auditory cortex after CM injections (see Figs.9, 12). Injections of RM did not label cells in the entorhinal cortex.B: Patches of BDA-labeled cells and terminals in Ri, CM, and CL(arrows) after an injection into the posterior parietal cortex just cau-dal to the end of the lateral sulcus (LS). The concentration of cells andterminals in layer III of CM coincide with the CTB injection of CM(see case 3, #177). Asterisk marks the pipette track made by that CTBinjection. Dashed lines, borders between areas. For abbreviations, seelist. Scale bar � 1 mm in A; 500 �m in B.


to almost zero in Ri and CM. Labeling stopped abruptlyrostral to the A1/R border (#105).

In case 6 only a few labeled cells were found in con-tralateral CM, CL, and Ri of the most caudal sections(#206–164), reflecting the more rostral placement of theinjection within CM in the opposite hemisphere (Fig. 14).Instead, overlapping cells and terminals were concen-trated more laterally in layer III of CM and A1. Nearer thehomotopic location of the injection site at section (#128),labeled cells became more numerous in CM and Ri butdiminished in A1 (#152–140). Rostrally, at the border ofA1 and R, labeling continued to dominate in CM, extend-ing into RM, with additional cells appearing in Pro (#116–104). Thereafter, the number of labeled cells in RM ta-pered off until no more cells were found (#92–80). Anexample of overlapping cell and terminal labeling in thecontralateral hemisphere is shown in Figure 15, associ-ated with the CTB injection in this case. Anterogradebanding extended from layer III into layer I across somecolumns in CM. A cell-sparse band of terminal labelingdistinguished layer V of CM in this section. Terminallabeling in A1 and Ri was relatively light.

Summary of CM connections

Injections into CM caudal to A1 and CM medial to A1revealed similar patterns of connections overall, withsome interesting differences reflecting topographic varia-tions (Fig. 16, left). For all locations in CM, the strongestconnections involved the caudal areas of the auditory cor-tex and surrounding areas, including A1, ML, CL, CPB,Ri, and other portions of CM. Connections with rostralareas of the auditory cortex were topographic, dependingon the location of CM injection, and favored projections toCM from cells in the infragranular layers. Caudal CM hadrelatively weak connections with RM and Pro, mostly in-fragranular projections from RPB, and almost no connec-tions with the most rostral areas at the level of RT, in-cluding RT, RTM, and RTL. Rostral CM had strongreciprocal connections with RM, Pro, R, and AL. Connec-tions with RT were minimal, and connections with RTM,RTL, and RPB tended to become infragranular-dominantwith distance from the injection site. Both rostral andcaudal sites in CM exhibited continuous reciprocal connec-tions within CM that extended into the RM, althoughinjections of rostral CM resulted in a greater extension oflabeling into RM and RTM. This was consistent with weakconnections with caudal CM observed after injections ofRM (see below).

All locations in CM had strong reciprocal connectionswith parts of A1, but an interesting topographic patternwas revealed involving the core. Caudal CM had broadconnections with A1L and A1M caudally, weak connectionswith rostral A1M, and no connections with R or RT. Ros-tral CM had strong connections only with the medialhalves of A1, R, and RT near the injection and rostrally,but dense broad connections with A1M and A1L caudally.Thus, the contrast between continuous label in the lateraland medial belt coupled with the absence of label in partsof the core formed elongated gaps between the lateral andmedial belt regions along the rostrocaudal axis of theauditory cortex. A comparable pattern was also observedafter injections of RM (see below).

Beyond the auditory cortex, connections were consis-tently found with the cortex in the lower layers of the STSand the entorhinal cortex. There was no clear evidence of

an anterograde projection from CM to either of these re-gions. Connections with the somatosensory cortex wereclearly established with Ri, but connections with S2 andother somatosensory areas remain uncertain. Finally,moderate numbers of labeled cells were consistently ob-served in the posterior parietal cortex after CM injections,indicating a reliable connection with areas in that region.

The interhemispheric connections of CM (Fig. 16, right)favored contralateral CM, and additional strong connec-tions with A1 and Ri. The connections with CM and A1were largely reciprocal. Other connections included inputsfrom RM, Pro, and a weak projection from ML. Connec-tions with all areas were concentrated in layer III.

Ipsilateral connections of RM

In case 1 tracer injections were made into RM, R, andAL after multiunit recordings were used to identify thereversal in the tonotopic gradient between areas A1 and R(Fig. 17A). The BDA injection was made into RM by avertical penetration that passed through the overlyingparietal cortex 0.5 mm medial to the edge of the lateralfissure at AP �10 mm. This was accompanied by cell andterminal labeling of the somatosensory cortex not ob-served in case 2, in which the injection was made directlyinto RM after sulcus retraction. The injection site waspoorly responsive to pure tones but responded well to1/3-octave bandpass noise, with a best center frequency of6.9 kHz.

In the most caudal sections near the caudal edge of A1(#219–209), labeled cells were scattered in A1, ML, CL,CM, and Ri. Rostrally, as A1 emerged (#194), a densefocus of labeled cells and terminals labeled the lateraldivision of A1 (A1L) in the infragranular and supragranu-lar layers, whereas the medial part of A1 (A1M) containeda few supragranular cells. This patch was separated froma secondary patch of label in the supragranular CPB byML, which was mostly devoid of label (Fig. 5). In a seriesof sections rostral to this zone (#184–159), the concentra-tion of labeled cells and terminals in CPB was slightlygreater in the infragranular layers. In A1M at this level,labeled terminals were both infragranular and supra-granular, and labeled cells were more numerous in layerIII. CM also contained a few labeled cells. In section #144RM injection spanned all cortical layers, stopping justshort of the white matter in lateral RM. The diffusion zoneappeared to include the medial edge of R, although label-ing in R did not extend significantly beyond this point. Theappearance of labeled cells in the ventral medial genicu-late (MGv) of this case suggested that there was someinvolvement of medial R (de la Mothe et al, 2006). Denselabeling continued in RM for about 2 mm to its rostralterminus near section #89, where lighter connections con-tinued in RTM. By contrast, R and parts of AL were nearlydevoid of label across this range, forming an elongated gapbetween the medial and lateral belt regions (Fig. 10).

The heaviest connections within RM included the su-pragranular and infragranular layers, with a lighter bandof anterograde projections in layer IV. Labeled cells wereconsistently found medial to RM and RTM in Pro, betweenthe medial belt and insula, where labeled cells were usu-ally infragranular. Along part of this range (#144–119), adense patch of labeled cells and terminals was located inthe supragranular and infragranular AL. Anterogradeterminal labeling in AL was densest in a cell-free bandcorresponding to layer IV. As this patch diminished


Fig. 12. Ipsilateral cortical connections of area CM, case 6. Seriesof serial sections are arranged from caudal (upper left) to rostral(lower right). Labeled cells (filled circles) and terminals (shading) aredrawn onto each section, showing borders between areas identified by

architectonic criteria. Inset: Schematic of marmoset auditory cortexshowing location of CTB injection in rostral CM. For abbreviations,see list.

(#119), a different patch of labeled cells and terminalsbegan to emerge in RPB, which became quite dense overseveral sections (#119–99). Anterograde projections toRPB favored the supragranular layers, although the ter-minal band in layer IV remained prominent.

A focal projection to the ventral caudate nucleus (Fig.18C,D) was also present in several of these sections(#129–109). In the most rostral sections (#69–44), thestrongest labeling was within RTM. Whereas terminalsand cells were in both supragranular and infragranularlayers, layer IV received the densest terminal projection.Labeled cells were more numerous below layer IV. RT hadalmost no labeling, maintaining the gap between the me-dial and lateral belts observed caudally. In presumptiveRTL, a dense focus of label was contained across corticallayers, with a dense terminal band in layer IV. A separatepatch of label with similar connections was located inRPB. A focalized projection to the lateral nucleus of theamygdala (Fig. 18A,B) was present in a few rostral sec-tions (#89–79).

In case 2 BDA was injected directly into RM after re-traction of the lateral fissure (Fig. 19). The injection ex-tended across all cortical layers and labeled cells in thecore, belt, and parabelt regions. In this same case, otherareas were injected with different tracers (R, CL/A1), asreflected in the reconstructions. In the most caudal section(#131), a patch of labeled cells and terminals overlappedin A1L at its border with ML, similar to case 1. Rostrally,(#143–155), a patch of label was found in A1M, withweaker extension into CM. Labeling in CM increased ros-trally (#161) as the border with RM neared. Just caudal tothe border of A1 and R (#173), labeled cells and terminalswere concentrated in CM. Layer III contained most of thelabeled cells, but anterograde labeling was heavy in bothsupragranular and infragranular layers. Weaker cell andterminal labeling extended into A1M. Lighter labeling wascontained within the CPB at this level, as in case 1.Further rostrally (#185–197), labeling was strong in RM,the medial portion of R, and RPB. These two zones wereseparated by sparse labeling in the intervening cortex,

corresponding to AL and the lateral portion of R, where adifferent tracer injection was located.

The patch of intense label in RPB extended across thecortical layers for about 1.5 mm along the rostral-caudalaxis of the sulcus. RM injection was located rostral to thiszone (#215–227). The heaviest label remained in RM butalso included the medial portion of R and Pro, locatedbetween RM and the insula. The injection may have in-volved the medial edge of R, as some labeled cells werefound in the MGv near the MGpd border (de la Mothe etal., 2006). A patch of anterograde label extended across afew sections in the ventral caudate nucleus at this level(#197–215), as in case 1 (Fig. 18). Nearing the estimatedborder of R and RT (#257–251), labeling in RM and medialR continued to be strong. However, in contrast to sectionscaudal to the injection site, anterograde labeling in rostralRM was concentrated in layer IV, whereas it was sparse inlayer IV of R, reflecting a strong feedforward projection tothis area. Another patch of labeled cells and mostly ter-minals appeared in RPB and AL/RTL at this level, al-though labeling was less dense than in more caudal sec-tions. At this same level, anterogradely labeled terminalswere also found in the lateral division of the amygdala(Fig. 18). In the most rostral sections (#269–281) contain-ing the core area, RT, labeled terminals continued to fillthe medial belt (RTM), with concentrations in layer IVand lower V/VI. Weaker anterograde projections could beseen under high magnification in the lateral belt andparabelt as RT was displaced by the merging of RTL andRTM.

Interhemispheric connections of RM

In case 1, cells labeled by the RM injection were concen-trated in layer III of the contralateral medial belt and coreregions, as in case 2, but additional numbers of cells werefound in the core due to encroachment of the injectionupon R at its medial border with RM (Fig. 20). In the mostcaudal sections containing A1 (#184–154), labeled cellswere located in A1 and rostral CM. Cells were infragranu-lar (#184) or distributed among infragranular and supra-



granular layers. In more rostral sections containing A1(#144–109), cells were nearly all supragranular. Ros-trally, in sections containing R (#99–54), labeled cellsoccupied layer III of R and RM.

In case 2, BDA-labeled cells from the RM injection wereconcentrated in layer III of the contralateral medial beltand core regions (Fig. 21). In caudal sections containingA1 (#107–167), labeled cells were confined to rostral CMand A1m. Rostral to A1, the same pattern continued withcells limited to layer III of RM and R. Two cells were foundin RPB of one section (#203).

In both cases, there was no evidence of anterogradeterminal labeling in the contralateral hemisphere, despitedense projections to the thalamus (de la Mothe et al.,2006), and throughout the ipsilateral hemisphere. It can-not be determined whether this was due to poor inter-hemispheric transport of BDA, or whether this reflected aunique property of RM.

Summary of RM connections

Although there were some minor variations between thetwo cases in which injections were made into RM, theoverall patterns were the same (Fig. 22, left). Dense re-ciprocal interconnections extended continuously along themedial belt from CM through RTM, although the connec-tion with CM became weaker with distance from theRM/CM border. Connections with the lateral belt andparabelt were reciprocal and dense, but focalized, reflect-ing patchy rather than continuous connections. Parabeltconnections included patches in both CPB and RPB. Con-nections with the rostral lateral belt (AL, RTL) were densebut very sparse with the caudal belt (ML, CL), indicatingtopographic specificity favoring the rostral belt areas. Theprojections from RM to the belt and parabelt areas rostralto the injection site typically included strong terminallabeling of layer IV, whereas projections to caudal areas

Fig. 13. Interhemispheric cortical connections of area CM, case 3.Series of serial sections are arranged from caudal (upper right) torostral (lower left). Labeled cells (filled circles) and terminals (shad-ing) are drawn onto each section, showing borders between areas

identified by architectonic criteria. Inset: Schematic of marmosetauditory cortex showing location of CTB injection in caudal CM in thecontralateral hemisphere. For abbreviations, see list.


were usually weak in layer IV. Connections with the corewere characterized by a continuous gap in which RT andthe lateral portions of both A1 and R were nearly devoid oflabeled cells and terminals. Only the medial parts of theseareas were labeled, possibly reflecting slight encroach-ment of RM injections into medial R. On the other hand,significant involvement of R should have produced morewidespread labeling within the core. In any event, therewas a continuous gap in the connections between themedial and lateral belts extending from one end of the coreto the other, suggesting that RM has restricted connec-tions with the core.

Compared with CM, RM had fewer connections beyondthe auditory cortex. In both cases focal projections tar-

geted the lateral amygdala and ventral caudate nuclei. Incontrast with CM, RM does not appear to have significantconnections with the somatosensory cortex or the STS.

Interhemispheric connections of RM (Fig. 22, right)strongly favored contralateral RM, with secondary inputsfrom R, A1M, and CM. Labeled cells in all areas wereconcentrated in layer III. As noted above, there was noevidence of anterograde projections to contralateral RM ineither case, as predicted from the connections of CM. Thisobservation requires further investigation.

Ipsilateral connections of A1

In case 4, injections were made into the core areas A1and R (Fig. 23). The CTB injection of A1 extended across

Fig. 14. Interhemispheric cortical connections of rostral area CM,case 6. Series of serial sections are arranged from caudal (upper right)to rostral (lower left). Labeled cells (filled circles) and terminals (shad-ing) are drawn onto each section, showing borders between areas

identified by architectonic criteria. Inset: Schematic of marmosetauditory cortex showing location of CTB injection in rostral CM in thecontralateral hemisphere. For abbreviations, see list.


all layers of the cortex (#260) and mainly labeled cells andterminals in the core and belt areas ipsilaterally. A smallnumber of cells were labeled in layer V of presumptiveparabelt areas, reflecting a weak feedback projection fromthat region. Caudal to A1 and CM, labeled cells werelocated in the infragranular layers of temporoparietal ar-eas that were architectonically distinct from CM and Ri(#196). The lateral area may correspond to the gyral por-tion of Tpt observed in the macaque monkey (Galaburdaand Pandya, 1983). Just rostral to these fields, labeledcells and terminals were concentrated in the infragranu-lar and supragranular layers of CM (#220–244), reflectinga strong reciprocal connection between A1 and CM. Byinspecting consecutive sections (#228–276), it can be seenthat the density of connections between A1 and CM waspatchy, reflecting topographic variation. This feature isillustrated in Figure 24.

Within A1, labeled cells and terminals were continu-ously distributed along its entire rostrocaudal axis (#228–276), extending rostrally into R and RT of the core (#284–340). Rostral and caudal to the injection site (#260),reciprocal connections were revealed with supragranularand infragranular layers of the lateral belt areas. Densitydecreased with distance from the injection site, but con-nections were maintained along the entire rostrocaudalaxis of the belt. Connections with the rostral and caudaldivisions of the parabelt were characterized by the label-ing of infragranular cells, but not terminals, indicating afeedback projection from the parabelt to A1. In the rostralmedial belt region, A1 had dense reciprocal connectionswith RM (#284–292) that extended across cortical layers.In contrast to the continuous band of connections observedfor most other areas, the connection with RM spanned arelatively small rostrocaudal distance, indicative of focaltopography. Labeled cells and terminals were continuousin Pro over a broad range (#284–340) and tended to dom-inate in the infragranular layers, whereas RM had fewlabeled cells in its most rostral sections. In the areas

medial to the medial belt region, including Ri, and theinsula, labeled cells, but not terminals, were consistentlyfound and these were generally concentrated in the infra-granular layers. Like the projections to A1 from the para-belt region, this pattern reflects feedback to A1 from, butnot to, these areas. Given the sensitivity of CTB and denseparallel projections to core and belt areas, it is unlikelythat any significant connection of A1 was not representedin this case.

In case 5 the FR injection was placed into A1 (Fig. 25).The most caudal sections (#188–220) included the FRinjection into lateral A1. Labeled cells at this level weremostly contained within A1, with scattered cells in MLand CM. Moving rostrally from the injection (#228–252),labeled cells were most dense in layers III and V of CM.Labeling in the lateral A1 was less dense and favoredlayer III. In ML, labeled cells were more numerous in theinfragranular layers. In CPB, labeled cells were sparseand mostly infragranular. Rostrally, at the level of R(#260–308), labeled cells in R were almost exclusivelyinfragranular, whereas sparse labeling in AL was slightlymore infragranular. Labeled cells in RM were in infra-granular and supragranular layers.

Ipsilateral connections of R

In case 1, two injections were placed within R, justrostral to the A1/R border (Fig. 17). The FR injection wasmade about 0.5 mm from the edge of the lateral sulcus.The FE injection was about 1.5 mm from the sulcus nearAL but also included some of the white matter below thesite. In the most caudal sections containing labeled cells(#194–184), the greatest number of cells was found inA1M. Cells from both injections and double-labeled cellsoverlapped in this zone, especially in the supragranularlayers. Labeled cells in A1L were fewer in number. A fewcells were found in ML and CM. Rostrally, up to the borderwith R (#169–159), the concentration of cells continued tofavor A1 over other areas but was more balanced betweenA1M and A1L. CM contained a moderate number of labeledcells from both injections and double-labeled cells, as well.A small number of cells were located in ML and scatteredin CPB. The strongest connections were observed rostralto A1 (#144–119), where most of the labeled cells werefound within R, followed by RM, and then AL. Both single-and double-labeled cells were found in these areas. Fur-ther rostral (#109–59), few cells were found in R. Instead,labeled cells were most numerous in RM, and these weremostly FR, rather than FE. This is consistent with thestronger connections between RM and medial part of Robserved after injections of RM (see above). Finally, in themost rostral sections containing RT, only scattered cellswere found in either RT or RM.

In case 2 the FR injection was made into R laterally, atits border with AL (Fig. 19). The diffusion zone appearedto encroach slightly upon AL, although labeled cells in thethalamus were restricted to a narrow band in the MGv,reflecting a clean core injection. In the most caudal sec-tions (#143–161), labeled cells were concentrated in A1L,and just a few cells in were found in ML. Rostral to theborder of A1 and R (#173–215), labeled cells were concen-trated in both R and AL, especially around the injectionsite. A few cells were found in supragranular RPB, possi-bly reflecting slight involvement of AL by the injection.Further rostral (#227–251), cells were found only in R andRT. From rostral to caudal, only scattered cells were found

Fig. 15. CTB-labeled cells and terminals in Ri, CM, and A1Mcontralateral to the injection of CM (case 7). Labeled cells are concen-trated in layer III of all areas. In CM, terminals formed bands inlayers III and V and also radial columns that spanned layers I–III.For abbreviations, see list. Scale bar � 500 �m.


in either CM or RM, consistent with weak connectionbetween lateral R and RM.

In case 4, injections were made into the core areas A1and the lateral half of R (Fig. 23). In caudal sectionscontaining A1 (#244 –284), FR cells were located in thesupragranular A1 and became concentrated mediallynear the border with R. Additional cells were in CM andML. Rostrally, most of the labeled cells were clusteredin supragranular R (CTB, #292–332). Some cells werealso found in caudal AL and were scattered in RM andPro. In the most rostral section (#CTB 340), most of thecells were in supragranular RT, with additional cells inRTL and Pro. Scattered cells were found in infragranu-lar RPB.

Interhemispheric connections of A1and R

In case 4, the injections of CTB and FR into A1 and Rlabeled cells primarily in the core and medial belt regionsof the opposite hemisphere (Fig. 26). In the most caudalsections of the contralateral hemisphere (#172–148), cellsand terminals labeled by the CTB injection of A1 werelargely confined to the lower half of layer III in area CM(#148–172), reflecting a strong reciprocal connection be-tween A1 and contralateral CM. A few cells, but no ter-minals, were found in the deep middle layer of Ri, medialto CM (#156), indicating a one-way projection to A1. Ros-tral to these sections, the greatest concentration of labeled

Fig. 16. Summary of ipsilateral (left) and interhemispheric (right)connections of CM. Top panels illustrate connections (arrows) of CMon schematic diagram of marmoset auditory cortex. Arrow size isproportional to connection strength, as indicated in the histogramsbelow each panel. Double arrows indicate reciprocal connection. Sin-gle arrows indicate unidirectional projections. Dashed lines indicate

infragranular projection. White arrows (R, RT) indicate connectionswith CM confined to the medial half of each area. Summary does notreflect absent infragranular projections observed in rostral fields aftercaudal CM injection (case 3, see text). Bottom left, white bar indicatesthat cell counts for ipsilateral CM may be inaccurate (deflated) due tomasking by the tracer injection. For abbreviations, see list.


Fig. 17. Ipsilateral cortical connections of areas RM and R, case 1.Series of serial sections are arranged from caudal (upper left) torostral (lower right). BDA-labeled cells (filled squares) and terminals(shading) are drawn onto each section, showing borders betweenareas identified by architectonic criteria. Cells labeled by FR (opentriangles), FE (open circles), and double-labeled cells (asterisk). FE

and FR injections indicated by dashed outlines in sections 144, 159.FB*, location of FB injection extending into white matter below areaML (not plotted). Inset: Schematic of marmoset auditory cortex show-ing location of BDA injection in RM, FR in medial R, and FE in lateralR. For abbreviations, see list.

cells and terminals was found in layer III of A1 laterally,matching the position of the injection in the oppositehemisphere (#108–132). The patch of homotopic label inA1 extended rostrally into R/RT for about 2 mm (#36–92).In one section there were labeled cells in caudal AL andRM, but not in sections further rostral, suggesting thatthe interhemispheric connection with A1 was focal, ratherthan continuous as in the core. For the FR injection into R,transport was relatively weak, covering a narrower range(CTB, #116–76). The cells in these sections were mostlydistributed along the core in supragranular R and A1.

In case 5, most of the labeled cells from the FR injectionof A1 labeled cells in contralateral A1M, with muchweaker extension rostrally into A1 (Fig. 27). The secondgreatest concentration of cells was found in CM, followedby Ri. Scattered cells were located in ML or CL of thelateral belt. Cells were exclusively located in layer III.

In case 1, the two injections of R resulted in a similarpattern of label, although greater numbers of cells werelabeled by the more lateral FE injection (Fig. 20). Overall,most of the labeled cells were contained within A1 and R,with fewer cells in the lateral and medial belts. Caudally,in sections containing A1 (BDA, #154–99), labeled cellswere concentrated in layer III of A1. A few cells in A1 weredouble-labeled by both injections. There were no cells inCM across this range until near the border between A1and R (BDA, #119–99), where cells were found in layer III.Rostrally, in sections containing R (BDA, #79–99), labeledcells were located in layer III of R and RM. A few double-labeled cells were in R. A small number of cells were

scattered in layer III of ML. There were no labeled cells inRi of any sections.

In case 2, no labeled cells were found in the oppositehemisphere after FR injection of R, even though labeledcells were numerous ipsilaterally and in the thalamus (dela Mothe et al., 2006).

Summary of A1 and R connections

After injections of A1 in two cases, the densest connec-tions were within the core, where labeled cells and termi-nals were continuously distributed along its entire rostro-caudal extent, including R and RT (Fig. 28, left).Connections with the lateral belt were reciprocal and alsocontinuous from rostral to caudal but were less denseoverall. Focal, but strong reciprocal projections were re-vealed with portions of the medial belt areas RM and CM,which were otherwise lightly labeled. There were no in-jections of the medial A1, so it was not possible to confirmwhether it had denser connections with most of the medialbelt, as suggested by injections of RM and CM. Connec-tions with the parabelt region and areas medial to themedial belt (i.e., Ri and Pro) were mostly characterized byfeedback projections from cells in the infragranular lay-ers, but there was some evidence of a forward projection toPro from A1. There were no connections between A1 andcortex ventral to the parabelt in the STS.

Injections of R primarily labeled cells in the core in acontinuous band that extended away from the injectionsites into A1 caudally and RT rostrally (Fig. 29, left). Therelative medial-lateral position of the injection was gen-erally reflected by the positions of the labeled cells in thecore of both hemispheres. Connections with the lateralbelt favored AL, with fewer cells in either ML or RTL.Labeling in the medial belt was dependent on injectionlocation. Injections of lateral R produced scattered label-ing in RM, CM, or RTM. In contrast, one injection ofmedial R produced strong labeling in RM, with additionallabeling in CM. This pattern is consistent with the gap inconnections with the lateral core areas after injections inRM (see above). Very few cells were found in RPB after Rinjections.

Interhemispheric homotopic and heterotopic connec-tions were generally reciprocal and were limited to thedeep part of layer III after injections of A1 or R (Figs. 28,29, right). The densest connections of both areas werehomotopic, but both divisions of the core were stronglyinterconnected between hemispheres. The densest heter-otopic connections of A1 were with CM, whereas R wassimilarly linked to CM and RM. A1 also received inputsfrom Ri, but no connections were found after injections ofR in the opposite hemisphere. Connections with the lat-eral belt were very sparse for A1 and especially R. Therewere no connections with the parabelt or insula.

Laminar specificity of connections

Because BDA and CTB transport is both anterogradeand retrograde, it was possible to visualize the laminardistribution of labeled cells and terminals in areas con-nected to the injection site. In the present study, severaldistinct types of connection patterns were observed. Ipsi-laterally, most of the connections between and withinareas were characterized by groups of labeled pyramidalcells in supragranular (layer II/III) and infragranular(layer V/VI) layers, overlapped by a haze of anterogradelabel (Fig. 24). The density of the anterograde terminal

Fig. 18. Anterograde BDA projections (arrows) to subcorticalstructures from an RM injection (case 1). A: Patch of BDA-labeledterminals in lateral nucleus of the amygdala. B: Acetylcholinesterase(AChE) stain of section corresponding to A. C: Elongated strip ofBDA-labeled terminals in the ventral caudate nucleus. D: AChE stainof section corresponding to C. BA, basal nucleus of the amygdala. Forother abbreviations, see list. Scale bar � 1 mm in B (applies to A,B);2 mm in D (applies to C,D).


labeling was commensurate with the density of labeledcells in the area.

A second pattern was commonly observed near injectionsites and sometimes in dense projections to distant areas.Here, patches of anterograde label formed a continuouscolumn that spanned all layers. These columns weresometimes adjacent to columns exhibiting a different pro-jection pattern but within the same architectonic field(Fig. 10B).

A third pattern was restricted to interhemispheric con-nections. These were generally characterized by overlap-ping labeled cells and terminals confined to layer III. Insome instances, the anterograde label formed a continu-ous column that spanned layers I–III (Fig. 15). A second-ary band of anterograde label was sometimes observed inlayer V, associated with very dense cell labeling above inlayer III. Often, there were labeled cells without clearevidence of anterograde labeling (e.g., projections from Rito A1).

In a fourth pattern, observed ipsilaterally, connectionsbetween certain areas were characterized by projectionsfrom labeled cells, but not terminals, located in the infra-granular layers of one or more areas. This pattern typifiedprojections from the parabelt to the core, from rostralauditory areas to caudal CM (Fig. 10), and from entorhinalcortex to CM (Fig. 11A). A similar pattern was notedpreviously after injections of the caudal lateral belt/parabelt region in macaques (Galaburda and Pandya,1983). This type of projection probably strictly reflectsfeedback to one or more layers of the target areas. Becauseour injections spanned all cortical areas in most cases, wecould not determine the laminar targets of those projec-tions.

In some areas, a fifth pattern was sometimes found.After RM injections, dense projections to layer IV werefound in sections containing RM, RTM, AL, and RPBrostral to the injection site (Fig. 10A,B). The heavy layerIV projection was in addition to normal (type 1) labeling inthe other layers. This pattern indicates that an exception-ally strong feedforward projection overlaps with thethalamocortical projections in layer IV in those areas. Incontrast, caudally directed projections from injections inRM produced the typical dense labeling above and belowlayer IV, but weak terminal labeling within. This fifthpattern was also observed after CM injections in projec-tions to ML, and to a lesser extent, CPB, but not withinCM (Fig. 10C,D). Galaburda and Pandya (1983) also de-scribed a strong layer IV projection to rostral fields fromthose located caudally. In the caudal direction, they re-ported that the lower laminae of rostral areas projected tolayer I of caudal areas. They also found a layer I projectionfrom the medial belt to the core and lateral belt. We didnot observe a prominent projection to layer I in this studyfrom any of our injections in the core or medial belt,suggesting that methodological differences may accountfor the discrepancy between studies.

DISCUSSION

In the present study, the anatomical organization of theauditory cortex in primates was studied in marmoset mon-keys by concurrent analysis of architectonic features andconnections of areas in the core (A1, R) and medial belt(CM, RM) regions. Overall, these findings indicate thatthe organization of the marmoset auditory cortex is com-

parable to that of other New and Old World primates. Inaddition to confirming regional distinctions between thecore and medial belt, the data revealed clear differencesbetween RM and CM. These findings are discussed belowalong with their functional implications.

RM and CM are functionally distinctauditory areas

The main finding of the present study was that RM andCM represent anatomically distinct areas of the auditorycortex. Placed within the context of several other observa-tions, we conclude that RM and CM are functionally dis-tinct areas as well. First, RM and CM are architectonicallydissimilar. CM is much more primary-like, as revealed bydense myelination across layers III–VI and elevated ex-pression of parvalbumin and cytochrome oxidase in thethalamo-recipient layers. The attenuation of these fea-tures in RM was more similar to that of the lateral beltareas. The architectonic profiles of RM and CM are con-sistent with descriptions of corresponding areas in otherprimates (Pandya and Sanides, 1973; Jones and Burton,1976; Imig et al., 1977; Galaburda and Pandya, 1983;Morel and Kaas, 1992; Morel et al., 1993; Jones et al.,1995; Kosaki et al., 1997; Hackett et al., 1998a).

Second, thalamic inputs to RM and CM arise from dif-ferent divisions of the MGC (de la Mothe et al., 2006). CMwas dominated by projections from the MGad and multi-sensory nuclei, whereas RM received inputs mainly fromthe MGpd. The architecture and inputs to the MGad andMGpd in macaque monkeys suggest that they may relayinformation to the cortex from distinct subcortical audi-tory pathways (Hashikawa et al., 1995; Molinari et al.,1995; Jones, 1997, 2003).

Third, the connections of RM and CM within the audi-tory cortex of both hemispheres were topographically dis-tinguishable. CM was more strongly interconnected withA1 and caudal areas outside the core (ML, CL, CPB),whereas the connections of RM favored areas rostral tothese. The most caudal portion of CM had especially weakconnections with the rostral fields. Overlap in the connec-tions of RM and CM occurred mainly in the middle third ofthe auditory cortex and then became increasingly diver-gent toward its rostral and caudal poles. Similar trendshave been noted after injections of the core, lateral belt,and parabelt in other primates, suggesting that there islimited direct communication between the most rostraland caudal domains of the auditory cortex (Galaburda andPandya, 1983; Luethke et al., 1989; Morel and Kaas, 1992;Morel et al., 1993; Hackett et al., 1998a). Compared withCM or RTM, however, RM appears to have more wide-spread connections with caudal and rostral fields, consis-tent with its more central location along the rostrocaudalaxis (Jones et al., 1995; Hackett et al., 1998a).

Fourth, RM and CM have unique connections with ar-eas beyond the auditory cortex. RM projected to the lateralamygdala and ventral caudate nuclei. CM did not projectto these nuclei but received strong inputs from the ento-rhinal cortex and had dense reciprocal connections with Riand posterior parietal cortex. RM had no significant con-nections with any posterior parietal or somatosensoryfield. These results provide indirect anatomical supportfor observations of bimodal auditory and somatosensoryactivity in CM of macaque monkeys (Schroeder et al.,2001; Fu et al., 2003). The results are also consistent withstudies in macaques that demonstrated topographic seg-


Fig. 19. Ipsilateral cortical connections of areas RM and R, case 2.Series of serial sections are arranged from caudal (upper left) torostral (lower right). BDA-labeled cells (filled squares) and terminals(shading) are drawn onto each section, showing borders betweenareas identified by architectonic criteria. FR-labeled cells are indi-

cated by open triangles. Hatching, diffusion and local tissue damagefrom FB injection involving the CL and A1 (cells not plotted). Inset:Schematic of marmoset auditory cortex showing location of BDAinjection in RM and FR injection in lateral R. For abbreviations, seelist.

regation of connections between the rostral and caudalbelt and the parabelt with functionally distinct regions ofthe prefrontal and posterior parietal cortex (Raczkowskiet al., 1976; Hackett et al., 1999; Romanski et al., 1999a,b;Lewis and Van Essen, 2000). As noted for connectionswithin the auditory cortex, the segregation of connectionswith auditory related fields becomes more strict with ros-tral or caudal distance from the “pivotal center” of theauditory cortex, which we loosely define as the border ofA1 and R.

Comparisons of anatomical and physiological profilesacross studies indicate that area RM of marmosets corre-sponds to the following areas identified in other primates:

“proA” (Pandya and Sanides, 1973; Galaburda and Pan-dya, 1983); “a” (Merzenich and Brugge, 1973); “A-m” or“M” (Jones et al., 1995; Kosaki et al., 1997); “Pi” (Burtonand Jones, 1976; Cheung et al., 2001); and “RM” (Imig etal., 1977; Morel and Kaas, 1992; Morel et al., 1993; Hack-ett et al., 1998a; Romanski et al., 1999a). Accordingly,area CM of marmosets corresponds to the following areas,at least in part: “paAc” (Pandya and Sanides, 1973; Gala-burda and Pandya, 1983); “P-m” (Jones et al., 1995; Ko-saki et al., 1997); “PA” (Jones and Burton, 1976; Robinsonand Burton, 1980a); and “CM” or “C” (Merzenich andBrugge, 1973; Imig et al., 1977; Pfingst and O’Connor,1981; Brugge, 1982; Morel and Kaas, 1992; Morel et al.,



Fig. 20. Interhemispheric cortical connections of areas RM and R,case 1. Series of serial sections are arranged from caudal (upper right)to rostral (lower left). BDA- labeled cells (filled squares) and terminals(shading) are drawn onto each section, showing borders betweenareas identified by architectonic criteria. Cells labeled by FR (open

triangles), FE (open circles), and double-labeled cells (asterisk). Inset:Schematic of marmoset auditory cortex showing location of BDAinjection in RM, FR in medial R, and FE in lateral R in the contralat-eral hemisphere. For abbreviations, see list.

1993; Rauschecker et al., 1997; Romanski et al., 1999a).With respect to differences in the size and extent of CMbetween studies, these are most likely due to differencesin interpretation, rather than differences between speciesor individual animals of the same species. We noted gra-dients in the both the architecture and connections of CMthat could be used to justify its division into areas medialand caudal to A1 (e.g., MM, middle medial; CM, caudome-dial). This distinction has been proposed and illustrated insummary diagrams of the macaque monkey but so far notverified (Kaas and Hackett, 1998, 2000). Further studieswill be required to resolve this issue.

Serial and parallel processing in the coreand medial belt

Injections of the core and medial belt areas in this studyrevealed that RM and CM have strong reciprocal connec-tions with infragranular and supragranular layers of thecore. Rostrocaudal topography was evident in these con-nections, such that RM was more densely connected with

R and RT, whereas CM had stronger connections with A1.In addition, RM and rostral CM had connections with allthree core areas, whereas caudal CM had only sparseinfragranular inputs from the rostral core. In addition toinputs from the core, RM and CM had topographic con-nections with the belt and parabelt. These connectionsinvolved cells in supragranular and/or infragranular lay-ers and were generally reciprocal. Remarkably, the stron-gest connections of RM and CM were from within themedial belt, accounting for 40% of all labeled cells, versus27% in the core. The remainder was mostly distributedamong the lateral belt and parabelt. In contrast, injectionsof the core revealed reciprocal suprgranular and infra-granular connections with the medial and lateral beltareas but only sparse connections with infragranular cellsin the parabelt. Thus, it appears that the core region ofmarmosets is instructing the medial and lateral belt areasvia strong reciprocal connections at all rostrocaudal levels,whereas the belt and parabelt regions are also stronglyinterconnected. This is consistent with findings in other



Fig. 21. Interhemispheric cortical connections of areas RM and R,case 2. Series of serial sections are arranged from caudal (upper right)to rostral (lower left). BDA-labeled cells (filled squares) and terminals(shading) are drawn onto each section, showing borders between

areas identified by architectonic criteria. Cells labeled by FR (opentriangles). Inset: Schematic of marmoset auditory cortex showinglocation of BDA injection in RM and FR in lateral R in the contralat-eral hemisphere. For abbreviations, see list.


primates (Aitkin et al., 1988; Morel and Kaas, 1992; Morelet al., 1993; Hackett et al., 1998a), from which it has beenconcluded that the parabelt region receives auditory cor-tical inputs through an intermediate stage of processing inthe belt region (Rauschecker et al., 1997; Kaas and Hack-ett, 1998).

To these results it should be added that both RM andCM receive dense inputs from the MGpd and MGad, re-spectively, whereas the primary inputs to R and A1 arisefrom the MGv (de la Mothe et al., 2006). These connectionsreflect inputs from at least two separate subcortical audi-tory pathways. At present, it is not clear from either theanatomy or physiology whether neurons in the core orthalamus represent the primary drive to neurons in the

medial belt. Most physiological studies indicate that ei-ther a reversal or disruption in the tonotopic gradientoccurs at the A1/CM border and that neurons in CM aremore broadly tuned than those in A1 (Merzenich andBrugge, 1973; Imig et al., 1977; Kosaki et al., 1997; Raus-checker et al., 1997; Recanzone et al., 2000a; Kajikawa etal., 2005). Recanzone and colleagues added that neuronsin CM had longer response latencies and were relativelymore selective for spatial location than neurons in A1(Recanzone, 2001; Recanzone et al., 2000b), as previouslynoted (Rauschecker et al., 1997), similar to the caudola-teral belt area (Tian et al., 2001). These response proper-ties support the conclusions of Rauschecker et al. (1997)that auditory information is processed in series between

Fig. 22. Summary of ipsilateral (left) and interhemispheric (right)connections of RM. Top panels illustrate connections (arrows) of RMon a schematic diagram of the marmoset auditory cortex. Arrow sizeis proportional to connection strength, as indicated in the histogramsbelow each panel. Double arrows indicate reciprocal connection. Sin-gle arrows indicate unidirectional projections. Open arrowheads indi-cate probable reciprocal connections based on other injections in this

study. White arrows (RM–R, RM–RT) indicate connections betweenareas confined to the medial half of R and RT. There was no clearinterhemispheric BDA transport after either RM injection. Bottomleft, white bar indicates that cell counts for the ipsilateral RM may beinaccurate (deflated) due to masking by the tracer injection. Forabbreviations, see list.


Fig. 23. Ipsilateral cortical connections of area A1 and R, case 4.Series of serial sections are arranged from caudal (upper left) torostral (lower right). CTB-labeled cells (filled circles) and terminals(shading) are drawn onto each section, showing borders between

areas identified by architectonic criteria. Cells labeled by FR (opentriangles). Inset: Schematic of marmoset auditory cortex showinglocation of CTB injection in A1 and FR in R. For abbreviations, seelist.

A1 and CM. In that key study, responses to tones wereabolished in CM after A1 ablation, but responses to com-plex sounds remained, whereas responses in R were un-affected by the A1 lesion. Responses to complex sounds inCM were thought to be preserved because they were me-diated by intact inputs from the MGd.

This conclusion may be consistent with the results of arecent study in macaques, in which latencies for toneswere longer in CM than the A1, whereas latencies fornoise bursts were shorter (Lakatos et al., 2005). In con-trast, Kajikawa et al. (2005) reported that average mini-mum response latencies for tones and noise were shorterin CM of marmosets. Short latency responses in CM havebeen reported by others as well (Bieser and Muller-Preuss, 1996; Scott et al., 2000).

Thus, although the precise nature of the relationshipbetween the medial belt and core remains to be deter-mined, it can be concluded that responses in CM are atleast partly dependent on intact feedforward inputs fromthe A1, consistent with its position in the auditory corticalhierarchy (Rauschecker et al., 1997). Simultaneous re-cordings from the core, medial belt, and perhaps thalamuswould be especially useful in resolving some of these is-sues, particularly if laminar array electrodes could beemployed to examine timing across laminae in the cortex(Schroeder and Foxe, 2002; Lakatos et al., 2005).

Sources of somatosensory input to theauditory cortex

Perhaps the clearest difference between RM and CMrevealed by the results of the present study was the strongreciprocal connection between CM and Ri. This somato-sensory area occupies the fundus of the lateral sulcuscaudal to the insula, separating S2 on the upper bankfrom CM on the lower bank, and appears to correspond tothe ventral somatosensory area (VS). The somatosensoryfeatures of Ri have been fairly well studied in both Newand Old World primates (Leinonen, 1980; Friedman andMurray, 1986; Friedman et al., 1986; Cusick et al., 1989;Krubitzer et al., 1995; Burton et al., 1995; Qi et al., 2002;Disbrow et al., 2003). There is also evidence that parts of

Ri may have a vestibular function as well (Grusser et al.,1990a,b; Akbarian et al., 1992, 1994; Guldin et al., 1992).

The significance of the connections between Ri and CMis that they represents the most likely source of somato-sensory input to CM, and perhaps other areas of auditorycortex, as observed in macaque monkeys. In studies ofcaudal somatosensory areas, Robinson and Burton(1980a–c) reported unimodal and bimodal auditory andsomatosensory responses in Ri and Pa, which correspondsto CM. In Ri, 74% of 199 units were responsive to somato-sensory stimulation. In Pa, 57% of 75 units were respon-sive to cutaneous stimulation. Most of these responseswere confined to the upper body, and nearly half of thereceptive fields were bilateral. About 16% of the neuronsin Pa and the extension of Ri onto the lower bank of thelateral sulcus were responsive to auditory or convergentauditory-somatosensory stimulation, although Pa was notcompletely mapped. Neurons responsive only to somaticstimulation were intermingled with those responsive onlyto sound. In addition, three neurons responded to audi-tory, visual, and somatic stimulation in the caudal part ofthe Pa, whereas additional auditory or visual responseswere located further caudal. These neurons may havebeen located in the portion of the Tpt that wraps onto thelower bank of the lateral sulcus from the STG (Leinonen etal., 1980).

These results were recently confirmed by additionalstudies of CM in macaques (Schroeder et al., 2001; Schr-oeder and Foxe, 2002; Fu et al., 2003) and a correspondingfield in humans (Foxe, 2002; Caetano, 2005). In thesestudies, a majority of neurons in CM were responsive toboth auditory and somatic stimulation in the form of elec-trical stimulation of the median nerve in the hand ormechanical stimulation of the upper body. Both cutaneousand proprioceptive responses were observed at short la-tencies, matching those evoked by auditory stimulation.In contrast, control recordings in the adjacent core area,A1, revealed no significant response modulation by so-matosensory stimulation, suggesting that the inputs re-sponsible for bimodal activity in CM are not likely tocharacterize all auditory areas equally.

Given the proximity of Ri to CM, it is not surprisingthat Ri could be a source of somatosensory input to CM,as well as other nearby areas, including CL and Tpt. Incats, auditory-somatosensory interactions were found ina comparable zone located between the suprasylvianand anterior ectosylvian sulci (Berman, 1961a,b; Car-reras and Andersson, 1963; Dehner et al., 2004) andalso within the AES (Clemo and Stein, 1983). In pri-mates, it has often been overlooked that Ri has connec-tions with the cortex in the vicinity of the caudal beltregion. Following injections of wheat grerm agglutinin-horseradish peroxidase (WGA-HRP) into physiologi-cally defined locations along the lateral sulcus of mar-moset monkeys, labeled cells were found in the caudalfundus, corresponding to Ri (Aitkin et al., 1988). Oneinjection, placed into presumptive A1 (BF � 8 kHz),labeled a narrow band that spanned layers in Ri. A morecaudal injection in either A1 or CM (BF � 16 kHz) alsolabeled cells and terminals in Ri and extended slightlyonto the upper bank. In macaque monkeys, degenerationwas observed in Ri and posterior insula after lesions of thecortex corresponding to the lateral belt and parabelt (Pan-dya et al., 1969; Pandya and Sanides, 1973; Pandya andRosene, 1993). Tracer transport studies have also revealed

Fig. 24. Patchy distribution of labeled cells and terminals in ad-jacent sections of area CM after injection of CTB into A1. A: Densityof label is highest in layer II/III. A secondary zone of labeled cells andterminals overlap in layers V/VI. B: Adjacent section showing weakeranterograde projections to layer II/III. Scale bar � 500 �m in B(applies to A,B).


sporadic evidence of such connections in several primatespecies (Galaburda and Pandya, 1983; Friedman et al.,1986; Morel and Kaas, 1992; Morel et al., 1993; Hackett etal., 1998a). In contrast, there is little evidence for signif-icant connections between S2 and auditory cortex (Jonesand Powell, 1969; Friedman et al., 1986; Krubitzer andKaas, 1990; Burton et al., 1995; Lewis and Van Essen,2000; Qi et al., 2002; Disbrow et al., 2003).

In the present study labeled cells were consistentlyfound after two of four injections involving CM. Whenlabeled cells were found in S2, such labelappeared to bethe result of spread of the tracer into a lesion or track inthe upper bank, although we could not rule out a legiti-mate projection with certainty. However, the absence ofconnections with S2 from the other two CM injectionssuggests that significant connections are unlikely. In onecase, the heavy labeling of Ri stopped cleanly at the borderwith S2 on the upper bank.

The discovery of multisensory covergence in CM iseven less surprising, given recent evidence of visualinteractions involving the auditory cortex. Neurons inlayer VI of the core, belt, and parabelt areas have beenfound to project to layer I of areas 17 and 18 of thevisual cortex (Falchier et al., 2002; Rockland and Ojima,2003). There is also limited evidence that the connec-tions may be reciprocal. These findings echo reportsthat eye position modulates the responses of neurons inthe inferior colliculus, core, and caudal belt (Groh et al.,2001; Werner-Reiss et al., 2003; Fu et al., 2004).

Considered together, all these findings suggest thatmultisensory influences of one variety or another may bediscovered in other areas of auditory cortex. The findingsof the present study indicate that Ri is the most likelysource of somatosensory input to CM and perhaps othercaudal fields. Given the absence of significant projectionsto RM, however, any multisensory interactions that may

Fig. 25. Ipsilateral cortical connections of area A1, case 5. Series ofserial sections are arranged from caudal (upper left) to rostral (lowerright). FR-labeled cells (open triangles) are drawn onto each section,showing borders between areas identified by architectonic criteria.

Black shading, core of FR injection. Gray shading, heavy labeling anddiffusion of FR injection. Inset: Schematic of marmoset auditorycortex showing location of FR injection into lateral A1. For abbrevia-tions, see list.


be identified in this field are likely to arise from anothersource.

Significance of connections with areasoutside the auditory cortex

After injections of RM in this study, strictly anterogradeprojections were discovered in the lateral nucleus of the

amygdala and the ventral caudate. Although these projec-tions were not observed after CM or core injections, onlyCM received inputs from the entorhinal cortex, furtherdistinguishing the functional segregation of the rostraland caudal auditory areas.

The projections from RM to the amygdala have not beenobserved in previous studies, although some projections

Fig. 26. Interhemispheric cortical connections of area A1 and R,case 4. Series of serial sections are arranged from caudal (upper right)to rostral (lower left). CTB-labeled cells (filled circles) and terminals(shading) are drawn onto each section, showing borders between

areas identified by architectonic criteria. FR-labeled cells (open trian-gles). L, lesion. Inset: Schematic of marmoset auditory cortex show-ing locations of CTB injection in A1 and FR in R in the contralateralhemisphere. For abbreviations, see list.


have been found in the insula and medial temporal pole(Herzog and Van Hoesen, 1976; Aggleton et al., 1980). Theprojection from RM is in line with previous findings inprimates in that connections with the amygdala tend to bestronger among rostral fields of the auditory cortex, reach-ing a maximum in the temporal pole, but are weak orabsent with caudal areas corresponding to the lateral beltor parabelt (Herzog and Van Hoesen, 1976; Aggleton etal., 1980; Turner et al., 1980; Kosmal et al., 1997; Yukie,2002). In that regard, the absence of CM projections to theamygdala in the present study is consistent with the to-pographic gradients observed laterally. Similarly, the lack

of inputs from the core is also consistent with previousstudies in primates. The inputs from the rostral areas ofthe auditory cortex to the amygdala may influence othercortical areas that receive projections from this structure(Romanski et al., 1993).

The RM projection to the tail of the caudate nucleus hasalso not been previously reported, although the existenceof this input is not surprising because most of the lateralbelt and parabelt areas of the primate auditory cortex areknown to project to some part of the caudate or putamen(Borgmann and Jurgens, 1999; Yeterian and Pandya,1998), similar to what has been observed in other mam-

Fig. 27. Interhemispheric cortical connections of area A1, case 5.Series of serial sections are arranged from caudal (upper right) torostral (lower left). FR-labeled cells (open triangles) are drawn ontoeach section, showing borders between areas identified by architec-

tonic criteria. Inset: Schematic of marmoset auditory cortex showinglocations of FR injection into A1 in the contralateral hemisphere. Forabbreviations, see list.


mals (Reale and Imig, 1983; Romanski and LeDoux, 1993).The striatal projections of the core are less clear. Borg-mann and Jurgens (1999) found no evidence of projectionsto the striatum after injections of the core despite strongprojections from areas corresponding to the lateral beltand parabelt. In apparent contrast, Yeterian and Pandya(1998) reported a “modest” projection after injection of alarge part of the core that appeared to involve A1 and R.However, this injection also extended into area RM (areaProA) and labeled parts of the putamen and head and tailof the caudate. Therefore, it is not clear whether the labelresulted from projections from the core, RM, or both. Theabsence of striatal connections after injection of CM in thepresent study was somewhat surprising, because Yeterianand Pandya (1998) found projections to the putamen andhead and tail of the caudate after an injection that in-

cluded this area. However, that injection also involved theTpt, which has strong projections to the striatum.

Area CM was the only area in this study to exhibitconnections with the entorhinal cortex. After CTB injec-tions in the rostral and caudal CM, labeled cells weredistributed along most of the rostrocaudal extent of theinferior temporal lobe but were restricted to the lowerlayers of this cortex. There was no evidence of an antero-grade projection to this region, despite dense anterogradeprojections to auditory areas of the temporal lobe in thesame tissue sections. Injections of RM produced neitheranterograde nor retrograde labeling in the entorhinal cor-tex. These findings are only partly consistent with otherstudies in primates. After injections in various divisions ofthe entorhinal cortex in macaque monkeys, labeled cells inthe vicinity of the auditory cortex are typically located in

Fig. 28. Summary of ipsilateral (left) and interhemispheric (right)connections of A1. Top panels illustrate connections (arrows) of A1 ona schematic diagram of the marmoset auditory cortex. Arrow size isproportional to connection strength, as indicated in the histogramsbelow each panel. Double arrows indicate reciprocal connection. Sin-

gle arrows indicate unidirectional projections. Dashed lines indicateinfragranular projection. Bottom left, white bar representing ipsilat-eral A1 indicates that cell counts may be inaccurate (deflated) in thearea injected by the tracer. For abbreviations, see list.


the insula, temporal pole, and rostral parts of the STGcorresponding to rostral areas of the medial belt, lateralbelt, and parabelt (Van Hoesen and Pandya, 1975; Insau-sti et al., 1987). Caudally, at the level of A1 or caudal CM,labeled cells are not found in the auditory cortex but tendto be limited to the upper bank of the STS. These resultsindicate that the rostral auditory cortex projects to theentorhinal cortex, although our BDA injections into RMrevealed no projections in marmosets. The strong entorhi-nal projection to CM in the present study may not havebeen observed in previous studies involving retrogradetracer injections in the entorhinal cortex or superior tem-poral lesions. At present, the known connections suggestthat rostral auditory areas project to the entorhinal cor-

tex, but only some caudal auditory fields (i.e., the CM)receive inputs from this region.

Additional connections were found between CM and theposterior parietal cortex after injections of both regions.Labeled cells after CM injections were plotted but notanalyzed or reconstructed because of concern that some ofthe labeling was due to unintended uptake by the cortex inthe upper bank of the lateral sulcus. However, a BDAinjection just behind the caudal terminus of the lateralsulcus in one case revealed the presence of retrogradelylabeled cells in CM, CL, and Ri, as well as light antero-grade projections to these same areas (Fig. 11). Thesefindings may be consistent with previous observations ofconnections between the caudal belt region (i.e., CM, CL,

Fig. 29. Summary of ipsilateral (left) and interhemispheric (right)connections of R. Top panels illustrate connections (arrows) of R on aschematic diagram of the marmoset auditory cortex. Arrow size isproportional to connection strength, as indicated in the histogramsbelow each panel. Double arrows indicate reciprocal connections ver-ifed by other injections in these areas, as results were based onretrograde tracers. Open arrowheads indicate that the projection is

assumed to be reciprocal based on laminar distribution of cells butcould not be verified using retrograde tracers. Single arrowheadsindicate unidirectional projections. Dashed lines indicate infragranu-lar projection. White arrow (RM–R) indicates that the connection withRM favored the medial half of R. Bottom left, white bar indicates thatcell counts for ipsilateral R may be inaccurate (deflated) due to mask-ing by the tracer injection. For abbreviations, see list.


Tpt) and the intraparietal sulcus in macaques (Lewis andVan Essen, 2000). The rostral auditory areas do not ap-pear to have posterior parietal connections.

Finally, we are uncertain about prefrontal connectionsof the medial belt region in marmosets. Blocks of tissuecontaining parts of the prefrontal cortex were removedfrom the main block prior to sectioning and generally notprocessed. Interested readers are referred to related stud-ies in macaques (Romanski et al., 1999a,b).

Correspondence with the auditory cortex ofother mammals

In our analysis of the results of this study and thoserelated to it (Kajikawa et al., 2005; de la Mothe et al.,2006), we were impressed by similarities in the organiza-tion of the monkey and cat auditory cortex noted earlier byJones and Burton (1976). They suggested that the primateparainsular field (RM) may represent the homologue ofarea AII in the cat, and the postauditory field (CM) mayrepresent the homologue of the anterior ectosylvian regionsituated between the AII and SII in the cat, known as theanterior auditory field (AAF). A number of findings in thepresent and previous studies support this hypothesis.First, AAF and CM occupy similar positions in the audi-tory cortex, relative to AI. In the cat, ferret, and severalother species, AAF and A1 share a high characteristicfrequency border (Knight, 1977; Imig and Reale, 1980;Phillips and Irvine, 1982; Rouiller et al., 1991; Wallace etal., 1991; Lee et al., 2004). In primates, the tonotopicgradient in A1 has been found to reverse or be disrupted atits caudal border with CM (Merzenich and Brugge, 1973;Imig et al., 1977; Kosaki et al., 1997; Rauschecker et al.,1997; Recanzone et al., 2000a; Cheung et al., 2001; Ka-jikawa et al., 2005). Second, the response properties ofneurons in AAF and CM are very similar to those in A1,except for significantly broader tuning bandwidth(Knight, 1977; Tian and Rauschecker, 1994; Kowalski etal, 1995; Eggermont, 1998; Kajikawa et al., 2005).

Third, although they are primary-like in the ways de-scribed above, CM and AAF are not primary fields. Inprimates, the MGv is the major input to the core areas(A1, R, RT), and in cats, the MGv projects strongly to theA1, PAF, and VPAF. In contrast, CM and AAF have stronginputs from AI and thalamic inputs that include the pos-terior nuclei and the rostral pole of the MGC (i.e., MGad,Pol; Lee et al., 2004; de la Mothe et al., 2006). Fourth, AAFand CM adjoin the somatosensory cortex, including apoorly defined region of the cortex in which auditory andsomatosensory representations appear to converge (seeabove discussion).

On the basis of these comparisons, it is quite clear thatCM resembles AAF more than any other area of auditorycortex. Although these common features are not sufficientto establish homology between these areas (i.e., inheritedfrom a common ancestor), the areas could aptly be de-scribed as corresponding. To the extent that AAF and CMare corresponding areas, like A1, comparisons of auditorycortex organization across taxa become more meaningfuland findings in one species can be more broadly applied.With time, evidence may accumulate supporting the cor-respondence of other areas, as well (e.g., the PAF-R;VPAF-RT; AII-RM). In any event, it is not anticipated thatcorresponding areas will be identical, rather, they aresimply more likely to have retained common features than

other areas. In that sense, the identification of correspond-ing areas is meaningful and instructive.

Correspondence of CM with posteromedialfields in other primates

Whereas the correspondence between areas identifiedas CM is fairly well established for monkeys, it is lesscertain for other primates, including humans. In a recentstudy, the core region in monkeys was identified in chim-panzees and humans on the basis of common architectonicfeatures (Hackett et al., 2001). In that same study, corre-spondence was also proposed between CM in monkeys anda distinct field in chimpanzees and humans located at themedial terminus of Heschl’s gyrus, known as the TD (vonEconomo and Koskinas, 1925). Additional areas extendposterior and laterally to fill out the planum temporaleregion, which is larger in chimpanzees and humans. Thus,on architectonic grounds, parts of the planum temporaleappear to correspond to CM, CL, ML, AL, and Tpt inmonkeys (Hackett, 2002; Sweet et al., 2005). In functionalimaging studies, the planum temporale, and subregionswithin, are activated during a variety of tasks, includingperception of speech and environmental sounds, speechproduction, and spatial perception (Griffiths and Warren,2002). Of special interest is functional magnectic reso-nance imaging evidence of auditory-somatosensory con-vergence in this area in humans (Foxe et al., 2002). Al-though further elaboration of these ideas is beyond thescope of this paper, it is important to continue to identifylinks between taxonomic groups in future studies of theauditory cortex to broaden the applicability of researchfindings regardless of species and thereby improve ourunderstanding of the auditory system.

Consistency with a working model of theprimate auditory cortex

A general conclusion reached in the present study wasthat the organization of the marmoset auditory cortexconformed well to the working model based on studies ofNew World and Old World primates. Although the core-belt-parabelt schema is the most well developed for themacaque monkey, it is important to note that findings inNew World primates (e.g., owl monkey, squirrel monkey)were equally influential in the development of this model(Imig et al., 1977; Morel and Kaas, 1992). In marmosets,however, anatomical studies have largely focused on theorganization of the core region, especially AI (Aitkin et al.,1988; Luethke et al., 1989). Thus, one motivation for con-ducting the current study in marmosets was to determinewhether the core-belt-parabelt model might also charac-terize this species. This appears to be especially importantbecause the marmoset has become a popular model forneurophysiological study of the auditory cortex (Aitkin etal., 1986; Luethke et al., 1989; Wang et al., 1995; deC-harms and Merzenich, 1996; deCharms et al., 1999; Lu etal., 2001a,b; Wang and Kadia, 2001; Liang et al., 2002;Nagarajan et al., 2002; Lu and Wang, 2004; Luczak et al.,2004; Eliades and Wang, 2005; Kajikawa et al., 2005;Kajikawa and Hackett, 2005; Philibert et al., 2005; Ben-dor and Wang, 2005).

With some exceptions, the present study revealed thatthe architectonic characteristics of the marmoset superiortemporal region were comparable to those described forthe macaque monkey (Pandya and Sanides, 1973; Jonesand Burton, 1976; Galaburda and Pandya, 1983; Morel et


al., 1993; Jones et al., 1995; Hackett et al., 1998a; Hackettet al., 2001) and other New World primates (Jones andBurton, 1976; Imig et al., 1977; Luethke et al., 1989; Moreland Kaas, 1992). The size of the auditory cortex wassmaller, about one-third the size of that of macaques, butmajor architectonic features were qualitatively similar,reflecting the typical medial-lateral and rostral-caudal ar-chitectonic gradients observed in other primates. On theother hand, a comparative quantification of architectonicdetails (e.g, cell types, cell size, cell or fiber density) mayultimately reveal significant species differences that werenot addressed by this study. To date, however, detailedarchitectonic analyses of these features have not beenconducted for the auditory cortex of primates other thanhumans. One surprising difference between marmosetsand other primates was that acetylcholinesterase (AChE)expression was relatively uniform across major regions ofauditory cortex. Typically, elevated AChE density in thelayer III/IV band is a reliable and robust marker of thecore region, where it is coextensive with dense expressionof cytochrome oxidase and parvalbumin. In this study,however, its expression was greatly reduced comparedwith tissue from other primates processed in our labora-tory using the same histochemical protocol (Hackett et al.,2001). In contrast, adjacent sections processed for cyto-chrome oxidase and parvalbumin revealed the expectedpattern of expression in the core (see Figs. 2, 3, 5). It is notknown whether species differences or methodological fac-tors account for this finding.

Additional support for a core-belt-parabelt system oforganization was found in the patterns of connectionsbetween regions. Injections of the medial belt revealedtopographic connections with the medial belt, lateral belt,core, and parabelt regions, as expected. Likewise, injec-tions of the core labeled cells and terminals within thecore and belt regions, consistent with previous results inmarmosets and tamarins (Aitkin et al., 1988; Luethke etal., 1989). The discovery of labeled cells, but not terminals,in the infragranular layers of the parabelt after injectionsof A1 or R was unexpected, given previous findings inmacaques and owl monkeys (Morel and Kaas, 1992; Morelet al., 1993; Hackett et al., 1998a). However, an infra-granular projection from the parabelt to the core wouldnot have been observed after injection of retrograde trac-ers into the parabelt. In addition, the absence of antero-grade label in the parabelt after core injections and theabsence of retrogradely labeled cells in supragranular lay-ers of the parabelt in this study indicate that the projec-tion from the parabelt is strictly feedback in nature. Thus,the overall pattern of connections in the marmoset isconsistent with the working model in that informationflow proceeds outward from the core to the parabelt via anintermediate stage of processing in the belt (Kaas andHackett, 1998). However, the model may need to beamended to include the present observation that feedbackprojections from the parabelt directly targeted the core inmarmosets (Figs. 28, 29).

Finally, physiological studies of the marmoset corrobo-rate some of the subdivisions identified anatomically inthe present study. On the basis of tonotopic reversals,Bendor and Wang (2005) identified A1, R, and RT on thesurface of the STG near the lateral sulcus, extendingprevious findings in this species concerning the locationand tonotopic organization of A1 (Aitkin et al., 1986, 1988;Luethke et al., 1989; Kajikawa et al., 2005; Philibert et al.,

2005). Both the size and extent of the core areas identifiedmatch the present findings. We would add, however, thatup to about one-third of the core extends medially into thelateral sulcus and has not been mapped in some studies.The significance of this may relate to the present obser-vations concerning differences between the medial andlateral halves of the core. In brief, the lateral halves of thecore areas had sparse connections with RM or rostral CM,whereas relatively dense connections were concentratedmedially in the core. The pattern was different in caudalCM, which had somewhat strong connections with A1M

and A1L caudally and A1L rostrally.Some evidence for such patterns can be found in a

previous study of marmosets (Aitkin et al., 1988). Al-though these patterns may simply reflect strict topo-graphic constraints, the patterns could also reflect func-tional specificity within the medial and lateral halves ofthe core that is preserved in its output to other areas. Asnoted above (Fig. 4), the medial half of the core, at leastin A1, was more densely myelinated than the lateralhalf, consistent with previous distinctions made inother primates, including humans (Pandya andSanides, 1973; Galaburda and Pandya, 1983; Hackett etal., 2001). The division of A1M and A1L near the edge ofthe lateral sulcus cuts across the rostrocaudal gradientof isofrequency contours; thus, there may be functionaldifferences between these divisions of A1. For example,recordings in A1 of owl monkeys and cats indicate thatthe representation of response parameters other thancharacteristic frequency may be spatially representedin maps that do not coincide with isofrequency contours(Schreiner, 1998; Recanzone et al., 1999; Read et al.,2002). Further study will be needed to clarify the func-tional significance of these details.

CONCLUSIONS

The results of this study indicate that the organiza-tion of the marmoset monkey auditory cortex closelymatches that of other primates, in line with our workingmodel of the primate auditory cortex. The medial beltareas RM and CM represent functionally distinct areasof the auditory cortex and of the medial belt region.Both areas receive dense projections from the core andare broadly connected with medial belt, lateral belt, andparabelt regions. Individually, RM and CM have dis-tinctive architectonic features and patterns of connec-tions. In particular, CM receives somatosensory inputsfrom the retroinsular somatosensory area (Ri) and hasadditional connections with the STS and posterior pa-rietal and entorhinal cortex. RM does not appear tohave connections with somatosensory fields but doesproject to the lateral nucleus of the amygdala and tail ofthe caudate nucleus. In addition, the collective findingssuggest that primate CM may correspond to areas TD inhumans and AAF in other mammals. Architectonic fea-tures and connections distinguish the core areas A1 andR from the belt and parabelt regions of the auditorycortex. Projections to the core from the parabelt origi-nated from infragranular cells, but there was no evi-dence that the core projects directly to the parabelt.These findings suggest minor revisions to the model.


ACKNOWLEDGMENTS

The authors thank Laura Trice and Mary Varghese foroutstanding technical assistance with histology and Jes-sica Stefanovic for help with cell plotting.

LITERATURE CITED

Aggleton JP, Burton MJ, Passingham RE. 1980. Cortical and subcorticalafferents to the amygdala of the rhesus monkey (Macaca mulatta).Brain Res 190:347–368.

Aitkin LM, Merzenich MM, Irvine DR, Clarey JC, Nelson JE. 1986. Fre-quency representation in auditory cortex of the common marmoset(Callithrix jacchus jacchus). J Comp Neurol 252:175–185.

Aitkin LM, Kudo M, Irvine DR. 1988. Connections of the primary auditorycortex in the common marmoset, Callithrix jacchus jacchus. J CompNeurol 269:235–248.

Akbarian S, Grusser OJ, Guldin WO. 1992. Thalamic connections of thevestibular cortical fields in the squirrel monkey (Saimiri sciureus).J Comp Neurol 326:423–441.

Akbarian S, Grusser OJ, Guldin WO. 1994. Corticofugal connections be-tween the cerebral cortex and brainstem vestibular nuclei in the ma-caque monkey. J Comp Neurol 339:421–437.

Beck E. 1928. Die myeloarchitektonische Felderung es in der SylvischenFurche gelegenen Teiles des menschlichen Schlafenlappens. J PsychNeurol 36:1–21.

Bendor D, Wang X. 2005. The neuronal representation of pitch in primateauditory cortex. Nature 436:1161–1165.

Berman AL. 1961a. Interaction of cortical responses to somatic and audi-tory stimuli in anterior ectosylvian gyrus of cat. J Neurophysiol 24:608–620.

Berman AL. 1961b. Overlap of somatic and auditory cortical responsefields in anterior ectosylvian gyrus of cat. J Neurophysiol 24:595–607.

Bieser A, Muller-Preuss P. 1996. Auditory responsive cortex in the squirrelmonkey: neural responses to amplitude-modulated sounds. Exp BrainRes 108:273–284.

Borgmann S, Jurgens U. 1999. Lack of cortico-striatal projections from theprimary auditory cortex in the squirrel monkey. Brain Res 836:225–228.

Brodmann K. 1909. Vergleichende Lokalisationslehre der Grosshirnrinde.Leipzig: Barth.

Brugge JF. 1982. Auditory cortical areas in primates. In: Woolsey CN,editor. Cortical sensory organization, vol 3. Multiple auditory areas.Clifton, NJ: Humana Press. p 59–70.

Burton H, Jones EG. 1976. The posterior thalamic region and its corticalprojection in New World and Old World monkeys. J Comp Neurol168:249–301.

Burton H, Fabri M, Alloway K. 1995. Cortical areas within the lateralsulcus connected to cutaneous representations in areas 3b and 1: arevised interpretation of the second somatosensory area in macaquemonkeys. J Comp Neurol 355:539–562.

Caetano. 2005.Carreras M, Andersson SA. 1963. Functional properties of neurons of the

anterior ectosylvian gyrus of the cat. J Neurophysiol 26:100–126.Cheung SW, Bedenbaugh PH, Nagarajan SS, Schreiner CE. 2001. Func-

tional organization of squirrel monkey primary auditory cortex: re-sponses to pure tones. J Neurophysiol 85:1732–1749.

Cipolloni PB, Pandya DN. 1989. Connectional analysis of the ipsilateraland contralateral afferent neurons of the superior temporal region inthe rhesus monkey. J Comp Neurol 281:567–585.

Clemo HR, Stein BE. 1983. Organization of a fourth somatosensory area ofcortex in cat. J Neurophysiol 50:910–925.

Cusick CG, Wall JT, Felleman DJ, Kaas JH. 1989. Somatotopic organiza-tion of the lateral sulcus of owl monkeys: area 3b, S-II, and a ventralsomatosensory area. J Comp Neurol 282:169–190.

de la Mothe LA, Blumell S, Kajikawa Y, Hackett TA. 2002. Connections ofauditory medial belt auditory cortex in marmoset monkeys. Soc Neu-rosci Abstr 261.266.

de la Mothe LA, Blumell S, Kajikawa Y, Hackett TA. 2006. Thalamicconnections of auditory cortex in marmoset monkeys: core and medialbelt regions. J Comp Neurol 496:72–96.

deCharms RC, Merzenich MM. 1996. Primary cortical representation ofsounds by the coordination of action-potential timing. Nature 381:610–613.

deCharms RC, Blake DT, Merzenich MM. 1999. A multielectrode implantdevice for the cerebral cortex. J Neurosci Methods 93:27–35.

Dehner LR, Keniston LP, Clemo HR, Meredith MA. 2004. Cross-modalcircuitry between auditory and somatosensory areas of the cat anteriorectosylvian sulcal cortex: a ’new’ inhibitory form of multisensory con-vergence. Cereb Cortex 14:387–403.

Disbrow E, Litinas E, Recanzone GH, Padberg J, Krubitzer L. 2003. Cor-tical connections of the second somatosensory area and the parietalventral area in macaque monkeys. J Comp Neurol 462:382–399.

Eggermont JJ. 1998. Representation of spectral and temporal sound fea-tures in three cortical fields of the cat. Similarities outweigh differ-ences. J Neurophysiol 80:2743–2764.

Eliades SJ, Wang X. 2005. Dynamics of auditory-vocal interaction in mon-key auditory cortex. Cereb Cortex 15:1510–1523.

Falchier A, Clavagnier S, Barone P, Kennedy H. 2002. Anatomical evi-dence of multimodal integration in primate striate cortex. J Neurosci22:5749–5759.

Foxe JJ, Wylie GR, Martinez A, Schroeder CE, Javitt DC, Guilfoyle D,Ritter W, Murray MM. 2002. Auditory-somatosensory multisensoryprocessing in auditory association cortex: an fMRI study. J Neuro-physiol 88:540–543.

Friedman DP, Murray EA. 1986. Thalamic connectivity of the secondsomatosensory area and neighboring somatosensory fields of the lat-eral sulcus of the macaque. J Comp Neurol 252:348–373.

Friedman DP, Murray EA, O’Neill JB, Mishkin M. 1986. Cortical connec-tions of the somatosensory fields of the lateral sulcus of macaques:evidence for a corticolimbic pathway for touch. J Comp Neurol 252:323–347.

Fu KM, Johnston TA, Shah AS, Arnold L, Smiley J, Hackett TA, GarraghtyPE, Schroeder CE. 2003. Auditory cortical neurons respond to somato-sensory stimulation. J Neurosci 23:7510–7515.

Fu KM, Shah AS, MN OC, McGinnis T, Eckholdt H, Lakatos P, Smiley J,Schroeder CE. 2004. Timing and laminar profile of eye position effectson auditory responses in primate auditory cortex. J Neurophysiol.92:3522–3531.

Galaburda AM, Pandya DN. 1983. The intrinsic architectonic and connec-tional organization of the superior temporal region of the rhesus mon-key. J Comp Neurol 221:169–184.

Galaburda A, Sanides F. 1980. Cytoarchitectonic organization of the hu-man auditory cortex. J Comp Neurol 190:597–610.

Gallyas F. 1979. Silver staining of myelin by means of physical develop-ment. Neurol Res 1:203–209.

Geneser-Jensen FA, Blackstad TW. 1971. Distribution of acetyl cholines-terase in the hippocampal region of the guinea pig. I. Entorhinal area,parasubiculum, and presubiculum. Z Zellforsch Mikrosk Anat 114:460–481.

Griffiths TD, Warren JD. 2002. The planum temporale as a computationalhub. Trends Neurosci 25:348–353.

Groh JM, Trause AS, Underhill AM, Clark KR, Inati S. 2001. Eye positioninfluences auditory responses in primate inferior colliculus. Neuron29:509–518.

Grusser OJ, Pause M, Schreiter U. 1990a. Localization and responses ofneurones in the parieto-insular vestibular cortex of awake monkeys(Macaca fascicularis). J Physiol 430:537–557.

Grusser OJ, Pause M, Schreiter U. 1990b. Vestibular neurones in theparieto-insular cortex of monkeys (Macaca fascicularis): visual andneck receptor responses. J Physiol 430:559–583.

Guldin WO, Akbarian S, Grusser OJ. 1992. Cortico-cortical connectionsand cytoarchitectonics of the primate vestibular cortex: a study insquirrel monkeys (Saimiri sciureus). J Comp Neurol 326:375–401.

Hackett T. 2002. The comparative anatomy of the primate auditory cortex.In: Ghazanfar A, editor. Primate audition: behavior and neurobiology.Boca Raton: CRC Press. p 199–226.

Hackett TA, Stepniewska I, Kaas JH. 1998a. Subdivisions of auditorycortex and ipsilateral cortical connections of the parabelt auditorycortex in macaque monkeys. J Comp Neurol 394:475–495.

Hackett TA, Stepniewska I, Kaas JH. 1998b. Thalamocortical connectionsof the parabelt auditory cortex in macaque monkeys. J Comp Neurol400:271–286.

Hackett TA, Stepniewska I, Kaas JH. 1999. Prefrontal connections of theparabelt auditory cortex in macaque monkeys. Brain Res 817:45–58.

Hackett TA, Preuss TM, Kaas JH. 2001. Architectonic identification of thecore region in auditory cortex of macaques, chimpanzees, and humans.J Comp Neurol 441:197–222.


Hackett TA, Karmos G, Schroeder CE, Ulbert I, Sterbing-D’Angelo SJ,D’Angelo WR, Kajikawa Y, Blumell S, de la Mothe L. 2005. Neurosur-gical access to cortical areas in the lateral fissure of primates. J Neu-rosci Methods 141:103–113.

Hashikawa T, Molinari M, Rausell E, Jones EG. 1995. Patchy and laminarterminations of medial geniculate axons in monkey auditory cortex.J Comp Neurol 362:195–208.

Herzog AG, Van Hoesen GW. 1976. Temporal neocortical afferent connec-tions to the amygdala in the rhesus monkey. Brain Res 115:57–69.

Hopf P. 1954. Die myeloarchitektonic des isocortex temporalis beim men-schen. J Hirnforsch 1:208–279.

Imig TJ, Reale RA. 1980. Patterns of cortico-cortical connections related totonotopic maps in cat auditory cortex. J Comp Neurol 192:293–332.

Imig TJ, Ruggero MA, Kitzes LM, Javel E, Brugge JF. 1977. Organizationof auditory cortex in the owl monkey (Aotus trivirgatus). J Comp Neurol171:111–128.

Insausti R, Amaral DG, Cowan WM. 1987. The entorhinal cortex of themonkey: II. Cortical afferents. J Comp Neurol 264:356–395.

Jones EG. 1997. The relay function of the thalamus during brain activa-tion. In: Steriade M, Jones E, McCormick D, editors. Thalamus. NewYork: Elsevier. p 393–531.

Jones EG. 2003. Chemically defined parallel pathways in the monkeyauditory system. Ann N Y Acad Sci 999:218–233.

Jones EG, Burton H. 1976. Areal differences in the laminar distribution ofthalamic afferents in cortical fields of the insular, parietal and tempo-ral regions of primates. J Comp Neurol 168:197–247.

Jones EG, Dell’Anna ME, Molinari M, Rausell E, Hashikawa T. 1995.Subdivisions of macaque monkey auditory cortex revealed by calcium-binding protein immunoreactivity. J Comp Neurol 362:153–170.

Jones EG, Powell TP. 1969. Connexions of the somatic sensory cortex of therhesus monkey. I. Ipsilateral cortical connexions. Brain 92:477–502.

Kaas JH, Hackett TA. 1998. Subdivisions of auditory cortex and levels ofprocessing in primates. Audiol Neurootol 3:73–85.

Kaas JH, Hackett TA. 2000. Subdivisions of auditory cortex and processingstreams in primates. Proc Natl Acad Sci U S A 97:11793–11799.

Kaas JH, Hackett TA, Tramo MJ. 1999. Auditory processing in primatecerebral cortex. Curr Opin Neurobiol 9:164–170.

Kajikawa Y, Hackett TA. 2005. Entropy analysis of neuronal spike trainsynchrony. J Neurosci Methods 149:90–93.

Kajikawa Y, de la Mothe LA, Blumell S, Hackett TA. 2005. A comparisonof neuron response properties in areas A1 and CM of the marmosetmonkey auditory cortex: tones and broad band noise. J Neurophysiol93:22–34.

Knight PL. 1977. Representation of the cochlea within the anterior audi-tory field (AAF) of the cat. Brain Res 130:447–467.

Kosaki H, Hashikawa T, He J, Jones EG. 1997. Tonotopic organization ofauditory cortical fields delineated by parvalbumin immunoreactivity inmacaque monkeys. J Comp Neurol 386:304–316.

Kosmal A, Malinowska M, Kowalska DM. 1997. Thalamic and amygdaloidconnections of the auditory association cortex of the superior temporalgyrus in rhesus monkey (Macaca mulatta). Acta Neurobiol Exp (Wars)57:165–188.

Kowalski N, Versnel H, Shamma SA. 1995. Comparison of responses in theanterior and primary auditory fields of the ferret cortex. J Neuro-physiol 73:1513–1523.

Krubitzer LA, Kaas JH. 1990. The organization and connections of somato-sensory cortex in marmosets. J Neurosci 10:952–974.

Krubitzer L, Clarey J, Tweedale R, Elston G, Calford M. 1995. A redefini-tion of somatosensory areas in the lateral sulcus of macaque monkeys.J Neurosci 15:3821–3839.

Lakatos P, Pincze Z, Fu KG, Javitt DC, Karmos G, Schroeder CE. 2005.Timing of pure tone and noise-evoked responses in macaque auditorycortex. Neuroreport 16:933–937.

Lee CC, Imaizumi K, Schreiner CE, Winer JA. 2004. Concurrent tonotopicprocessing streams in auditory cortex. Cereb Cortex 14:441–451.

Leinonen L. 1980. Functional properties of neurones in the parietal retro-insular cortex in awake monkey. Acta Physiol Scand 108:381–384.

Leinonen L, Hyvarinen J, Sovijarvi AR. 1980. Functional properties ofneurons in the temporo-parietal association cortex of awake monkey.Exp Brain Res 39:203–215.

Lewis JW, Van Essen DC. 2000. Corticocortical connections of visual,sensorimotor, and multimodal processing areas in the parietal lobe ofthe macaque monkey. J Comp Neurol 428:112–137.

Liang L, Lu T, Wang X. 2002. Neural representations of sinusoidal ampli-

tude and frequency modulations in the primary auditory cortex ofawake primates. J Neurophysiol 87:2237–2261.

Lu T, Wang X. 2004. Information content of auditory cortical responses totime-varying acoustic stimuli. J Neurophysiol 91:301–313.

Lu T, Liang L, Wang X. 2001a. Neural representations of temporallyasymmetric stimuli in the auditory cortex of awake primates. J Neu-rophysiol 85:2364–2380.

Lu T, Liang L, Wang X. 2001b. Temporal and rate representations oftime-varying signals in the auditory cortex of awake primates. NatNeurosci 4:1131–1138.

Luczak A, Hackett TA, Kajikawa Y, Laubach M. 2004. Multivariate recep-tive field mapping in marmoset auditory cortex. J Neurosci Methods136:77–85.

Luethke LE, Krubitzer LA, Kaas JH. 1989. Connections of primary audi-tory cortex in the New World monkey, Saguinus. J Comp Neurol285:487–513.

Merzenich MM, Brugge JF. 1973. Representation of the cochlear partitionof the superior temporal plane of the macaque monkey. Brain Res50:275–296.

Molinari M, Dell’Anna ME, Rausell E, Leggio MG, Hashikawa T, JonesEG. 1995. Auditory thalamocortical pathways defined in monkeys bycalcium-binding protein immunoreactivity. J Comp Neurol 362:171–194.

Morel A, Kaas JH. 1992. Subdivisions and connections of auditory cortex inowl monkeys. J Comp Neurol 318:27–63.

Morel A, Garraghty PE, Kaas JH. 1993. Tonotopic organization, architec-tonic fields, and connections of auditory cortex in macaque monkeys.J Comp Neurol 335:437–459.

Nagarajan SS, Cheung SW, Bedenbaugh P, Beitel RE, Schreiner CE,Merzenich MM. 2002. Representation of spectral and temporal enve-lope of twitter vocalizations in common marmoset primary auditorycortex. J Neurophysiol 87:1723–1737.

Pandya DN. 1995. Anatomy of the auditory cortex. Rev Neurol (Paris).151:486–494.

Pandya DN, Rosene DL. 1993. Laminar termination patterns of thalamic,callosal, and association afferents in the primary auditory area of therhesus monkey. Exp Neurol 119:220–234.

Pandya DN, Sanides F. 1973. Architectonic parcellation of the temporaloperculum in rhesus monkey and its projection pattern. Z Anat En-twicklungsgesch 139:127–161.

Pandya DN, Hallett M, Mukherjee SK. 1969. Intra- and interhemisphericconnections of the neocortical auditory system in the rhesus monkey.Brain Res 14:49–65.

Pfingst BE, O’Connor TA. 1981. Characteristics of neurons in auditorycortex of monkeys performing a simple auditory task. J Neurophysiol45:16–34.

Philibert B, Beitel RE, Nagarajan SS, Bonham BH, Schreiner CE, CheungSW. 2005. Functional organization and hemispheric comparison ofprimary auditory cortex in the common marmoset (Callithrix jacchus).J Comp Neurol 487:391–406.

Phillips DP, Irvine DR. 1982. Properties of single neurons in the anteriorauditory field (AAF) of cat cerebral cortex. Brain Res 248:237–244.

Qi HX, Lyon DC, Kaas JH. 2002. Cortical and thalamic connections of theparietal ventral somatosensory area in marmoset monkeys (Callithrixjacchus). J Comp Neurol 443:168–182.

Raczkowski D, Diamond IT, Winer J. 1976. Organization of thalamocorti-cal auditory system in the cat studied with horseradish peroxidase.Brain Res 101:345–354.

Rauschecker JP. 1998. Parallel processing in the auditory cortex of pri-mates. Audiol Neurootol 3:86–103.

Rauschecker JP, Tian B. 2004. Processing of band-passed noise in thelateral auditory belt cortex of the rhesus monkey. J Neurophysiol91:2578–2589.

Rauschecker JP, Tian B. 2000. Mechanisms and streams for processing of“what” and “where” in auditory cortex. Proc Natl Acad Sci U S A97:11800–11806.

Rauschecker JP, Tian B. 2004. Processing of band-passed noise in thelateral auditory belt cortex of the rhesus monkey. J Neurophysiol91:2578–2589.

Rauschecker JP, Tian B, Hauser M. 1995. Processing of complex sounds inthe macaque nonprimary auditory cortex. Science 268:111–114.

Rauschecker JP, Tian B, Pons T, Mishkin M. 1997. Serial and parallelprocessing in rhesus monkey auditory cortex. J Comp Neurol 382:89–103.


Read HL, Winer JA, Schreiner CE. 2002. Functional architecture of audi-tory cortex. Curr Opin Neurobiol 12:433–440.

Reale RA, Imig TJ. 1983. Auditory cortical field projections to the basalganglia of the cat. Neuroscience 8:67–86.

Recanzone GH. 2001. Spatial processing in the primate auditory cortex.Audiol Neurootol 6:178–181.

Recanzone GH, Schreiner CE, Sutter ML, Beitel RE, Merzenich MM. 1999.Functional organization of spectral receptive fields in the primaryauditory cortex of the owl monkey. J Comp Neurol 415:460–481.

Recanzone GH, Guard DC, Phan ML. 2000a. Frequency and intensityresponse properties of single neurons in the auditory cortex of thebehaving macaque monkey. J Neurophysiol 83:2315–2331.

Recanzone GH, Guard DC, Phan ML, Su TK. 2000b. Correlation betweenthe activity of single auditory cortical neurons and sound-localizationbehavior in the macaque monkey. J Neurophysiol 83:2723–2739.

Robinson CJ, Burton H. 1980a. Organization of somatosensory receptivefields in cortical areas 7b, retroinsula, postauditory and granular in-sula of M. fascicularis. J Comp Neurol 192:69–92.

Robinson CJ, Burton H. 1980b. Somatic submodality distribution withinthe second somatosensory (SII), 7b, retroinsular, postauditory, andgranular insular cortical areas of M. fascicularis. J Comp Neurol 192:93–108.

Robinson CJ, Burton H. 1980c. Somatotopographic organization in thesecond somatosensory area of M. fascicularis. J Comp Neurol 192:43–67.

Rockland KS, Ojima H. 2003. Multisensory convergence in calcarine visualareas in macaque monkey. Int J Psychophysiol 50:19–26.

Romanski LM, LeDoux JE. 1993. Information cascade from primary audi-tory cortex to the amygdala: corticocortical and corticoamygdaloid pro-jections of temporal cortex in the rat. Cereb Cortex 3:515–532.

Romanski LM, Clugnet MC, Bordi F, LeDoux JE. 1993. Somatosensory andauditory convergence in the lateral nucleus of the amygdala. BehavNeurosci 107:444–450.

Romanski LM, Bates JF, Goldman-Rakic PS. 1999a. Auditory belt andparabelt projections to the prefrontal cortex in the rhesus monkey.J Comp Neurol 403:141–157.

Romanski LM, Tian B, Fritz J, Mishkin M, Goldman-Rakic PS, Raus-checker JP. 1999b. Dual streams of auditory afferents target multipledomains in the primate prefrontal cortex. Nat Neurosci 2:1131–1136.

Rouiller EM, Simm GM, Villa AE, de Ribaupierre Y, de Ribaupierre F.1991. Auditory corticocortical interconnections in the cat: evidence forparallel and hierarchical arrangement of the auditory cortical areas.Exp Brain Res 86:483–505.

Sanides F, Krishnamurti A. 1967. Cytoarchitectonic subdivisions of senso-rimotor and prefrontal regions and of bordering insular and limbicfields in slow loris (Nycticebus coucang coucang). J Hirnforsch 9:225–252.

Schreiner CE. 1998. Spatial distribution of reponses to simple and complexsounds in the primary auditory cortex. Audiol Neurootol 3:104–122.

Schroeder CE, Foxe JJ. 2002. The timing and laminar profile of converginginputs to multisensory areas of the macaque neocortex. Brain Res CognBrain Res 14:187–198.

Schroeder CE, Lindsley RW, Specht C, Marcovici A, Smiley JF, Javitt DC.2001. Somatosensory input to auditory association cortex in the ma-caque monkey. J Neurophysiol 85:1322–1327.

Scott BH, Malone BJ, Semple MN. 2000. Physiological delineation ofprimary auditory cortex in the alert rhesus macaque. 2000 AbstractViewer/Itinerary Planner. Washington, DC: Society for Neuroscience.

Sweet RA, Dorph-Petersen KA, Lewis DA. 2005. Mapping auditory core,lateral belt, and parabelt cortices in the human superior temporalgyrus. J Comp Neurol 491:270–289.

Tian B, Rauschecker JP. 1994. Processing of frequency-modulated soundsin the cat’s anterior auditory field. J Neurophysiol 71:1959–1975.

Tian B, Rauschecker JP. 2004. Processing of frequency-modulated soundsin the lateral auditory belt cortex of the rhesus monkey. J Neurophysiol92:2993–3013.

Tian B, Reser D, Durham A, Kustov A, Rauschecker JP. 2001. Functionalspecialization in rhesus monkey auditory cortex. Science 292:290–293.

Turner BH, Mishkin M, Knapp M. 1980. Organization of the amydalopetalprojections from modality-specific cortical association areas in the mon-key. J Comp Neurol 191:515–543.

Van Hoesen GW, Pandya DN. 1975. Some connections of the entorhinal(area 28) and perirhinal (area 35) cortices of the rhesus monkey. I.Temporal lobe afferents. Brain Res 95:1–24.

Vogt C, Vogt O. 1919. Allgemeinere Ergebnisse unserer Hirnforschung.J Psych Neurol 24:279–462.

von Economo C, Koskinas G. 1925. Die Cytoarchitectonik der Hirnrindedes erwachsenen menschen. Berlin: Julius-Springer.

Wallace MN, Kitzes LM, Jones EG. 1991. Intrinsic inter- and intralaminarconnections and their relationship to the tonotopic map in cat primaryauditory cortex. Exp Brain Res 86:527–544.

Wang X, Kadia SC. 2001. Differential representation of species-specificprimate vocalizations in the auditory cortices of marmoset and cat.J Neurophysiol 86:2616–2620.

Wang X, Merzenich MM, Beitel R, Schreiner CE. 1995. Representation ofa species-specific vocalization in the primary auditory cortex of thecommon marmoset: temporal and spectral characteristics. J Neuro-physiol 74:2685–2706.

Werner-Reiss U, Kelly KA, Trause AS, Underhill AM, Groh JM. 2003. Eyeposition affects activity in primary auditory cortex of primates. CurrBiol 13:554–562.

Wong-Riley M. 1979. Changes in the visual system of monocularly suturedor enucleated cats demonstrable with cytochrome oxidase histochem-istry. Brain Res 171:11–28.

Yeterian EH, Pandya DN. 1998. Corticostriatal connections of the superiortemporal region in rhesus monkeys. J Comp Neurol 399:384–402.

Yukie M. 2002. Connections between the amygdala and auditory corticalareas in the macaque monkey. Neurosci Res 42:219–229.


cortical connections of the auditory cortex in marmoset monkeys: core and medial belt regions

Documents