Cortical network for representing the teeth and tongue in primates



Sensory information from the tongue and teeth is used to evaluate and distinguish food and nonfood items in the mouth, reject some and masticate and swallow others. While it is known that primates have a complex array of 10 or more somatosensory areas that contribute to the analysis of sensory information from the hand, less is known about what cortical areas are involved in processing information from receptors of the tongue and teeth. The tongue contains taste receptors, as well as mechanoreceptors. Afferents from taste receptors and mechanoreceptors of the tongue access different ascending systems in the brainstem. However, it is uncertain how these two sources of information are processed in cortex. Here the parts of somatosensory areas 3b, 3a, and presumptive 1 that represent the mechanoreceptors of the teeth and tongue are identified, and evidence is presented that the representations of the tongue also get information from the taste nucleus of the thalamus, VPMpc. As areas 3b, 3a, and 1 project to other areas of somatosensory cortex, and those areas to additional areas, some or all of the currently defined somatosensory areas of cortex may be involved in processing gustatory, as well as tactile, information from the tongue and thus have a role in the biologically important function of evaluating food in the mouth. © 2006 Wiley-Liss, Inc.

Sensory inputs from the tongue and teeth are obviously very important to the health and survival of most mammals. Based on inputs from receptors in the tongue and periodontal receptors around the teeth, distinctions are made between food and nonfood objects in the mouth, how much pressure can be delivered via the teeth on such items in the mouth, and how to guide tongue and jaw movements in food processing (Jacobs et al.,1998). For judgments about the desirability of food objects, taste information is integrated with somatosensory information on texture, compliance, and item size. Thus, inputs from taste receptors and mechanoreceptors in the mouth, in part, must be evaluated together in order to facilitate the basic function of eating. But where and how is this done? As neocortex is obviously important in the processing of sensory information, how are sensory systems, which are related to sensory afferents from the tongue and teeth, organized at the cortical level? Given the obvious importance of this issue, it may come as a surprise that little is known about the anatomical framework in cortex for processing information, and most of this understanding is focused on the cortical processing of taste. Because the system for processing somatosensory information at the cortical level has been well studied in monkeys, and many of the recent studies of cortical structures related to taste have been in monkeys, this review is focused on these primates. We argue that it is likely that the rather extensive cortical network for processing tactile information from the hand in monkeys provides an attractive model for how information from the teeth and tongue might be processed. We also provide evidence that inputs from tactile and taste receptors are integrated at the earliest levels of cortical processing and are processed to a large extent in the same cortical structures. First, we briefly review what is known about the cortical system devoted to processing information from the hand and then outline current portrayals of the cortical system that mediates taste. Next, we review recent studies that indicate how the teeth and tongue are represented in primary somatosensory cortex (area 3b) and in adjoining parts of area 3a and presumptive area 1. Finally, we describe inputs from the somatosensory thalamus to these representations, and how these representations project to other subdivisions of cortex. While more research is needed to achieve a comprehensive understanding of the cortical network devoted to the tongue and teeth receptors, an overview of the early stages of this network is starting to emerge.


Most of the research on the organization of somatosensory cortex in primates has focused on New World owl monkeys, because of certain technical advantages, and Old World macaque monkeys, because they are more closely related to humans. Here we discuss features of somatosensory cortex organization that were more easily determined in owl monkeys, but likely apply to most or all other primates. The most important technical advantage offered by owl monkeys is that they have no central sulcus (Fig. 1). Thus, areas of anterior parietal cortex are directly accessible for experimental study as they are exposed on the dorsolateral surface of the brain, rather than being extensively buried in the central sulcus as in macaques and humans.

Figure 1.

A lateral view of the brain of an owl monkey with some of the somatosensory, motor, visual, and auditory areas indicated. Area 3b in primates corresponds to primary somatosensory cortex of other mammals. This area represents the body from foot to hand in a mediolateral sequence, and then the face, teeth, tongue, teeth, and tongue again in a caudodorsal to rostroventral sequence. Histologically, this representation is discontinuous, with ovals of cortex representing different parts of the face (see text), the contralateral teeth, the contralateral tongue, the ipsilateral teeth, and the ipsilateral tongue. In a parallel fashion, cortex adjoining these ovals in area 3a and presumptive area 1 represents the tongue and teeth as illustrated. Other somatosensory areas that are largely or completely in the lateral sulcus (opened for viewing in this illustration) include the second somatosensory area (S2), the parietal ventral area (PV), caudal and rostral divisions of the ventral somatosensory field (VSc and VSr), and the parietal rostral area (PR). While the somatotopic organization, if any, of PR is unknown, the other four areas all contain systematic representations of cutaneous and other body receptors. Representations of the tongue and teeth have not been identified in these areas, but they can be presumed to exist. Thus, there could be many representations of the tongue and teeth that are involved in the important process of guiding the selection and processing of food. Premotor areas, in turn, likely receive some of this processed information to guide tongue and jaw movements in these behaviors. Here, primary motor cortex (M1), dorsal and ventral premotor areas (PMD and PMV), the supplementary motor area (SMA), and the frontal eye field (FEF) are indicated. Primary (V1) and secondary (V2) visual areas and primary (A1), rostral (R), rostrotemporal (RT), and caudomedial (CM) auditory fields, as well as the auditory (Aud) belt and parabelt, are outlined for reference. In owl monkeys, the most ventral part of area 3b and adjoining field is on the ventral surface, where it is shown as a folded-out flap.

There have been several advances in our understanding of the organization of somatosensory cortex that relate to the present issue of identifying networks limited to teeth and tongue. First, the puzzle has been resolved of why primary somatosensory cortex of monkeys includes four distinct architectonic fields, area 3a, 3b, 1, and 2 of Brodmann (1909). Rather than these fields being a part of a single primary area, S1, as originally claimed, each of these fields contains a separate representation of body receptors (Merzenich et al.,1978), and each is a functionally distinct area (for review, see Kaas,2004). Only area 3b corresponds to S1 as described in most nonprimate mammals (Kaas,1983), and area 3b receives activating inputs from tactile mechanoreceptors via the ventroposterior (VP) nucleus of the somatosensory thalamus. Area 1 constitutes a second level of processing tactile information as it is relayed from area 3b, although area 1 also receives direct thalamic inputs from VP. Area 1 relays to area 2, as a third level of processing, where tactile information is mixed with proprioceptive information relayed from the ventroposterior superior nucleus of the thalamus (VPS). Area 3a receives proprioceptive information from VPS as well. Other connections, such as those between areas 3b and 3a, and areas 2 and 3a, enrich the processing in these areas. In addition, all four areas project to two somatosensory representations in the lateral sulcus, the long-known second somatosensory area, S2, and the more recently discovered parietal ventral area, PV (Krubitzer et al.,1986). Each of these areas forms a separate representation of the body surface, and they in turn activate the adjoining ventral somatosensory area, VS, which now appears to contain rostral and caudal parts with separate representations (Coq et al.,2004), and the less well-understood parietal rostral area, PR. Areas S2, PV, PR, and VS also have connections with other more poorly understood portions of cortex of the lateral sulcus, somatosensory regions of the rostral half of posterior parietal cortex, and motor and premotor areas of the frontal lobe.

The main point of this brief review is that studies of somatosensory cortex in monkeys have revealed the early stages of a complex processing system, one that includes at least 9 or 10 areas in anterior and lateral parietal cortex alone, as well as additional fields in posterior parietal, frontal, and even cingulate cortex (for review, see Kaas,2004). The general assumption is that each station in this complex array of interconnected areas includes some neurons responsive to any given part of the body, but this is largely uncertain, especially for such parts of the body that are seldom stimulated in microelectrode mapping studies of somatosensory areas. Because it is more difficult to stimulate structures in the mouth, there is not much evidence of how periodontal and tongue receptors are represented in this extensive somatosensory network of cortical areas. Yet a reasonable assumption is that cortical areas usually contain complete representations, and that all or most of the known somatosensory fields process information from the teeth and tongue, as well as from the rest of the body. To what extent is this assumption supportable, and to what extent is it at odds with experimental results? To start to answer this question, we need to review briefly current views on the organization of the cortical system for gustation.


In contrast to the complexity of the somatosensory cortical network outlined above, the cortical system proposed for taste is relatively simple. The current view is that the cortical gustatory system consists of a tongue representation in S1, a large primary gustatory area (G) in rostral cortex of the lateral sulcus, and a more rostral hedonic (H) taste region in orbitofrontal cortex (Fig. 2), with possibly one or two intermediate areas in between (for review, see Pritchard and Norgren,2004). Early evidence for this scheme came from cortical recordings during electrical stimulation of the tongue in squirrel monkeys (Benjamin and Burton,1968; Benjamin et al.,1968). Two regions of cortex were activated, one was attributed to rostrolateral S1, and the other, on the rostral tip of the cortex of the upper bank of the lateral sulcus, became known as area G, the proposed primary taste region (Sanides,1968). The source of taste information to the cortex was known from other studies (e.g., Blomquist et al.,1962) to be from a thalamic nucleus known as the parvocellular division of the ventral posterior medial nucleus or complex, VPMpc (VPM proper and VPL are subdivisions of the ventroposterior nucleus, VP, that respectively represent the head or body; VPMpc is also known as the basal ventral medial nucleus, VMb). Thalamocortical connections were determined at that time by making cortical lesions and seeing where neurons degenerated in the thalamus. As a large lesion of cortex, involving both the S1 tongue region and opercular-insular cortex of the region G, was needed in order to produce extensive degeneration of neurons in VPMpc, Benjamin and Burton (1968) concluded that both S1 and region G received inputs from VPMpc and were involved in taste perception. Later, Pritchard et al. (1986) supported this supposition when an injection of a tracer in VPMpc labeled both regions, but G more densely than S1. However, the significance of the label transported to S1 has been recently questioned, as the injected tracer may have partially involved VPM proper, thereby labeling projections to S1 related to tactile sensations, rather than taste (Pritchard and Norgren,2004). Thus, region G was considered to be the only area with inputs from the thalamic taste nucleus. A third cortical location that is clearly involved in taste has been described more recently after recordings in orbital frontal cortex (OFC) in macaque monkeys encountered neurons responsive to taste substances as well as other stimuli (e.g., Rolls et al.,1989,1990; Rolls and Baylis,1994; Rolls,2000). Neurons in OFC are thought to reflect the hedonic or pleasurable aspects of taste (for reviews, see Kringelbach,2004; Pritchard and Norgren,2004). Evidence from differences in the response properties of neurons in the region suggested to Rolls (2000) that the caudolateral part of the orbitofrontal region is secondary taste cortex, while rostrally adjoining cortex constitutes a third level of processing that is more multisensory. As a further complication, it is not clear how taste information would get from region G to OFC taste cortex. Pritchard and Norgren (2004) suggest that inputs may be indirect, involving one or two intervening areas (see also Sewards and Sewards,2001). Nevertheless, taste processing in cortex would seem to involve only a few stages, while the processing of information from tactile mechanoreceptors would seem to involve a much more elaborate network of interconnected areas.

Figure 2.

The locations of proposed taste areas of cortex in New World monkeys. In earlier studies, anterior parietal cortex was presumed to contain the single representation, the primary area (S1), and a large lateral portion of that S1 was activated by electrical stimulation of the tongue. In addition, another region in cortex of the rostral end of the lateral sulcus responded to electrical stimulation of the tongue (see text) and this region became known as primary gustatory cortex or area G. Later research identified a region of orbitofrontal cortex (OBF) in macaque monkeys where neurons were responsive to taste. This higher-order taste region, sometimes referred to as a hedonistic taste area (H), is shown here as a flap of cortex unfolded from the ventral surface of the frontal lobe. The taste nucleus of the thalamus, the parvocellular ventral posterior medial nucleus (VPMpc), projects to both the S1 tongue and the G regions. The G region is thought to relay taste information to OBF. As direct connections are sparse, this is thought to involve one or two additional relay areas. Note the differences in the interpretation of cortical organizations in Figures 1 and 2.

In the next section, we reevaluate and reconsider the evidence and suggest a different scheme, not only for processing information from the tongue, but also from the important periodontal receptors for the teeth. We do this by considering both published and incompletely reported observations.


Early microelectrode mapping studies of the somatotopy of area 3b in New World owl monkeys (Merzenich et al.,1978) and squirrel monkeys (Sur et al.,1982) and Old World macaque monkeys (Nelson et al.,1980) did not disclose the full extent of the field because the ventrolateral portion of area 3b, where the tongue and teeth are represented, is less accessible, and because there was less interest in the representation of the oral structures. In addition, it was never very certain how far area 3b extended ventrolaterally and rostrally, as histological distinctions are best made in the most favorable plane of section, and the sagittal or near sagittal planes that best revealed the boundaries of dorsomedial area 3b were not suitable for ventrolateral 3b. However, later it became apparent that the histological boundaries of area 3b were quite visible in appropriately stained sections cut parallel to the cortical surface after cortex had been artificially flattened so that dorsomedial and ventrolateral portions of area 3b were forced into the same plane (Tootell et al.,1985). Such brain sections in owl and squirrel monkeys revealed that area 3b extends much further rostrally than ever suspected, and that the myelin-dense architecture of area 3b contains a number of interruptions that relate to somatotopy (Jain et al.,2001). First, there is a long mediolateral band of dense myelination that extends from near the medial wall to near the lateral fissure. This band constitutes the portion of 3b that represents the body from tail to hand in a mediolateral sequence that has been described in great detail (Merzenich et al.,1978; Sur et al.,1982). A narrow myelin-poor septum cuts across the lateral margin of this band to separate the representation of the body from that of the face and oral skin (Fig. 3C). Less noticeable septa separate representations of other body parts in the part of 3b devoted to the body as well. Most notably, thin septa separate territories for the representation of each digit of the hand in cortex just medial to the hand-face junction (Jain et al.,1998). These isolating septa are also apparent in area 3b of macaque monkeys, where they can be identified as early as 2 weeks after birth (Qi and Kaas,2004). Just ventrolateral to the hand-face septum, three partially separated ovals of densely myelinated cortex represent different parts of the face (Fig. 3). We have numbered the caudorostral sequence of face-related ovals F1, F2, and F3, with F1 corresponding to the representation of the upper face, F2 to the region of the upper lip, and F3 to the lower lip, chin, and lower face. The ovals are present in owl monkeys (Fig. 3A), squirrel monkeys (Fig. 3B), and marmosets (data not shown). These face ovals are also present in prosimian galagos, where they are less obvious (Fig. 4), and they have been partially described in macaque monkeys, where difficulties in flattening the lateral part of area 3b confounded a complete description (Qi and Kaas,2004). Rostroventral to the three face ovals, a series of myelin-dense ovals in owl and squirrel monkeys represent receptors in the oral cavity, largely those of the tongue and the teeth (Fig. 3). In a caudorostral sequence, we have numbered these ovals O1, O2, O3, and O4. Ovals 3 and 4 are sometimes fused. Microelectrode recordings in these ovals indicate that O1 represents the upper and lower contralateral teeth. Oval O2 responds to touch on the tongue and, to a lesser extent, hard palate. Much of the oval is devoted to the contralateral tongue, but a more rostral part is devoted to the ipsilateral tongue. O3 responds to ipsilateral teeth, while O4 responds to ipsilateral tongue (for a somewhat different interpretation of the representation of the face and oral structures in squirrel monkeys, see Manger et al.,1995). A similar arrangement appears to exist in prosimian galagos (Fig. 4), where recordings have been obtained only from the more accessible O1 and O2 ovals. The cheek pouch, an important specialization of the oral cavity in macaques for short-term food storage, is well represented in cortex just anterior to the tongue representation (Manger et al.,1996). The more ventral region where the ipsilateral tongue and the ipsilateral teeth are expected to be represented has not been explored in macaques (O3 and O4).

Figure 3.

The locations of face and mouth ovals of area 3b in (A) owl monkeys and (B) squirrel monkeys. The ovals represent different parts of the face (F1, F2, F3). The O1, O2, O3, and O4 ovals for the oral cavity successively represent the contralateral teeth, the contralateral tongue, the ipsilateral teeth, and the ipsilateral tongue as shown in Figure 1. Note that the ovals extend onto the ventral surface of the frontal lobe in owl monkeys, but not squirrel monkeys. C: Part of a brain section from unfolded and flattened cortex cut parallel to the surface and stained for myelin to show the F and O ovals of area 3b in an owl monkey. Holes in the hand representation and the F2, F3, O1, and O2 ovals are microlesions placed to identify recording sites (see Jain et al.,2001). The hand is represented in cortex just above the F1–F3 ovals.

Figure 4.

The locations of face and oral cavity ovals in a prosimian primate, Galago garnetti (also known as Otolemur garnetti). The drawing above shows in black the medial portion of area 3b representing the body, the F1–F3 ovals for parts of the face, and the O1–O4 ovals for teeth and tongue. In a myelin-stained brain section cut parallel to the brain surface, some of the boundaries of these ovals are apparent. The presence of these ovals and their responsiveness to touch and taps on the teeth and tongue suggests that the face and tongue are represented in a highly similar manner in both prosimian and simian primates.

In summary, New World monkeys appear to have four myelin-dense ovals in ventrolateral area 3b that form a caudorostral sequence successively representing contralateral teeth, contralateral and ipsilateral tongue, ipsilateral teeth, and ipsilateral tongue. At least some aspects of this arrangement, the O1 and O2 ovals for teeth and tongue, are also found in prosimian galagos and Old World macaques. Thus, they may be basic features of somatosensory cortex in primates.

The recordings in squirrel and owl monkeys (Jain et al.,2001), as well as those more recently in galagos (data not shown), revealed additional representations of the tongue and teeth in cortex adjoining area 3b (Fig. 1). In brief, a narrow strip of cortex along the dorsorostral border of the slanted array of 3b ovals paralleled the 3b ovals in somatotopic organization, although this cortex was much less myelinated and had no ovals or other architectonic correlates of the representations of teeth and tongue. Thus, neurons dorsorostral to O1 were activated by taps on the teeth, those dorsorostral to O2 responded to touch on the tongue, those adjoining O3 responded to the ipsilateral teeth, and the few neurons recorded next to O4 responded to the ipsilateral tongue. This cortex has been architectonically defined in squirrel monkeys as area 3a (Sanides,1968). Other parts of area 3a respond to deep receptors involved in proprioception, largely muscle spindle receptors, via inputs from the ventroposterior superior nucleus, and touch via area 3b inputs (e.g., Krubitzer et al.,2004; for review, see Kaas,2004). In macaque monkeys, the tongue representation in area 3b (corresponding to 02) is also bordered rostrally by a tongue representation in area 3a (Krubitzer et al.,2004). In prosimian galagos, tongue and teeth ovals of area 3b (Fig. 4) are also bordered dorsorostrally by cortex responsive to stimulation of the tongue and teeth (data not shown). Thus, most or all primates appear to have tongue and teeth representations in area 3a that adjoin those in 3b. When the thalamic connections of these 3a tongue and teeth representations become known, we expect them to be with the ventroposterior superior nucleus, consistent with other parts of area 3a. Possibly, the tongue representation in 3a will also receive inputs from VPMpc.

A strip of cortex just ventral to the area 3b ovals also contains neurons responsive to touch and taps on the teeth and tongue in a pattern that parallels that of the 01–04 ovals. Thus, neurons in cortex just ventral to 01 respond to taps on the contralateral teeth and so on (Jain et al.,2001). This cortex could be an extension of area 1, but this is uncertain as areas S2 and PV appear to adjoin the face ovals of area 3b in a manner that would leave little room for area 1 (e.g., Krubitzer and Kaas,1990; Qi et al.,2002; Coq et al.,2004). However, this is difficult to determine electrophysiologically, as face representations in S2 and PV are represented near or next to the face ovals in area 3b, and a band of area 1 separating area 3b from S2 and PV would also be expected to respond to touch on the face. Thus, it is difficult to assign this cortex with certainty to an area 1 or S2 plus PV. In either case, the cortex ventral to the tongue and teeth representation in area 3b also responds to tongue and teeth, much as one would expect from a ventrorostral extension of area 1. In squirrel monkeys, this cortex was previously found to be responsive to tongue and teeth and was assigned to area 1 (Cusick et al.,1986). In macaque monkeys, cortex caudal to the tongue and teeth representations in area 3b also respond to the tongue or teeth, and this more caudal responsive zone was considered to be area 1 (Toda and Taoka,2002). Krubitzer et al. (1995) also described representations of the teeth and tongue in area 1 of macaque monkeys, while also finding teeth and tongue representations in PV and teeth in S2.

In summary, there is solid evidence for three representations of the teeth and tongue in monkeys, one in area 3b, another in area 3a, and a third in presumptive area 1. As areas S2, PV, and VS also represent the body surface, including the head, teeth and tongue representations are expected in these fields as well. There is yet little evidence for these additional representations of teeth and tongue, except from Krubitzer et al. (1995). However, neurons activated by tapping the teeth have been reported for the head portion of area VS and for a small portion of cortex caudal to S2 in owl monkeys (Cusick et al.,1989; for related observations on macaque monkeys, see also Ogawa et al.,1989; Ito and Ogawa,1994).

A remaining question is how do the three parallel teeth and tongue representations relate to the portion of S1 that was activated by electrical stimulation of the nerves of the tongue by Benjamin et al. (1968)? In brief, the activations recorded with brain surface electrodes included the region of the O2 tongue oval in area 3b, the adjoining cortex of area 3a and presumptive area 1, and perhaps a bit more (compare Figs. 1 and 2). The O4 oval for ipsilateral tongue was not included, as it was apparently out of the explored section of cortex. Benjamin et al. (1968) did not attribute the zone of activation as evidence for three representations of the tongue, as areas 3a, 3b, 1, and 2 were considered at that time as architectonic variations within a single representation, S1 (see Kaas,1983). Benjamin et al. (1968) considered the S1 tongue region as potentially involved in taste, although lesions of the region did not alter taste thresholds or cause retrograde degeneration in the thalamic taste nucleus, VPMpc. Now we turn to the issue of the connections of area 3b tongue and teeth ovals.


In owl monkeys, squirrel monkeys, marmosets, and, more recently, galagos, we have been able to inject tracers in and around O1 (teeth) and O2 (tongue) ovals and demonstrate connections with the thalamus and with other parts of cortex. The results are incompletely analyzed and only reported in the form of an abstract (Iyengar et al.,2002). Yet a few conclusions are well supported. First, injections of tracers that were largely or completely confined to the O1 oval representing teeth labeled neurons in a very medial part of the medial division of the ventral posterior nucleus, VPM. This location corresponds closely with the location in VPM of the teeth representation determined in microelectrode recording experiments in squirrel monkeys (Kaas et al.,1984). Part of this teeth representation responds to the ipsilateral teeth (Bombardieri et al.,1975), indicating that there is a thalamic source for the representation of ipsilateral teeth in O1 and again in O3. The label in VPM of VP indicates that the O1 teeth oval gets input from the same nucleus as the rest of area 3b, in support of the contention based on architecture and cortical recordings that the O1 oval is part of area 3b. The relay from VP to area 3b is one of low-threshold mechanoreceptors, and the periodontal receptors for the teeth respond to light pressure and taps on the teeth (van Steeberghe,1979). As one would expect, no neurons in the parvocellular VPM nucleus, VPMpc, which is the thalamic taste nucleus, were labeled by injections in the O1 oval representing the teeth. Second, injections in the O2 oval for the tongue labeled neurons largely in VPMpc, with other neurons in the most medial section of VPM. VPMpc can be recognized in New World monkeys as a most medial part of the ventral posterior complex that has smaller neurons than VPM, stains less densely for cytochrome oxidase, and expresses more parvalbumin and less calbindin (calcium-binding proteins). These differences in histochemical properties are even more marked in macaque monkeys, where VPMpc is largely ventral rather than medial to VPM (Jones and Hendry,1989). In galagos, VPM and VPMpc are also easily recognized in most preparations (Fig. 5). VPMpc is largely medial to VPM, and it can be identified by a greater expression of parvalbumin.

Figure 5.

The architectonic features of the taste nucleus, VPMpc, of the thalamus of prosimian galagos. The thalamus has been cut in the coronal plane, and sections processed for different histological markers. In standard Nissl preparations (middle panel), the ventroposterior nucleus, representing mechanoreceptors of the skin, stands out as a nucleus of large, darkly stained neurons. Cell-poor septa divide the nucleus according to the representation of body parts. Thus, a less distinct, narrow zone on the lateral margin of the nucleus represents the tail, and hindlimb (foot) representation is in the adjacent lateral cortex. More medially, two obvious dorsoventral septa mark the borders of the forelimb (hand) representation. These representations constitute the lateral division, VPL, of the nucleus. The larger medial division, VPM, represents the face and oral skin in a series of modules that are also outlined by septa, with the most medial module representing the teeth and tongue. Medial to this module, and after a cell-poor transition zone, neurons of VPMpc are apparent. In galagos, the cells in VPMpc are not notably smaller or less darkly stained than those in VPM. Subdivisions of VP and VPMpc are also apparent in sections stained for cytochrome oxidase (top panel). Typically, VPMpc expresses less cytochrome oxidase than VPM. In sections processed for the calcium-binding protein, calbindin, VPL, and VPM express very little calbindin, while VPMpc expresses a moderate amount (bottom panel). This clear difference also characterizes a difference between VPM and VPMpc in other primates. The adjoining nuclei, the ventroposterior inferior nucleus, VPI, and the ventroposterior superior nucleus, VPS, also express moderate amounts of calbindin. In each panel, an arrow points to VPMpc. In the top panel, dorsal (D) and medial (M) are indicated.

The evidence that VPMpc projects to the O2 oval of area 3b implicates this part of area 3b in gustatory functions, but this is somewhat uncertain. The VPMpc region is where Blomquist et al. (1962) activated neurons, both ipsilaterally and contralaterally, via electrical stimulation of nerves of the tongue in squirrel monkeys. This identified VPMpc as representing the tongue and quite likely representing taste afferents from the tongue. In macaque monkeys, VPMpc has been shown to receive direct inputs from the nucleus of the solitary tract in the brainstem (Pritchard et al.,1986), which receives taste afferents from the tongue (Pritchard and Norgren,2004). However, VPMpc in macaques has neurons that respond to taste and neurons that respond to touch (Pritchard et al.,1989). Thus, the projections to area 3b could provide both taste and touch information or touch only. There seem to be no recordings from neurons in the area 3b tongue representation while taste stimuli were delivered. Nevertheless, the involvement of area 3b in both touch and taste on the tongue would allow an integration of touch and taste information at the level of primary sensory cortex. At the very least, the questioned evidence of Benjamin and Burton (1968) and Pritchard et al. (1986) for projections from VPMpc to area 3b or S1 has been fully supported by our present results.

The anatomical sources of activation of neurons in ventrorostral parts of area 3a and presumptive area 1 via stimulating the tongue and teeth are not fully established, but one likely route is through connections with area 3b. Injections in the O1 teeth oval in monkeys and galagos labeled adjoining regions of cortex, including dorsally adjoining area 3a and ventrally adjoining area 1. Similarly, injections in the O2 tongue oval of area 3b labeled adjoining parts of areas 3a and 1. Another possible source is direct input from the thalamus. Injections of tracers into the presumptive area 1 tongue representation labeled neurons in VP, as expected for area 1 injections, and neurons in VPMpc as well. These direct thalamic inputs may modulate the responses of neurons in presumptive area 1, as above threshold activation of area 1 appears to depend on area 3b inputs rather than thalamic inputs, at least for the hand representation (Garraghty et al.,1990).

Other more distant connections of the teeth and tongue ovals have not been fully determined, but area 3b projects to both S2 and PV (Krubitzer and Kaas,1990; Qi et al.,2002; Coq et al.,2004) and area 3a projects to primary motor cortex (Huerta and Pons,1990). If similar patterns of connections are maintained for all parts of area 3b and 3a, we expect PV and S2 to be involved in processing information from the tongue and teeth, and distributing this information further to such areas as VSr, VSc, and PR, and motor cortex (Fig. 1). Thus, the area 3b ovals for teeth and tongue are likely to be the start of a vast cortical network for processing at least tactile information from these mouth structures, and quite possibly taste information as well.

While we have started to investigate the connections of the O1 and O2 ovals in area 3b, the connections of the O3 oval for ipsilateral teeth and of the O4 oval for ipsilateral tongue are totally unknown. They did not seem to have connections with their counterparts in the same hemisphere, the O1 and O2 ovals, and they did not have callosal connections with the O1 and O2 ovals. Instead, the callosal connections of the O1 and O2 ovals were focused on homologous ovals and adjoining parts of area 3a and presumptive 1.

By position, the O3 and O4 ovals appear to be near the rostral tip or slightly beyond area 3b as previously defined by Sanides (1968) in squirrel monkeys. This position is very close to orbitofrontal cortex that is involved in taste perception, and these ovals could provide a direct source of taste information.


The approximate location of area G in owl monkeys, based on the description of Sanides (1968) in squirrel monkeys, is shown in Figure 2. Benjamin and Burton (1968) concluded that VPMpc projects to both area G and S1 because a large lesion of cortex that included both the S1 tongue representation (the O2 portion of area 3b and adjoining cortex in Fig. 1) and area G (the proximity of the two areas makes it easy to include both in a single lesion) resulted in the retrograde degeneration of neurons in VPMpc, while a lesion of either zone alone resulted in no clear degeneration. The hypothesis was that neurons in VPMpc had branching axons to both targets, and lesions of either alone would allow cell survival via the remaining branch. Presumably, a bilateral lesion of S1 tongue plus G would remove all taste inputs to cortex, but strangely, Benjamin and Burton (1968) found no change in tests of taste thresholds after such lesions.

In macaque monkeys, Pritchard et al. (1986) were able to restrict injections of tracers mostly to VPMpc, and they described transported tracers in ventrolateral S1 and in the lateral sulcus near the tip of the insula that by position appear to correspond to area G. While the projections of VPMpc to S1 were later questioned (Pritchard and Norgren,2004) as possibly resulting from the spread of the injected tracer into VPM, our results support the earlier interpretation that VPMpc projects to S1. Injection of tracers in the region of G labeled neurons in VPMpc (Pritchard et al.,1986), further demonstrating that projections exist from VPMpc to cortex in the region of G. As Benjamin and coworkers found in squirrel monkeys, Ogawa et al. (1985) reported that electrical stimulation of nerves of the tongue activated cortex in the region of G (as well as in ventrolateral area 3b) in macaque monkeys. Thus, it seems likely that the same region has been identified as G in both primates.

While G is widely considered to be primary gustatory cortex, it does not have the pronounced histological characteristics of primary somatosensory, auditory, or visual cortex (Sanides,1968). Instead, G resembles more closely such secondary sensory areas as S2 and PV. By location, G may partially overlap other previously, but poorly, defined somatosensory areas, especially the parietal rostral area, PR, known primarily because of connections with PV and S2 (e.g., Qi et al.,2002). Another surprise, at least in comparison with other primary sensory areas, is that G of macaques has been described as a rather large area where rather few neurons respond to taste (2–10%), while more respond to touch or movements of the mouth (e.g., Smith-Swintosky et al.,1991; Plata-Salamán et al.,1993; Scott et al.,1999). In part, this may reflect a peculiarity of the gustatory system, as the small percentage of taste-specific neurons in gustatory cortex of rats increased as qualifying criteria were broadened (Katz et al.,2001). Yet even neurons responsive to touch on the tongue appear to be rare and scattered over a large region of cortex in the lateral sulcus (Ogawa et al.,1989; Ito and Ogawa,1994). A possible interpretation of such results is that the region of G in macaques and perhaps other primates is not a single area, but rather recordings have been from a number of adjoining and nearby areas, each with a portion devoted to processing tactile and/or taste information from the tongue, while much of each area is devoted to other sensory inputs from other parts of the mouth, face, and body, or from other modalities such as temperature or pain. The suggestion here is that VPMpc projects to tongue ovals in area 3b and to one or more areas in the lateral sulcus, much as VPM projects to S1 (3b), S2, and possibly PV in most mammals (for review, see Garraghty et al.,1991). Such parallel pathways characterize many thalamocortical pathways, and they now appear to exist in thalamic gustatory pathways as well. If modules for gustatory processing were smaller than area G and scattered within and across areas, then few recording sites would be appropriately placed in the territory of area G, accounting for the infrequency of gustatory neurons in G, and the large size of G.

Orbitofrontal cortex, as studied in macaques, also presents a complex picture in that neurons responsive to taste substances are scattered over a large region, and neurons that respond to taste often also respond to odors, the texture of fat, or even visual stimuli (for review, see Pritchard et al.,2005). Because neurons are also influenced by the reward value of stimuli (for review, see Rolls,2000), the region has also been considered to be part of a hedonic system for linking food and other stimuli to hedonic or rewarding experiences (Kringelbach,2004). However, the sparseness of taste neurons in orbitofrontal cortex may be a problem of not recording in the right place. Pritchard et al. (2005) recently reported that in a 12 mm2 area of medial orbitofrontal cortex, nearly 20% of the neurons responded to taste stimuli. Thus, this medial zone seems to be a multisensory area (or areas) that is more focused on taste than is the caudolateral orbitofrontal cortex, where most of the recordings related to taste have been obtained. How this region gets its taste information is not clear, as direct inputs from the region of area G are at best sparse (e.g., Carmichael and Price,1995). For this reason, projections from G to either caudolateral or medial orbitofrontal taste cortex are presumed to be largely indirect (Pritchard and Norgren,2004). But how would taste inputs get concentrated from a small percentage in G to 20% in the medial orbitofrontal taste region? Again, it seems that the full richness of a distributed, perhaps multisensory, cortical taste system should be considered. Possibly the 02 and the 04 ovals that represent the tongue in area 3b are involved in taste as well as touch, and they could provide another source of taste information to orbitofrontal cortex. In addition, if smaller taste-responsive modules are distributed across several areas in the lateral sulcus, as suggested above, then converging projections from these modules could focus responsiveness to taste stimuli. What is missing is clear evidence for intermediate fields that are responsive to taste stimuli and are capable of relaying taste information from the region of area G to medial orbitofrontal cortex. Connection patterns have suggested that this relay involves dysgranular and agranular fields of the insula (Pritchard and Norgren,2004), but dysgranular and agranular fields are not usually thought of as involved in relays of relatively unprocessed sensory information. This suggests that other pathways, possibly those emanating from area 3b ovals, may be involved.