Jump to ContentJump to Main Navigation
Functional Connections of Cortical AreasA New View from the Thalamus$

S. Murray Sherman and Rainer W. Guillery

Print publication date: 2013

Print ISBN-13: 9780262019309

Published to MIT Press Scholarship Online: January 2014

DOI: 10.7551/mitpress/9780262019309.001.0001

Show Summary Details
Page of

PRINTED FROM MIT PRESS SCHOLARSHIP ONLINE (www.mitpress.universitypressscholarship.com). (c) Copyright The MIT Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in MITSO for personal use (for details see www.mitpress.universitypressscholarship.com/page/privacy-policy). Subscriber: null; date: 21 January 2019

Linking the Body and the World to the Thalamus

Linking the Body and the World to the Thalamus

(p.179) 7 Linking the Body and the World to the Thalamus
Functional Connections of Cortical Areas

S. Murray Sherman

R. W. Guillery

The MIT Press

Abstract and Keywords

This chapter and the next explore the pathways that link the thalamus and cortex to each other and to the rest of the brain, considering each in terms of the major points about the neural connections raised in earlier chapters. These connections provide the essential links between cortex, lower centers, action, and perception. Understanding the messages and knowing the functional nature of the relays is basic to understanding how the organism relates to the world. It is important to recognize that, although we raise the issue in a few instances only, this possible duality is relevant for all of the pathways that carry messages to the thalamus for relay to cortex.

Keywords:   thalamus, cortex, neural connections, duality, lower centers, action, perception

7.1 Introduction

This chapter and the next will be an exploration of the pathways that link the thalamus and cortex to each other and to the rest of the brain, considering each in terms of the major points about the neural connections raised in earlier chapters. These connections provide the essential links between cortex, lower centers, action, and perception. Chapter 4 considered the pathways as either drivers (Class 1) that transmit a message or modulators (Class 2) that change the way in which a message is or is not transmitted. The messages transmitted by the drivers need to be understood in terms of the informational content that they contribute to the computational functions of the neural stations they serve, and this will relate to events in the world or in other parts of the nervous system. The modulators (Class 2) need to be understood on the basis of how they affect the transmission of messages. Understanding the messages and knowing the functional nature of the relays is basic to understanding how the organism relates to the world.

In chapter 6, we showed that many of the axons carrying messages to thalamus for relay to cortex have branches that innervate motor centers so that the cortex receives a great many messages that have a dual significance, representing events in the body or the world on the one hand and instructions for upcoming movements on the other. It is important to recognize that, although we raise the issue in a few instances only, this possible duality is relevant for all of the pathways that carry messages to the thalamus for relay to cortex. It is relevant for first order thalamic relays, which relay messages from the sensory periphery or from lower (mamillary or cerebellar) centers, and also for higher order relays, which relay messages from one area of cortex to another.

Corticofugal axons from a great many and probably all cortical areas, including classical sensory areas and areas far beyond the areas that are labeled (p.180) motor in textbooks, carry motor outputs from layer 5 to lower centers with functions that are largely unknown and unexplored, and many of these outputs have branches to the thalamus. For a full view of the messages that thalamus sends to cortex, we need to recognize that there is this rich corticofugal output from all areas of cortex and that many of these outputs also send branches to the thalamus for relay to other areas of cortex (chapter 6). In this way, higher cortical areas can be informed about the outputs of lower areas. Each thalamic relay, whether first order or higher order, sends to the cortex messages that can have two meanings, one about the environment or about activity in other neural centers and the other about instructions currently on the way to motor centers.

We start by exploring the inputs to the thalamus that provide the cortex with its view of the world and of the body in relation to the world, then we look at the nature of the thalamocortical pathways that distribute this information to the cortex, and finally we consider the corticofugal pathways that provide the executive link between the cortex, the phylogenetically older lower motor centers and pattern generators (see chapter 6), the body, and the world. This is important not only for showing the extent to which any one cortical area can act directly on the lower centers but also for stressing that much of our behavior can depend on subcortical centers acting with minimal cortical controls. The question for understanding the role of cortex in movement control is not so much the classical question as to whether cortex controls muscles or movements, but one about the particular lower motor centers through which any cortical area can act.

The aim of chapters 7 and 8, as for most of this book, is to identify areas of ignorance as much as to demonstrate new facts or relationships and, although it may be no surprise that the former dominate the latter, we consider that an overall view of these connectivity patterns must form a basis for understanding cortical functions. The long-term aim is to understand how thalamus and cortex, or how any one thalamic nucleus and the cortical areas it innervates, relate to cognition and behavior, not only in terms of their neural connections but also in terms of the nature of the messages that are passing along these connections. This can then be related to information that is currently more readily available about the conditions under which a particular cortical area becomes active or about losses produced by silencing or removing one area. Such information is, at present, most commonly reported on the basis of functional imaging or lesion effects for particular cortical areas and produces a view of functional localization in particular cortical areas. This, however, has limited explanatory value. It provides information that can be regarded as “evidence-based phrenology”; it reveals nothing about the mechanisms concerned (p.181) or the nature of the messages involved. We need to learn how higher cortical functions relate to the particular messages that are arriving over defined pathways from other centers or from the peripheral sensory receptors and then to assess the nature of the cortical actions and their outputs. The following is an exploration of some thalamic relays that can serve as examples of the type of knowledge that eventually one should hope to obtain about all thalamic relays.

7.2 The Inputs to Thalamus

We start this review at the thalamic levels for which we have the most information, that is, with the classical primary sensory relays, and then use this information to move to other thalamic relays where more information is still needed.

7.2.1 The Inputs to the First Order Thalamic Relays The Inputs to the Major Classical Sensory Relays

We discussed the first order classical sensory relays in some detail in previous chapters (chapters 3 and 6). They provide examples of the information that needs to be obtained for any thalamic relay if the functions of the cortical area it supplies are to be understood. For the pathways to the primary sensory nuclei in the thalamus, we know the driver inputs and we can be reasonably clear about the messages that they carry about activities of the peripheral sensory receptors. This involves knowing first the nature of the sensory event and then knowing how this is converted to specific patterns of action potentials in the peripheral nerves or the ascending central pathways. For the somatosensory pathways, which we will use as a relatively simple example here, the early work of Adrian (1928) showed how particular sensory events are encoded in the peripheral nerves, and later studies, such as those of Mountcastle and Henneman (1952) provided evidence for the thalamus, where the activity was interpreted as representing the way in which the world is acting on the sensory receptors. However, as stressed in chapter 6, the activity must also be interpreted in terms of the actions that those same patterns of activity are currently about to produce through branches of the ascending axons innervating lower motor levels.

For inputs such as those passing toward the thalamic ventral posterior nucleus from the dorsal roots, the nature of the incoming sensory information from the receptors is clear: it may be muscle stretch or a nociceptive stimulus, and for these the neural code has been defined (see Adrian, 1928; Kandel et al., 2000). The instructions for future actions that these same incoming (p.182) axons are passing to the spinal circuits are also known to some extent and were discussed at the end of chapter 6. They provide information for the forward models that allow higher centers to anticipate forthcoming actions. That is, the information that is passed from the ventral posterior thalamic nucleus to the cortex must be expected to relate not only to the message produced by the changes in the sensory receptor but also to some aspects of upcoming movement patterns.

In addition to the branches that the incoming dorsal root axons send to the spinal cord, there are other branches given off at higher levels of the pathways to brain structures that include the superior colliculus, the periaqueductal gray, the midbrain reticular formation, the inferior olive, and the hypothalamus (summarized by Guillery and Sherman, 2002; Guillery, 2003; Sherman and Guillery, 2006). The details of the actions produced by such branches can all be considered as relevant for a full appreciation of the messages that the ventral posterior nucleus can pass to the somatosensory cortex. For example, the details of the inputs to the periaqueductal gray, which sends important descending pathways to spinal centers concerned with pain, will be of particular interest for a fuller appreciation of the thalamocortical involvement in sensory responses to nociceptive stimuli, whereas those to the hypothalamus, whose outputs innervate spinal visceral centers, will relate particularly to forthcoming autonomic responses. The functional role of these extrathalamic inputs will thus also be present in the information available to cortex. The cortical circuitry that receives these ascending messages will have information about the messages that are on their way to periaqueductal gray and hypothalamus. That is, they will have information about upcoming events in the nervous system, contributing to a view of the future that can play a crucial role in generating a model of the future needed in the brain of any organism before it can interact successfully with its environment.

Defining these relationships in the thalamus or cortex may prove elusive and will necessarily depend on studies of an unanesthetized preparation, but knowing that information about these spinal and brainstem events is necessarily available in thalamus and cortex should at the very least change the expectations of investigators and help to modify views of pure sensory functions in thalamus and cortex.

For the major sensory thalamic nuclei other than those receiving from the ascending spinal pathways, comparable relationships produced by branched incoming driver afferents arise, and we refer to chapter 6 for details of the connections on which this statement is based. We have stressed these conclusions from the previous chapter here, because they are essential not only for understanding the nature of the message that each primary sensory nucleus is (p.183) sending to the relevant cortical area(s) but also for understanding the functional organization of any thalamocortical pathway that is supplied by a branched axon; and it appears that most or all of them do have such branches (see chapter 6). The important point is that the functions of the extrathalamic branches are everywhere an essential part of understanding the nature of the messages that the thalamic nucleus is relaying to its cortical area(s). For example, knowing that the retinal inputs to the lateral geniculate nucleus have branches that innervate midbrain structures concerned with the control of gaze, of pupillary size, and of accommodation can serve as a clue for raising new questions about how visual perception can relate to action (Churchland et al., 1994; O'Regan and Noë, 2001) and can raise issues relevant for understanding how such links between actions and perception can be learned (Bompas and O'Regan, 2006a, b). The Inputs to the Anterior Thalamic Nuclei

We next consider the first order thalamic nuclei receiving inputs for which evidence of sensory functions are less obvious or lacking. These are the two large anterior thalamic nuclei receiving afferents from the mamillothalamic tract and the ventral lateral nucleus receiving afferents from the deep cerebellar nuclei. For the anterior thalamic nuclei, we showed in the previous chapter that the smallest, the anterior dorsal nucleus, is a relay on the pathway of vestibular messages about head position to the retrosplenial cortex and hippocampus (Taube, 2007). This part of the pathway through the anterior thalamus is comparable to the first sensory level. It raises comparable outstanding questions about the extent to which the messages about head position also relate to messages to motor centers and about their origin either from branches at the entrance of the vestibular nerve to the medulla or from the descending branches of the mamil-lotegmental tract to the medial pontine reticular formation; these were raised in chapter 6. However, we have no comparable information about the pathways to the limbic cortex through the two larger anterior ventral and anterior medial thalamic nuclei (figure 7.1), and current views about the functions of these pathways often depend on what is known about the functions of the cortical recipient areas, the anterior and posterior limbic areas, rather than information about the nature of the messages that the thalamic cells receive and send on to cortex. We consider the known evidence about the inputs to these nuclei in some detail because this illustrates the importance not only of understanding the nature of the messages that any one thalamic nucleus is relaying to cortex but also of knowing whether the input is a driver or a modulator. It also demonstrates the distinction between localizing particular functions to a cortical area (the “evidence-based phrenology” mentioned in section 7.1) as opposed (p.184)

Linking the Body and the World to the Thalamus

Figure 7.1 Schematic representation of the connections of the anterior thalamic nuclei. AD. AM. and AV are the anterior dorsal, anterior medial, and anterior ventral thalamic nuclei, respectively: LMN is the lateral mamillary nucleus; MMN the medial mamillary nucleus. The small arrows indicate regions where pathways that may not be made up of branching axons are shown as branched in order to reduce the number of lines in the figure; they may or may not be branched. X shows mamilloteg-mental branches that have not been traced to their midbrain terminal zones. The known driver pathways discussed in the text through the mamillary nuclei and the anterior thalamic nuclei are shown as thicker lines, whereas all of the other pathways shown as thin lines are not defined as drivers or modulators nor do we know about the messages that they are carrying except for the fornix inputs to the anterior thalamic nuclei, which are most reasonably regarded as modulators on the basis of the appearance of their terminals (see Somogyi et al.. 1978). n., nucleus.

to knowing the functions of the connections and the nature of the messages that are transmitted: the difference between functional localization and comprehension of the neural machinery.

The ascending inputs to the two larger anterior nuclei come from the medial mamillary nucleus and travel in the mamillothalamic tract (figure 7.1), which is the bundle that also carries inputs for relay to the anterior dorsal nucleus from the lateral mamillary nucleus. The three anterior nuclei are often treated as a group, which may not be justified. Not only are the cells in the anterior dorsal nucleus noticeably larger than those in the other two nuclei, but also they relate to the large-celled lateral mamillary input with thicker axons, and there may thus be significant functional differences. Further, the large cells in (p.185) the small lateral mamillary nucleus receive their driver inputs from a midbrain source that appears not to be shared by the smaller cells in the medial mamillary nucleus.

The mamillary bodies receive inputs from two sources (figure 7.1). One is the mamillary peduncle, which brings afferents from midbrain nuclei, and the other is the postcommissural fornix, which brings afferents from the hippocampal formation (the subiculum). The midbrain afferents to the medial mamillary nucleus come from the deep tegmental nucleus, whose functions are currently undefined (see Saunders et al., 2012); those that go to the lateral mamillary nucleus come from the dorsal tegmental nucleus, bringing vestibular information to the head position pathway (Taube, 2007).1 Although significant numbers of GABAergic neurons2 have been reported in these two nuclei (see Allen and Hopkins, 1989; Wirtshafter and Stratford, 1993; Gonzalo-Ruiz et al., 1999), there can be little doubt that the dorsal tegmental nucleus has a driver input to the lateral mamillary nucleus, because the transfer of information about head direction is known to be passed from the vestibular nerves to the anterior dorsal thalamic nucleus, first through the dorsal tegmental nucleus, then through the lateral mamillary nucleus, and then from the anterior dorsal thalamic nucleus to the retrosplenial cortex (Taube, 2007), and the inputs to the anterior dorsal nucleus have also been defined as Class 1, or driver (Petrof and Sherman, 2009). It would be reasonable to expect that the axons from the deep tegmental nucleus to the medial mamillary nucleus would also be acting as drivers on the route to cortex, like those from the dorsal tegmental nucleus to the lateral mamillary nucleus, but there is at present no experimental support for this view, and it is more common to consider the hippocampal input that reaches the mamillary bodies in the postcommissural fornix as the driver input to the mamillary bodies (Aggleton et al., 2010). These authors provide a detailed review and analysis of the pathways that link the hippocampus, the mamillary bodies, the anterior thalamus, and the midbrain nuclei, and they describe the memory losses that can be produced by lesions in these pathways. However, they do not raise any questions about the nature of the messages that are being passed along the pathways. On the basis of current knowledge, it is not possible to follow the message or identify the drivers to the medial mamillary nucleus in a way comparable to Taube's studies for the lateral mamillary nucleus. We lack the relevant information.

(p.186) The morphological evidence suggests that the fornix (but not the mamil-lothalamic tract) provides a modulatory input to the anterior thalamus (Somogyi et al., 1978). It could be argued that if the fornix provides modulators for the anterior thalamus, it might also provide modulators for the mamillary bodies, but that is a weak argument. In the first place, the mamillary bodies are not a part of the thalamus: they have a separate developmental history, and the rules that apply to thalamic nuclei may not apply to the mamillary bodies. Further, Wright et al. (2010) have shown that the thalamic and the mamillary inputs come from two distinct cell groups in the subiculum (see figure 7.1). In addition, the axon terminals from the fornix that have been described in the mamillary bodies (Allen and Hopkins, 1989) are somewhat larger and contain rather more mitochondria per terminal than the terminals from cortical layer 6 that are generally seen in the thalamus, and these larger terminals are features associated with Class 1 (driver) inputs at other sites (Covic and Sherman, 2011; Viaene et al., 2011b). These terminals also appear to make multiple synaptic contacts, another feature that is not expected for the layer 6 inputs to thalamus and is more like the appearance expected from a Class 1 input (see chapter 5).

The conclusion is that the drivers providing the inputs to the cells of the medial mamillary nucleus need to be defined. We need to learn about the details of their actions in terms that allow a distinction between Class 1 (driver) and Class 2 (modulator) inputs (see chapter 4), and, as for the lateral mamillary nucleus, we need to learn about the nature of the messages that the medial mamillary nuclei pass to the anterior thalamus and define where those messages come from. They may come from the fornix or from the mamillary peduncle, or from both. The mamillary peduncle would be likely to bring a link with lower centers and the world, but the nature of that input needs to be defined on the basis of the properties of cells in the deep tegmental nucleus. If the fornix provides the drivers, then this would probably be a link with established memory circuits. If both turn out to be drivers, then there would be an interesting question about exactly how the two relate to each other in the medial mamillary nucleus.

There is one further related unknown in these pathways, and that concerns the action of the fornix input on their target cells in the lateral mamillary nucleus. A reasonable guess would treat these as modulators of the head direction relay in the lateral mamillary nucleus, comparable to the input from the fornix to the anterior thalamic nuclei, and acting rather like layer 6 inputs to thalamic nuclei. However, at present this is an unknown. They could provide a second driver input.

We have explored the anterior thalamic nuclei in some detail, because they serve to illustrate two important questions. One is about the relation of the (p.187) thalamic relay to the lower centers and the world, a question that is answered for the anterior dorsal nucleus by the demonstration of head direction cells that depend on vestibular inputs, shown as thick lines in figure 7.1. That is, we know where the messages that are relayed to cortex are coming from and we know about the nature of these messages. The second question concerns the identification of drivers and modulators in the pathways. The transfer of information about head direction through the anterior dorsal nucleus also tells us about the drivers in this part of the pathways (thick lines), except that they leave open the possibility of a second driver input to the lateral mamillary nucleus from the postcommissural fornix (thin lines). In contrast to this, our current knowledge of the pathways through the medial mamillary system and the two large anterior thalamic nuclei does not provide clear information about which inputs are drivers and which are modulators, demonstrates no clear functional link with the world, and at present provides no information about the messages that are being passed along the pathways to the cortex; they are all shown as thin lines in figure 7.1.

Whereas in the previous section (, we stressed the need for information about the function of the extrathalamic branches of the afferents, in this section we have focused on the need to distinguish the drivers from the modulators and then to identify the message that the drivers carry. From this point of view, the anterior thalamic nuclei provide an example that can be usefully compared with the evidence discussed for the dorsal medial geniculate nucleus and ventral anterior thalamic nucleus in chapter 4.2 and 4.3. However, whereas for those nuclei the functional properties of the relevant pathways have been identified, for the mamillothalamic pathways that information still remains largely undefined for the two largest nuclei of the group. It should be noted that, in the mamillary pathways, the branching patterns are also important and that their functions are still unexplored (see chapter 6).

One alternative approach for elucidating the nature of the messages that are relayed by a thalamic nucleus to the cortex is to look at the functions of the cortical area(s) that receive the thalamic inputs and then use this information to gain insights into the messages carried by their inputs. For the anterior medial thalamic nucleus, studies of the anterior limbic cortex can suggest a range of possible functions. On the basis of human and animal experiments and of imaging and postmortem studies, the functions or functional losses that have been reported for the anterior limbic cortex relate to autism (Simms et al., 2009), geriatric depression (Gunning et al., 2009), borderline personality disorder (Whittle et al., 2009), psychosis (Fornito et al., 2008), and to mechanisms for predicting aversive events and terminating fear (Hayes and Northoff, 2011; Steenland et al., 2012) or evaluating the cost of foraging (Kolling et al., 2012). This cortical area has also been described as serving functions relating (p.188) to spatial working memory in rats (Mendez-Lopez et al., 2009), has been proposed as an area that relates to emotional processing in general (Etkin et al., 2011), as a mechanism supporting the selection and maintenance of learned options (Holroyd and Yeung, 2012); it has also been proposed as an area that relates to the production of a natural smile, as opposed to a smile posed for a photographer (Damasio, 2006). This is a list from a larger set of claims for the anterior limbic cortex, and a comparable list might be prepared for the posterior limbic cortex.

None of the items in this list is readily related to anything known about the mamillothalamic pathways except that degeneration of the mamillothalamic tract, fornix, or mamillary bodies is associated with memory losses (Aggleton et al., 2010) and that a transient Korsakoff's syndrome has been reported after a bilateral removal of the anterior limbic area (Brodal, 1981). That is, the listed functions may all be related in different ways to memory functions, and that would suggest that the fornix afferents coming from the hippocampus are the drivers in the mamillary bodies on the basis of the lost memory functions and probably modulators in the anterior thalamus on the basis of the fine structural evidence (Somogyi et al., 1978) (see figure 7.1). But at present this is a guess. We have no good evidence to come to any clear conclusions about the nature of the messages that the two larger anterior thalamic nuclei transmit to cortex or receive from the midbrain. We do not know whether the medial mamillary part of the system has any links with the world through the deep tegmental nucleus, and in terms of the activity within the pathways, we know nothing about the messages that are being transmitted. The contrast between the experimental and clinical reports for the anterior medial nucleus and the anterior limbic cortex on the one hand and the experimental reports for the anterior dorsal nucleus and the retrosplenial cortex on the other represents an important conceptual difference in approaches to study of the cortex. The former is an attempt to localize functions to cortical areas without concern for the thalamic inputs, or for any inputs, which may be bringing messages relevant for those functions to the cortex, whereas the latter involves tracing the driver inputs to thalamus and understanding the messages carried in those inputs from the body and the world through several defined relay stations. The Inputs to the Ventral Lateral Nucleus

The ventral lateral thalamic nucleus is included here, because it is another first order thalamic relay nucleus whose inputs do not come directly from sensory pathways but from the cerebellum, whose links to the world are complex. They need to be defined if we wish to understand the nature of the message that the cortex is receiving from the thalamus. The projection of the ventral lateral nucleus to the motor cortex (p.189) has often played a more important role in interpreting the nature of the thalamocortical message than has the cerebellar input. That is, the cerebellar input can be interpreted as the route for cerebellar influences on the motor cortex, often without specific questions being raised about the nature of the information sent to the cortex. However, it may well be that the message(s) can be understood on the basis of what the cerebellum sends to the thalamus rather than on the basis of the functions of the motor cortex itself: as already indicated, looking back from cortex to thalamus for a view of what the thalamus is transmitting to cortex may be more difficult than tracing the message the other way.

The inputs to the ventral lateral nucleus come from the deep cerebellar nuclei, mainly the dentate nucleus and also to some extent from the interpositus nucleus (Hoover and Strick, 1999; Teune et al., 2000). However, these inputs to the thalamus represent only a small part of the outputs from these cerebellar cell groups, which also send a rich innervation to many parts of the brainstem, including the red nucleus and the pontine reticular formation as well as more caudal parts of the brainstem concerned with motor control. The extent to which the thalamic inputs come from branches of these brainstem afferents is largely unknown, although we saw in chapter 6 that the Golgi method demonstrates a rich pattern of cerebellofugal axons giving rise to thalamic inputs with one branch and to brainstem centers with the other. One issue that will be important for understanding the message(s) that the ventral lateral nucleus passes to the motor cortex is to learn, in the first place, which of the cerebello-thalamic axons has branches to brainstem and, in the second place, to define these termination sites and the actions at those brainstem centers.

The inputs to the relevant deep cerebellar nuclei come from the lateral cerebellar hemisphere, and it has been shown that one of the functions of this input to the ventral lateral nucleus is a role in guiding limb movements to distant, visual targets (Stein and Glickstein, 1992). The cerebellar discharges related to visually guided movements precede activity in motor cortex by about 20 ms (Thach, 1975; Liu et al., 2003). On the basis of this delay, it has been suggested that the visual trigger for the movements passes from the parietal cortex to the cerebellum and is then passed through the thalamic relay to the motor cortex.3 Miall and King (2008) have proposed that this part of the cerebellum produces a “forward model” that predicts the actions that will be produced by currently active motor commands. They consider that “The (p.190) cerebellum receives ascending proprioceptive inputs and efferent copies of descending motor commands, and it outputs to cortical and brainstem motor nuclei.” The forward model is thus seen here as created in the cerebellum by messages coming in corticopontocerebellar pathways, and from ascending axons from the spinal cord that are likely to be efference copies as well (chapter 6.2).4 That is, efference copies are established in the cerebellum before the branched cerebellothalamic axons (see figure 6.6) send one branch to brainstem motor centers and another to the thalamus. Two points may be relevant for this schema. One is that that there are many possible ways of feeding copies of motor instructions into the pathways that eventually reach the thalamus and themselves represent copies of motor instructions as can be seen for the cerebellothalamic pathways that give off motor branches before entering the thalamus (figure 6.6). Another is that there is a rich transthalamic corticocortical pathway through the ventral anterior thalamic nucleus from and to motor and premotor areas in the frontal cortex and that the corticothalamic driver branches from layer 5 in this link are all likely to be branches of long descending corticofugal axons, including corticopontine axons. The cortico-pontocerebellar pathway may thus represent but one small part of the circuitry that is providing a forward model to the frontal cortex; the transthalamic corticocortical pathway through the ventral anterior nucleus may provide another and more direct link for visually guided movements. However, the extent to which the final smooth movement is based on components added by the cerebellum is largely unexplored in terms of the several other pathways that are likely to be relevant. These include, on the input side for vision, the retinal afferents to the tectum and pretectum as well as those to the cortex via the lateral geniculate nucleus, and for the arm, the spinocerebellar and the lemniscal pathways through the ventral posterior nucleus to the somatosensory cortex. In addition, there are the links established between vision and the arm movement not only in the parietal cortex but also in the superior colliculus and in brainstem motor centers, including the red nucleus. The importance of the tectum is easily ignored once we forget that frogs can catch flies; the projections from the cortex to the superior colliculus (Harting et al., 1992) and the tectopontine projections (see Schwarz et al., 2005) will also be relevant. We introduce these several pathways here, because, as indicated in chapter 6, several of the corticotectal and corticopontine axons also send branches to (p.191) higher order thalamic relays, and these are considered in the next section. The interim conclusion is that there are a great many pathways that are able to provide information about instructions for upcoming movements and contribute to forward models. The thalamocortical pathways are everywhere carrying efference copies, and contributions to forward models are available in many different pathways. Once a functional role is accepted for the branching axons, efference copies that can contribute to forward models will be seen as extremely common in all parts of the central nervous system.

7.2.2 The Inputs to the Higher Order Thalamic Relays

Our treatment of the higher order nuclei can be brief, primarily summarizing the limited knowledge about what amounts to the largest and still most mysterious parts of the primate thalamus. Higher order relays are defined as those that receive their driving inputs from layer 5 of cortex (see chapters 3 and 5). In several higher order thalamic nuclei, there is a possibility that first and higher order relays are present (see chapter 5), and to allow for this possibility, when we speak of first order and higher order nuclei, instead of relays, we regard first order nuclei as lacking corticothalamic drivers, whereas higher order nuclei have them. So far as the nature, driver or modulator, Class 1 or Class 2, of the corticothalamic axons is concerned, as indicated in chapters 3 and 5, in the thalamus we regard as drivers all of the axons that form large, excitatory terminals making many synaptic junctions often not close to astrocytic processes and often forming serial synaptic junctions (triads) and that have the physiological properties of Class 1 inputs as defined in chapter 4. Where an origin from cortical layer 5 rather than layer 6 is known, this strengthens the interpretation. However, for most higher order relays, this interpretation is still tentative, awaiting detailed information either about the nature of the message and its transmittal to cortex or about the properties of the transmission at the thalamic relay that would identify the input as Class 1 or Class 2 (or other). We need to ask questions (generally expecting few answers yet) about the nature of the messages that these layer 5 axons are carrying, and we expect that the message that reaches the thalamus will be relayed to cortex unless blocked at the thalamic gate (see chapter 5). We showed in chapter 3 that there are higher order relays that transmit visual messages through the lateral posterior nucleus or the pulvinar in rodents, cats, or monkeys, that higher order somatosensory messages are relayed through the posterior medial nucleus of the mouse and higher order auditory messages through the dorsal medial geniculate nucleus of the mouse, but for most of the other higher order nuclei, such as the lateral dorsal, medial dorsal, and central (p.192) medial nuclei, we have little or no information about the nature of the message that is relayed, nor can we be certain that a message is relayed, although the possibility that some thalamic nuclei transmit no message to cortex seems extremely unlikely. (We exclude the thalamic reticular nucleus, because it is part of ventral rather than dorsal thalamus; see chapter 3.2.)

Where a higher order pathway has been identified, information about the nature of the message that is passed to the thalamus may be gleaned from evidence about the activity in layer 5 of the cortical area from which these axons arise or, where a descending branch of the corticothalamic axon has been identified, from studies of the functional relationships established by these axons at their terminal sites. However, these are questions for future experiments; they have been explored to only a limited extent. We can take the layer 5 cells in area 17 as an example. They send axons to the superior colliculus with branches to the lateral posterior nucleus or pulvinar. They have complex receptive fields and are mainly binocular. Their axons terminate in the most superficial layers of the superior colliculus, far from the major motor outputs, which arise in the deeper layers of the colliculus (Harting et al., 1992). This suggests that these axons are likely to play a minor role in defining the output of the colliculus. However, in the superficial layers they overlap with the terminals of corticotectal axons from many higher visual and other cortical areas (including the frontal eye fields) that also have deeper terminals and that are likely to play a more significant role in the collicular control of gaze. These many visual inputs into a phylogenetically old center for the control of movements (Grillner et al., 2007) suggest that the inputs from layer 5 of area 17 are likely to make a minor contribution to the control of gaze, and that a message about that contribution will be passed from the higher order thalamic relay (pulvinar or lateral posterior nucleus) to higher visual areas, contributing to the relevant forward model. The observation that minimal localized stimulation of the deep layers of area 17 produces ocular movements (Tehovnik et al., 2003) lends some support to such a view.5

It should be stressed that we currently have very limited information about the extent to which thalamocortical pathways are drivers (Class 1) or modulators (Class 2); see chapters 4 and 5. As indicated in those chapters, the evidence that some thalamocortical relays are class 2 and thus not drivers transmitting messages raises major questions about many of the transthalamic cortical connections (p.193) considered here and in the following sections with the exception of those where there is clear evidence that a defined message is transmitted from the thalamus to the cortex or where the thalamic inputs have been identified on the basis of their morphological criteria as drivers, or functionally as either Class 1 or Class 2 (see chapter 4). At present, these are all pathways that are transmitting messages about the classical sensory inputs to the brain. There is an urgent need for information about which of the many remaining transthalamic pathways are relaying messages to the cortex and also for identification of the nature of the message. This is an important point to bear in mind in the rest of the book, but one that will not be repeatedly raised.

7.3 Outstanding Questions

  1. 1. For the ascending and cortical afferents to thalamus, the thalamocortical pathways, or the corticofugal axons to motor centers, which are drivers (Class 1) and which are modulators (Class 2) for each group of axons?

  2. 2. What exactly can we say about the messages carried in any of the pathway groups in question 1 apart from those that involve first order inputs to sensory nuclei? For example, what are messages that the ventral lateral nucleus is sending to the motor cortex or the anterior medial nucleus is sending to anterior limbic cortex?

  3. 3. Where we can identify the structures innervated by the nonthalamic branch of a driver input to thalamus, can we use information about the nature of the structures innervated by that branch, and their actions in response to stimulation of those branches, to provide clues about the nature of the message that the thalamic branches bring to the thalamus?

  4. 4. Where a cortical area has some descending layer 5 axons that send a branch to the thalamus and some that do not, what other functional characteristics distinguish these two types of axon arising from the same area of cortex?

  5. 5. Where the functions of a cortical area have been defined on the basis of losses produced by lesions or of conditions that produce activation of that area, is there a strategy that will allow us to trace the thalamic input to those functions back to a thalamic relay and its afferents?



(1) . It is important to record that at least some of the axons in the mamillary peduncle, not included in figure 7.1. branch to supply the lateral mamillary nucleus with one branch and the medial mamillary nucleus with the other (Cajal, 1911).

(2) . Which, as noted in chapter 4.2. are not suited to act as drivers.

(3) . It may be important to recognize that a transthalamic connection from parietal cortex to motor cortex may prove to be more direct (see chapter 5).

(4) . Not only do the axons that supply Clarke's column (which gives origin to the spinocerebellar tract) form as branches that are likely to be innervating other spinal centers, but also the spinocerebellar axons themselves give off branches in the caudal medulla before they enter the inferior cerebellar peduncle (Cajal. 1911).

(5) . The extent to which the strength of the motor outputs increases for higher levels of the corti-cortical hierarchies and may perhaps match the increase in the frequency with which forward receptive fields can be observed in the visual hierarchy merits investigation (Melcher and Colby, 2008).