Portal de Periódicos da CAPES

Cortical Function: A View from the Thalamus

2005; Cell Press; Volume: 45; Issue: 4 Linguagem: Inglês

10.1016/j.neuron.2005.01.035

1097-4199

Michele A. Basso, Daniel J. Uhlrich, Martha E. Bickford,

Functional Brain Connectivity Studies

Neuroscientists from across the country gathered at the University of Wisconsin, Madison in September to honor Ray Guillery and his seminal work on the thalamus. The meeting focused on three timely research topics, each of which inspired new thinking about thalamic function. Presentations on the organization and dynamic nature of thalamocortical pathways, the role of the thalamus in communication between cortical areas, and the relationship between sensory and motor pathways of the brain, including cognitive aspects of thalamocortical processing, made for lively discussions. The meeting revealed that communication between thalamus and cortex is so rich that we should no longer consider the operations of either structure separately from the other. Proceedings of the meeting will be published in Progress in Brain Research in 2005. In this report, we provide a general overview of the main themes of the meeting. Neuroscientists from across the country gathered at the University of Wisconsin, Madison in September to honor Ray Guillery and his seminal work on the thalamus. The meeting focused on three timely research topics, each of which inspired new thinking about thalamic function. Presentations on the organization and dynamic nature of thalamocortical pathways, the role of the thalamus in communication between cortical areas, and the relationship between sensory and motor pathways of the brain, including cognitive aspects of thalamocortical processing, made for lively discussions. The meeting revealed that communication between thalamus and cortex is so rich that we should no longer consider the operations of either structure separately from the other. Proceedings of the meeting will be published in Progress in Brain Research in 2005. In this report, we provide a general overview of the main themes of the meeting. On September 12–14, 2004, over 100 neuroscientists from across the country gathered at the University of Wisconsin, Madison to honor Ray Guillery and his seminal work on the functional organization of the thalamus. Organized by Murray Sherman and Vivien Casagrande, long-time friends and collaborators of Ray's, the meeting focused broadly on the dynamic interdependence of thalamus and cortex in perception and action. We know a considerable amount about the mechanisms of sensory organ function and basic mechanisms of motor control. With this background, neuroscience has made enormous strides in our understanding of sensory perception and movement based on a linear flow of information through thalamus to cortex. In this tradition, the thalamus is thought of as a pit stop, accumulating sensory information before shipping it off to cortex. The work presented at this meeting reflects a growing body of literature that challenges this traditional view in favor of an alternate view in which a complicated arrangement of neuronal connections organized in feedforward, feedback, and loop circuits participates in the dynamic regulation of information processing through the thalamus. As a result, even our assumptions about the linear processing of sensory information to guide action should be questioned. The meeting emphasized three areas, each of which inspire new thinking about thalamic function. The first was the organization and dynamic nature of thalamocortical pathways. The second was the role of the thalamus in communication between cortical areas. And the third was the relationship between sensory and motor pathways of the brain, including cognitive aspects of thalamocortical processing. The three-day meeting was lively and spirited throughout, exploring novel views of the importance of the thalamus. For interested readers, the proceedings of the meeting will be published in Progress in Brain Research in 2005. Here we review the main themes of the meeting. How do we become aware of our environment, and how does our brain sort the host of competing signals arising from our sensory receptors? Awareness of the world depends upon sensory information reaching the cerebral cortex. All sensory information (except olfactory) passes through the thalamus before it reaches the cortex. As a result, the thalamus can be expected to play an important role in conscious perception. However, the nature of this role is just beginning to be recognized. Historically, the thalamus was considered a simple relay of sensory information to the cortex. If this were true, why have a thalamus at all? Why not have the primary sensory afferents project directly to the cortex? A key point of the discussions at the meeting was that the thalamus is more than a passive relay. The thalamus controls the flow of information to cortex. Speakers such as Peter Ralston, Vivien Casagrande, Harvey Karten, Martin Deschênes, Barry Connors, Edward Callaway, Terrence Sejnowski, and Martin Usrey explored evidence for dynamic information processing through thalamus en route to cortex. Below, we review the basics of thalamocortical pathways and then review the presentations. Information from each of the sensory modalities travels along separate pathways through the thalamus to cortex. Within the thalamus of these pathways, there is specialized organization, reflecting the unique features of each modality. For example, the medial region of the ventral posterior nucleus (VPm) of the thalamus receives input from primary afferents that are connected to individual whiskers in the rodent. These afferents form a map of whisker space in the thalamic barreloids. In the visual system, the primary afferents from the retina terminate in the dorsal lateral geniculate nucleus of the thalamus (dLGN) where they form a map of visual space. Individual VPm and dLGN axons in turn project to the cortex, transmitting specific information about stimulation of the whiskers or retina, and in some cases, carry very specific information, such as that related to the wavelength of light. Although they are carrying specific sensory information, all thalamic neurons have features of signal processing in common. Perhaps most striking is that the input from the primary sensory afferents comprises only 5%–10% of the synapses onto thalamocortical neurons. The remaining 90%–95% of synapses arise from other locations, including the cortex, the brainstem, local interneurons, and the thalamic reticular nucleus (Sherman, 2001Sherman S.M. Tonic and burst firing: dual modes of thalamocortical relay.Trends Neurosci. 2001; 24: 122-126Abstract Full Text Full Text PDF PubMed Scopus (522) Google Scholar). The important point in this arrangement is that the receptive field properties of thalamic neurons are determined by the few primary afferent inputs rather than the other inputs. Because brain regions making up the many non-primary afferent inputs have been implicated in arousal and attention, a major focus of the discussions was that the thalamus may provide behaviorally relevant, dynamic control over the nature of the sensory information that is relayed to cortex. The challenge now is to identify the mechanisms of this control and their consequences for perception. Studies in the somatosensory system provided clear insight into the role GABAergic mechanisms play in the dynamic nature of processing within the thalamus. Three types of inhibitory influences on thalamocortical cells were discussed. In the first, the role of the intrinsic interneurons was explored. Studies of the pain pathways reveal that neurotoxic lesions of pain-transmitting afferents cause a dramatic downregulation of GABA expression in the interneurons of the lateral region of the ventral posterior thalamus (VPl), probably contributing to neuropathic pain. A second inhibitory pathway arises from the thalamic reticular nucleus. In the VPm, thalamic reticular neurons mediate recurrent inhibition as well as lateral inhibition of the thalamocortical neurons within the barreloids. Thalamic reticular neurons receive input from many regions of the cortex and brainstem. Additionally, thalamic reticular neurons are coupled chemically through synapses and electrically through gap junctions, which can mediate fast communication among local networks of thalamic reticular neurons. In the posterior thalamus, a third, inhibitory input arises from the zona incerta. This nucleus contains GABAergic neurons that may suppress primary afferent input that is not synchronized with movements of the whiskers. These observations strongly argue that inhibitory pathways contribute to the selection of information that is passed through the thalamus to the cortex. Discussions also focused on another common feature of thalamocortical cells that contributes to the dynamic transfer of information through the thalamus. Thalamocortical neurons can switch between two distinct modes of firing, called burst and tonic. These modes are contingent on the state of a voltage-dependent, low-threshold calcium channel. The burst mode predominates during sleep, and the tonic mode is more prevalent during waking. There once was controversy as to whether burst mode occurred at all during the waking state, but ample evidence for burst responses during waking was presented. Thus, much of the discussion addressed the function of burst firing during wakefulness. There were two important questions: first, what controls the firing mode of thalamocortical neurons? One accepted fact is that brainstem and cortical inputs control the mode of firing of thalamocortical neurons. An intriguing possibility was suggested that particular visual stimuli might be capable of switching the firing modes of thalamocortical neurons. The second important question was how do these modes transform the information conveyed to cortex? Addressing this, speakers discussed how burst and tonic mode affect signal content and how this might influence cortical neurons receiving these patterns of activity. It was shown that information in both spatial and temporal domains is conveyed differently by burst and tonic firing. For example, a neuron in tonic mode will accurately report the presence and persistence of a visual stimulus, whereas a neuron in burst mode will not. A goal for the future will be to understand what information is conveyed by the two modes and how it is interpreted by cortex. One scenario was proposed. In burst mode, groups of action potentials occur after periods of quiescence and are better at evoking responses in cortical neurons than are action potentials occurring in tonic mode. These observations led to the hypothesis that the burst itself may provide a unique signal, such as a "wake-up call" to cortex that something important has occurred. After this wake-up call, stimulus features can be conveyed in tonic mode. The studies described above provide strong support for the dynamic nature of information processing in the thalamus. Yet, the perceptual correlates of these dynamics are poorly understood, as few sensory recordings have been obtained in the thalamus of awake animals. One such experiment was presented at the meeting. It was shown that neuronal activity in the dLGN is enhanced at the end of a rapid eye movement. This enhancement might reflect attention to the spatial location to which the eyes move; something that would be important for scanning visual scenes and relevant for cognition. Experiments such as these provide an exciting direction to continue exploration of the dynamic nature of thalamic function. Speakers also discussed the possible function of alternate sensory pathways through the thalamus. One was described in birds, whose nucleus rotundus is innervated by the tectum. Neurons of the tectum in turn are innervated by multiple retinal ganglion neurons. Similar pathways innervate portions of the mammalian pulvinar nucleus (a "higher-order" thalamic nucleus discussed below). Because of the enormous receptive fields of rotundus and pulvinar neurons, this anatomical arrangement appears to sacrifice precise spatial information but may enhance motion detection. An analogous organization may occur in the posterior thalamus of the rat. This thalamic region receives afferents from trigeminal neurons that are activated by the deflection of multiple whiskers. In combination with inputs from the zona incerta, it was suggested that this arrangement may function to code signals related to whisker movement. In sum, there are commonalities across sensory thalamic nuclei despite the fact that these neurons encode very different information. Moreover, the processing within the thalamus is dynamic rather than stationary, and this type of information processing exists across sensory sectors of thalamus. The classic view of perception is corticocentric, with information traveling from the thalamus to the cortex, where sensory signals are processed leading to perception. A radically new view was discussed at the meeting, in which the role of the thalamus does not end after sensory signals are transmitted to the cortex. Rather, the thalamus itself may play a critical role in transmitting information between cortical areas or in coordinating the activity of different cortical areas. Presentations by Larry Abbott, Carol Colby, David Van Essen, Murray Sherman, and Harvey Swadlow explored the mutually dependent nature of cortical and thalamic activities. A recurring theme of these talks was the extent to which connections within the thalamus and cortex can be considered "drivers" or "modulators." Using the dLGN as a model, Sherman and Guillery, 1998Sherman S.M. Guillery R.W. On the actions that one nerve cell can have on another: Distinguishing "drivers" from "modulators".Proc. Natl. Acad. Sci. USA. 1998; 95: 7121-7126Crossref PubMed Scopus (467) Google Scholar described the driver inputs as those that define fundamental receptive field properties of thalamic cells (e.g., the retinal input to the dLGN), even though they constitute only a small percentage of the synaptic contacts on thalamic neurons. The term "modulator" was used to describe inputs that, although more numerous, influence thalamic activity in more subtle ways (e.g., layer 6 cortical inputs or cholinergic brainstem inputs). These definitions have provided a useful theoretical framework for discussions regarding information flow in sensory processing, and several new ideas utilizing these definitions were presented. First, the suggestion that some thalamic nuclei receive their driving input from the cortex (i.e., inputs originating from layer 5) introduced the idea that the cortex may strongly influence the activity of thalamic nuclei. Thalamic regions innervated by cortical drivers could in turn relay these signals to other cortical regions via ascending thalamocortical projections (Figure 1). Such an arrangement implies that the thalamus may be more actively involved in corticocortical communication than previously realized. In particular, the pulvinar nucleus was discussed as an especially intriguing thalamic region that could function to route signals from one cortical region to another. The impact of such cortico-thalamo-cortical loops was discussed in the context of cortical hierarchies defined by direct corticocortical connections. Since direct corticocortical projections far outnumber the layer 5 projections to the pulvinar nucleus, it was suggested, alternatively, that the pulvinar may not be the primary route for the transfer of sensory signals from one cortical area to another. Rather, it may serve to coordinate activity within the distributed cortical network, perhaps as a function of attention. A corollary of this discussion was an exploration of the extent to which thalamocortical inputs and corticocortical inputs may be considered drivers or modulators. In other words, is cortical activity defined primarily by corticocortical connections or thalamocortical connections? It was pointed out that thalamocortical terminals and thalamic drivers share a number of morphological and physiological features. Of particular interest is the fact that retinogeniculate and thalamocortical terminals exhibit a frequency-dependent depression. The outcome of this attribute is that thalamic spikes are more likely to evoke cortical spikes when they are preceded by long interspike intervals, a characteristic of the burst mode. Thus, the firing mode of thalamic neurons determines the efficacy of thalamocortical transfer. It was also proposed that the high spontaneous activity of the cortex supports a model in which drivers and modulators are not distinguished anatomically. In this scenario, drivers are defined as opposing excitatory and inhibitory inputs that can directly influence firing rate, whereas modulators are defined as balanced configurations of excitatory and inhibitory inputs that adjust the gain or sensitivity of cortical neurons. This model emphasizes the dynamic nature of corticocortical and thalamocortical connections, with individual inputs rapidly switching from driver to modulator status dependent on changes in their correlated inhibitory inputs. The dynamic nature of the dialogue between cortical and subcortical pathways was also emphasized in presentations of higher-level sensory processing related to attention and perception. Studies of pathways involved in the initiation and guidance of saccadic eye movements indicate a partnership of cortical and subcortical pathways rapidly exchanging information. Experiments using split-brain monkeys indicate that cortical and subcortical pathways collaborate to preserve a stable representation of visual space and show an extraordinary capacity for reorganization. Implicit in all these discussions was the recognition that to understand the function of the cortex or thalamus, neither can be considered in isolation. Perceptions obviously guide our movements so that we may interact with our environment. But, does what we do influence what we perceive? This question was examined by considering the relationship between sensory and motor pathways, particularly the processing of visual information used for the generation of saccadic eye movements. Presentations by Peter Schiller, John Reynolds, Mriganka Sur, Michael Paradiso, Melvyn Goodale, Ray Guillery, and Robert Wurtz revealed evidence for both serial and parallel processing of visual information to motor centers as well as the role of cognitive factors in defining the properties of neurons within the thalamocortical pathways. A guiding force for this topic was the classic visual hierarchy model, proposed by Felleman and Van Essen, 1991Felleman D.J. Van Essen D.C. Distributed hierarchical processing in primate cerebral cortex.Cereb. Cortex. 1991; 1: 1-47Crossref PubMed Scopus (5140) Google Scholar, in which the visual system is organized as a set of increasingly complex regions arranged in a feedforward hierarchy, culminating in the passage of information to motor centers to control movement. In this serially organized model, motor actions are carried out after sensory processing is completed. For example, reflexive saccades, made to briefly appearing stimuli, can be initiated by visual signals from cortical area V1 to superior colliculus. However, when given a choice among visual stimuli, the signals from V1 proceed through higher cortical areas to frontal cortex and then to the brainstem to initiate a particular eye movement. Indeed, choosing one stimulus from among many possible stimuli appears to involve competition between neuronal resources, and this competition may operate at multiple levels within the processing hierarchy. Superimposed on the serial arrangement is a parallel architecture. Neuropsychological experiments combined with fMRI on clinical populations reveal how information within cortical regions is processed independently for perception and action. Patients with damage to regions of the ventral visual cortex are unable to recognize visual objects but are nevertheless able to orient parts of their body accurately to interact with objects physically. Anatomical evidence presented also supports a parallel process for perception and action. Most primary sensory axons projecting to thalamus also branch to project to motor centers. Similarly, layer 5 cortical axons projecting back to the thalamus branch to innervate brainstem motor structures (Figure 1). Thus, a topic of discussion was that sensory and motor information could be conveyed by the same axons. Rather than conceiving of a system that processes sensory information, is directly linked to perception, and guides action, it was suggested that our actions can influence sensory signals and, thus, our perceptions. The idea that response properties of V1 neurons are in some linear and obvious way related to physical stimuli was challenged in two presentations. In one, if the relative contrast of a luminance patch and its surround were manipulated so that the surrounding light either decreased or increased, V1 neurons responded differently even though the patch of light stimulating the receptive field never changed. In the second presentation, an identical visual stimulus was presented in different conditions. In one condition, the stimulus indicated with a high likelihood that the subject should make an eye movement to it. In the second condition, the same visual stimulus was associated with a low likelihood that an eye movement would be required. V1 neurons responded preferentially for the condition in which an eye movement would be required. Thus, in both of these presentations it was clear that even at the very initial stages of visual processing, in V1, the neurons are not strictly coupled to the physical stimulus. Finally, the dependence of cortical areas on lower-level motor information was evident from the presentations on the control of saccadic eye movements. During fixation, when two visual stimuli are presented briefly in rapid succession, the locations of the stimuli are coded in retinal coordinates. Once an eye movement is made to acquire the first stimulus, determining the endpoint of the second saccade requires details about the initial eye movement. Such information could be provided to the frontal eye fields by the superior colliculus signals transmitted via the medial dorsal thalamus. Saccade-generating signals originating in the superior colliculus could be simultaneously transmitted to the brainstem and the medial dorsal thalamus (a corollary discharge). This information could then be combined with the information regarding the location of the second target to compute the metrics of the second saccade. This hypothesis was supported by the demonstration that inactivation of medial dorsal thalamus impaired the ability of monkeys to make an accurate saccade to the second stimulus. All of the presentations and discussions on the theme summarized here showed very compellingly that thalamocortical neurons are intimately involved in higher-order processing for perception and action. As Patricia Churchland pointed out, neuroscience is still in its infancy. Perhaps moving us out of this stage, this meeting revealed that our knowledge of the richness in communication between thalamus and cortex is now at a point where we can no longer consider the operations of either structure separately from the other. Along these lines, Sasha Nelson provided a view of an exciting future of discovering genetic markers that will identify neurons within thalamus and cortex as well as guide comparative studies of homologous neuronal classes. Everyone at the meeting agreed on a number of directions this field can move in. More studies of the anatomy and physiology of the pulvinar nucleus and its connections with cortex are needed. It was also clear that physiological investigations of thalamocortical, corticothalamic, and corticocortical pathways are needed, particularly in awake animals where perception can be related to action. Here it is important to explore the importance of functional processing loops rather than restrict our thinking to linear, hierarchal processing structures. It was also recognized that more complicated stimulus configurations would reveal the nuances of response properties that are not always evident in a paradigm in which single visual stimuli are presented in an otherwise dark room. Understanding the messages conveyed by thalamocortical pathways will be the real key. For example, the cortex must have a read-out mechanism for interpreting the very different patterns of neuronal activity it receives in the burst or tonic modes, and it must be able to interpret the timing of action potentials within spike trains of single neurons or across populations of neurons. Finally, many thanks to Dr. Ray Guillery who continues to move the field forward through his elegant experiments in the thalamocortical system and his thoughtful insights into the processes discussed here. A hearty thank you also to the University of Wisconsin - Madison, Murray Sherman, and Vivien Casagrande for giving us the opportunity to come to know Ray Guillery.