Adaptive behavior requires evaluation of environmental stimuli with respect to their behavioral significance. Our working hypothesis holds that the Dopamine (DA) and acetylcholine (ACh), acting in the striatum (the input stage of the basal ganglia), emit signals that can aid such an evaluation, and thus can be characterized as the teachers or critics of the basal ganglia. The cortex-striatum-pallidum would be the actor of this learning network.
We recorded from both divisions of the basal ganglia network in monkeys before (while performing a probabilistic delayed response task) and after induction of severe Parkinsonism by systemic MPTP treatment. While the dopaminergic response reflects the mismatch between expectation and outcome (reward prediction error), cholinergic neurons respond in the same manner to events with different level of reward prediction error. We therefore hold that the dopamine response is indicative of the predictive value of various events in relation to reward. The cholinergic signal, on the other hand, may provide a temporal frame, defining the time period in which the dopamine signal will be processed.
The spiking activity of the actor part (pallidal and SNr neurons) is modulated by events belonging to different domains of behavior (limbic, cognitive and motor). Moreover, the spiking activity of simultaneously recorded actor neurons is not correlated in the normal monkeys. The percentage of correlated pallidal pairs significantly increase following MPTP treatment and is restored to nearly normal values under dopamine replacement therapy.
We conclude that the normal basal ganglia activity represents an optimally compressed (uncorrelated) version of distinctive features (as selected by the complement messages of the DA and ACh signals) of cortical information. Changes in basal ganglia synchronization are caused by DA depletion and are correlated with the clinical manifestations of Parkinsonism and its pharmacological treatment.
Glutamatergic neurons of the subthalamic nucleus (STN) possess intrinsic membrane properties that are likely to underlie, in part, their normal operation in voluntary movement and their abnormal operation in Parkinson's disease (PD).
In vitro, STN neurons discharge autonomously in the absence of synaptic input, a property that presumably contributes to their tonic activity in vivo and their role as a major driving force of neuronal activity in the basal ganglia. Autonomous activity is driven by voltage-activated sodium (Nav) channels, which are active at subthreshold voltages. The precision of autonomous activity is determined by the functional coupling of apamin-sensitive small conductance (SK) calcium-activated potassium (KCa) channels to voltage-activated calcium (Cav) 2.2 channels that are activated by action potentials.
In vitro, STN neurons exhibit a primary low-sensitivity frequency-intensity (f-I) relationship at frequencies associated with quiet wakefulness in vivo (< 40 Hz) and a higher sensitivity secondary range f-I relationship at frequencies associated with voluntary movement in vivo (> 40 Hz). SK KCa channels help to maintain the low-sensitivity of STN neurons at frequencies < 40 Hz, whereas Cav1.2.-1.3 channels, through their weak coupling to SK KCa channels, contribute to the enhanced sensitivity of secondary range firing. Thus STN neurons possess a high-pass filtering property that only permits the transition to high-sensitivity, high-frequency activity during concerted/synchronous excitatory cortical/thalamic input that presumably occurs during voluntary movement.
In vitro approximately 75% and 25% of STN neurons exhibit respectively, short (< 100 ms) duration or long (> 100 ms) rebound burst firing that follows the termination of hyperpolarization. Cav3 channels recover from inactivation when hyperpolarized below the voltages associated with autonomous oscillation and are responsible for the initial phase of rebound burst firing. In 75% of neurons, the relatively strong activation of SK KCa channels during rebound activity terminates the rebound within 100 ms. In 25% of neurons, the weaker activation of SK KCa channels during rebound activity permits the generation of a Cav1.2-1.3 channel-mediated plateau potential that extends the duration of rebound activity beyond 100 ms.
As the equilibrium potential of GABA-A receptor mediated current in STN neurons is more hyperpolarized than the voltages associated with autonomous oscillation or the degree of hyperpolarization required for rebound burst firing, GABA-A synaptic potentials can partially or completely reset the phase of autonomous oscillation and generate rebound burst activity. The first effect is due to the deactivation and recovery from slow inactivation of pacemaker Nav channels and the second effect is through the priming of Cav channels. Thus, irregular, poorly correlated GABAergic input in health is likely to disrupt the autonomous oscillation of STN neurons and produce irregular firing. In contrast rhythmic, bursting GABAergic input in PD may generate rhythmic, rebound burst firing in STN neurons.
Vocal learning by songbirds provides a model for studying general mechanisms of sensorimotor learning with particular relevance to human speech learning. For both songbirds and humans, hearing the sounds of others, and auditory feedback of oneself, plays a critical role in learning. I will discuss evidence from lesion and physiology experiments for a role of an avian basal-ganglia circuit, the 'anterior forebrain pathway (AFP)', in this learning process.
To understand any complex brain circuit, a useful exercise is to start with the minimal circuit needed to perform a fundamental behavioral function, and then to iterate a process of adding further circuit features in order to enhance that basic function. The result of such an exercise is a set of progressively complex approximations to the in vivo circuit, coupled with a set of explicit hypotheses about how each new circuit feature improves behavioral competence when said feature is added to the pre-existing circuit. In this talk, I will highlight the set of hypotheses that emerged as we followed this procedure to construct a model of how the basal ganglia might interact with a laminar model of frontal cortex to satisfy the staging and gating requirements of conditional voluntary behavior.
Our recent modeling efforts have focused on modeling SK blocker-induced, calcium mediated "bistable" oscillations in the slice preparation, and on modifying a model of NMDA-induced, sodium-mediated burst firing in a slice preparation in order to approximate the in vivo conditions in an intact animal.
Our previous model of bistable oscillations (Amini et al., 1999) was based on the regenerative depolarization provided by the L-type calcium channels and repolarization of the plateau by instantaneous calcium inactivation of a calcium channel. The model predicted that calcium would continue to increase throughout the plateau, but subsequent simultaneous electrophysiological and calcium imaging (Callaway, Wilson, and Shepard, 2000) showed that calcium reaches a plateau and even decreases during the plateau. Our current model was adjusted to allow the steady state calcium concentration to remain nearly constant over a voltage range near that observed during the depolarized phase of the plateau, and the adjusted version displays periodic switching between two stable voltage levels even though calcium stays constant or decreases slightly during the plateau.
The extension of the in vitro model to the in vivo situation was carried out in two stages. First, a static, tonic level of activation of the GABA A receptor was added to the model, and the levels of tonic activation of both GABA A and NMDA receptor activation were varied in the simulated presence and absence of SK blockers. The effects observed in the experimental literature after the in vivo application of bicucculine, picrotoxin, and a specific SK channel blocker can be explained in terms of underlying model mechanisms. Second, the kinetics of AMPA and NMDA receptor activation were modeled (but not those of GABA A). Two extremes were studied, one in which all glutamatergic inputs to the dopamine neuron were synchronous, and one with the activation of AMPA and NMDA receptors set to the average values observed during periodic stimulation. In the synchronous case, frequencies at which the NMDA but not AMPA receptor activation summated temporally preferentially activated bursting. Bursting was still observed at the average values, the bursts were spaced farther apart. The in vivo models do not include the bistable oscillations at this time, but they will be incorporated in future versions.
A Hamiltonian framework for motor planning is presented that allows for compliant goal-directed (discrete) and repetitive motion. Motor planning among local joint groups can be achieved by trading off potential and kinetic energy. The potential energy is assumed to arise from a combination of external constraints (gravity, external forces), and internally (neuronally) generated components. This formulation is able to account for some symptoms of Parkinson's and Huntington's diseases in terms of defects in the potential energy function. We hypothesize that one component of the neuronally-generated potential energy arises from a diffusion-style computation within the striatum, which is able to integrate the required sensory information and provide guidance for motion execution.
Microelectrode recordings during functional stereotactic surgery provide a unique opportunity to examine the firing patterns of neurons located in thalamus, globus pallidus and subthalamic nucleus. Alterations in firing rates and firing patterns have been hypothesized to be associated with several neurological disorders and thus it is of interest to determine and compare the neuronal activity in the basal ganglia and thalamus in various patient groups. I will present data on 1) firing rates and patterns of neurons in globus pallidus (GP) in Parkinson disease (PD), Huntington and dystonia patients. 2) effect of apomorophine on firing rates and patterns in GP and subthalamic nucleus (STN) in PD patients. 3) oscillatory activity in GP and STN in PD patients (microelectrode single units and macroelectrode field potentials). 4) crosscorrelation analysis of neuronal activity recorded from pairs of electrodes in GP and STN.
A songbird basal ganglia circuit called the AFP is known to be critical for song learning and adult plasticity. I will discuss our studies showing that neural firing in this circuit is strongly modulated by social context, in a manner suggestive of neuromodulatory actions. These studies also show that the level of "noise" and bursting varies in a way that could influence learning. Simultaneous recordings of the activity of this circuit and of its target motor nucleus also suggest that the AFP consists of a highly interconnected network of neurons. Variations in the degree of correlation in this network could alter information processing, as has been suggested in normal and diseased basal ganglia of mammals.
A new model of subthalamic-pallidal (STN-GP) interaction is presented. Individual multicompartmental models of STN cells and of a subtype of GP cells tuned to rat physiological and anatomical data are connected together to form a recurrent network. The explicit representation of both neuron morphological and physiological characteristics in individual model cells in the network allows us to investigate how these characteristics combine to underlie emergent network behaviours.
The complex nature of this type of model imposes serious constraints on its usefulness. I will open for discussion how far this type of model can be effectively utilised and present early results. In our initial simulations we explore one of the behaviours predicted by a high-level dynamic network model, the behaviour of widespread subthalamic bursts (introduced by David Willshaw). In this more complex network the ability of the STN to generate such bursts can be more fully explored.
The basal ganglia have been centrally implicated in a range of cognitive and motor disorders. In normal individuals the basal ganglia may be essential to the development of behavioral routines and the kinds of relatively automatic behaviors that underlie the habits of everyday life. Still at issue, however, is what the basal ganglia do when considered as a major neural processing system of the forebrain. A key fact about the basal ganglia is that they lie as nodal points in a set of cortico-basal ganglia-thalamocortical circuits that interconnect many parts of the cortex with the basal ganglia. As the cortically directed outputs of these circuits preferentially target the frontal cortex, these basal ganglia circuits can be considered to exert a primary influence over executive areas of the cortex. In our laboratory, we have developed the hypothesis that the basal ganglia function as an adaptive mechanism to adjust cortical activity in response to detected behavioral contingencies. As a first step in examining this hypothesis, we have recorded with single electrode and multiple electrode methods in the striatum as animals learn tasks. We have found evidence for remarkable plasticity in the response properties of striatal units as animals undergo training in procedural learning tasks. Recordings in the striatum during successive bouts of learning, extinction and reacquisition indicate that the ensemble activity of striatal units can change, then can be reversed and then be reinstated. Some of these recordings have been done for identified neurons on the striatum, neurons that are thought to be local circuit interneurons. The fact that interneurons as well as projection neurons undergo such plastic changes indicates that there is a reconfiguration of network activity in the striatum during the course of learning. Our view is that in order to develop a semi-automatic behavioral routine or to form a habit, it may be necessary to "chunk" together sequences of actions by means of developing new neuronal firing patterns that represent the entire action sequence or, at minimum, the beginning and end of such sequences. Combined ensemble recording methods now being introduced should help to unravel the relative roles of the neocortex and the striatum in this process.
Recent experimental results suggest that the trans-subthalamic nucleus or hyperdirect pathway , Cortex- Subthalamic-Nucleus (STN) - Globus Pallidus internal (GPi) plays a major role in the function of Basal Ganglia (BG). We propose a model of the motor part of the BG consisting in two networks, each of them corresponding to the execution of an action or a movement, and comprising six interacting populations: the CTX, the Striatum, the internal and the external part of the Globus-Pallidus, the Subthalamic Nucleus and the Thalamus. The architectures of the different pathways within each network satisfy current knowledge about the convergent-divergent pattern of connectivity in BG. Interactions between networks occur via a divergent STN-GPi connectivity. A central hypothesis of our model is that all the direct, indirect and hyperdirect pathways embedded into the system are closed onto the cortex. In this framework we show that the competitions between these feedback loops of different polarities may be one of the basis of the BG functions and dysfunctions. For that purpose we analyze the different dynamical regimes of the overall system as a function of the parameters (synaptic properties, propagation delays, external inputs...). We find that it can perform action selection provided that the *feedbacks* in the direct and hyperdirect loops are strong enough. In our model, dopamine (DA) depletion reduces the strength of the cortico-striatal synapses and consequently the overall strength of the direct loops, but does not affect the hyperdirect loops. Subsequently, selectivity is lost for moderate to strong DA depletion. Moreover, hyperdirect loops, with strong enough negative feedback can develop oscillations. In that case for large DA depletion synchronous oscillations driven by the hyperdirect loops emerge. These oscillations can be supressed by an increased cortical input.
I plan to describe two models of basal ganglia function. The first is a minimalistic model of medium spiny neurons that uses biophysically realistic data to explore the mechanism whereby reward likelihood modulates single unit responses in awake monkey subjects. Burst firing of dopamine neurons in anticipation of reward induces bistability in the response properties of the model spiny neuron, and this is sufficient to account for single unit responses. The second model is an anatomically and physiologically constrained network incorporating the loop of connectivity between cortical area 46 and the basal ganglia. This model is used to explore how serial order encoding might result from a combination of competitive pattern classification in the striatum, working memory in the cortex and the recursion that results from the anatomical loop. Consideration will also be given to the enhanced processing that could result from a network model that incorporates bistability in its striatal layer.
We have produced models of the basal ganglia at multiple levels of description, all constrained by our hypothesis that the basal ganglia act as a central switch in action selection. Systems-level models using either leaky integrator or more complex spiking integrate-and-fire (IF) units have demonstrated that the basal ganglia have functional characteristics consistent with action selection. Results from the IF models also revealed a possible source of slow oscillatory activity within a normal (non-parkinsonian) basal ganglia. Increasing the complexity of the IF model units, through the addition of a phenomenological description of membrane current dynamics and shunting inhibition, allowed us to replicate and explain the different forms of bursting within the STN-GPe loop reported in a recent in vitro study. Thus, our IF-unit based computational models have allowed us to propose underlying causes of basal ganglia neural activity patterns and have demonstrated the usefulness of modelling neural circuits across a range of complexity levels.
We have found anatomical evidence showing clusters of sodium channels at excitatory synapses on GP dendrites. Whole cell recordings in vitro then showed that electrical stimulation of dendritic excitatory inputs leads to the generation of dendritic action potentials. The same excitatory inputs are too small to even generate a discriminable subthreshold EPSP at the soma. We explore the consequences of this novel way of input coding in a modeling study.
Deep brain stimulation (DBS) of the basal ganglia or thalamus represents a dramatically effective treatment for clinically intractable movement disorders such as essential tremor and Parkinson's disease. However, the underlying mechanisms of its therapeutic action remain unknown. My research program consists of the development of systems level computer models of the effects DBS. The goal of which is to augment experimental investigation on the therapeutic mechanisms of action and design new electrodes and stimulation paradigms that maximize therapeutic benefit. My work couples the results of functional imaging and basic neurophysiology to computer models of extracellular electric fields and their effects on the nervous system. These models consist of three basic stages. The first step is the development of 3D finite element models of the electric field generated by DBS electrodes where the electrical properties of the tissue are based on diffusion tensor MRI. The second step is coupling the electric field to 3D reconstructions of neurons surrounding the electrode where the ion channel biophysics and firing properties of the neuron models are based on experimental recordings. The outcome of steps 1 & 2 are predictions on the volume of tissue surrounding the electrode affected by the stimulation. The final step is then to apply those stimulation effects to large scale neuronal network models of the thalamo-cortical-basal ganglia system that DBS modulates thereby providing experimentally testable hypotheses on the effects of stimulation in the different nuclei of the network. In addition, the results of this work can be coupled to PET/fMRI to provide a continuum from the single cell to the network to the behavior.
We study chains of stongly electrically coupled oscillators of Morris-Lecar type. When individual oscillators are in the regime close to an Andronov-Hopf bifurcation, the coupled system exhibits a variety of oscillatory behavior. The structure and bifurcations of stable periodic solutions are investigated using numerical and analytic techniques. This work is motivated by the firing modes found in the Wilson-Callaway model of a dopaminergic neuron.
The tremendous increase in the amount and complexity of neurophysiological data now possible as the result of advances in electronic and computer technology is overwhelming our ability to intuit inferences based solely on empiric observation. Metaphors are needed with mathematical rigor to help organize and explain empiric observations and, perhaps most importantly, to structure future testable biological hypotheses. The capabilities of computational and mathematical models to help understand biological data have increased dramatically with developments in the analysis of non-linear systems. In fact, the power of mathematical and computational models is such that biological problems can be solved in the most un-biological manner.
Biological theory, which is the starting point for most computational and mathematical models, is now the rate-limiting step for future advances. Further, there is the question of which level of analysis of biological phenomena is most appropriate for mathematical and computational analysis. For which level of biological complexity is it necessary to use the Hodgkin Huxley equations? Perhaps there are levels of questions in which the timeseries of extracellular action potentials considered at the level of communication theory is not only appropriate but also actually feasible.
The concept of emergent properties would suggest that there are multiple levels of complexity and abstraction in any biological question appropriate for mathematical and computational analysis. Unfortunately, discussions of emergent properties occur on an ontological rather than epistemic basis. The result often is that those advocating analysis at a level of emergent properties are accused of anti-reductionism. Viewing the notion of emergent properties as an epistemological question posits that the same data set can be viewed from a number of perspectives. There may be a subset of perspectives in which the same behavioral data will demonstrate more readily identifiable structure and thus, analysis is more likely to be productive.
Mathematicians and computer scientists are very dependent on neurophysiological theories to provide the appropriate context and biological constraints for their analyses and modeling. The current now called classical theory of basal ganglia function as first advanced by Albin et al. (Albin R.L. et al. (1989) The functions anatomy of basal ganglia disorders. Trends Neurosci. 12, 366-375) and DeLong (DeLong, M.R. (1990) Primate models of movement disorders of basal ganglia origin Trends in Neurosci. 13, 281-285) no longer may be a suitable theoretical context by which to advance mathematical and computational models of basal ganglia-thalamic-cortical function. There have been and are increasingly observations for which the classical model is insufficient to explain. Most of these counter-observations are anatomical (Parent A, Sato F, Wu Y, Gauthier J, Levesque M, Parent M. Organization of the basal ganglia: the importance of axonal collateralization TINS (suppl.) 2000:23:S20-S27) or clinical (Obeso JA, Rodriguez-Oroz MC, Rodriguez M, Lanciego JL, Artieda J, Gonzalo N, Olanow W. The physiology of the basal ganglia in Parkinson's disease. TINS (suppl.) 2000:23:S8-S19). For example, the ability of pallidotomy to improve dyskinesia would be unexpected based on the classical model. Recent neurophysiological studies of deep brain stimulation (DBS) in human and non-human primates suggest that DBS of the subthalamic nucleus and globus pallidus internal segment drives the output from these structures (Anderson ME, Postupna N, Ruffo M. Effects of high-frequency stimulation in the internal globus pallidus on the activity of thalamic neurons in the awake monkey. J Neurophysiol 2003;89:1150-1160, Hashimoto T, Elder CM, Okun MS, Patrick SK, Vitek JL. Stimulation of the subthalamic nucleus changes the firing pattern of pallidal neurons. J Neurosci 2003;23:1916-1923), which, according to the classical view, should worsen rather than improve Parkinson's disease.
The question now is whether the necessary changes in the classical model are quantitative or qualitative. This presentation argues for a qualitative or radical change. This argument is based on the notion that the classical model is incapable of providing insights into the dynamics of the basal ganglia-thalamic-cortical network. Indeed, it is the previous attempts to extend the classical model, which is static or at best a one-dimensional push pull system, into the domain of physiological time that generate inconsistencies and paradoxes. The current extrapolation of the static classical model into the domain of physiological time results in notions of hierarchical and sequential processing were physiological function is uniquely represented in each substructure of the basic ganglia- thalamic-cortical network.
Research into the therapeutic mechanisms of action of DBS in non-human primates has provided an opportunity to study the dynamics of the basal ganglia-thalamic-cortical network at high temporal resolution. This has resulted in the development of a new theory of basal ganglia-thalamic-cortical network function based on dynamically coupled high frequency reentrant non-linear oscillations. These oscillations are manifest in high frequency periodic changes in neuronal discharge probabilities rather than the actual discharge rates. Evidence of these high frequency oscillations in neuronal discharge probability come from direct observations and from paired-pulse DBS experiments demonstrating resonance effects at high frequencies and will be presented. Implications of this new theory of dynamically coupled high frequency reentrant non-linear oscillators for function of the basal ganglia-thalamic-cortical network will be discussed.
Neurons in the nucleus accumbens exhibit non-linear properties in their processing of information. Intracellular recordings in vivo revealed a bimodal distribution of membrane potential values, reflecting up and down membrane potential states. Up states are depolarizing events driven by sufficient activation of glutamatergic inputs. The strong hippocampal afferent projection to the accumbens is sufficient and necessary to elicit transitions to up states. Inputs from the prefrontal cortex, on the other hand, are more effective during up states. These findings have been interpreted as indicating a gating mechanism by which limbic inputs allow processing of prefrontal cortical information. Activation of amygdala or hippocampal afferents enhances the amplitude of subsequent synaptic responses to prefrontal cortical stimulation. Conversely, activation of prefrontal afferents reduces the amplitude of limbic synaptic responses. This has been interpreted as a strong cortical activation closing the limbic-driven gate. When cortical and limbic afferents are activated simultaneously, the intrinsic variability of synaptic responses is significantly reduced, suggesting increase of information in those conditions. Thus, the interactions between limbic and prefrontal cortical inputs depend on non-linear membrane properties and their nature varies with the timing of inputs. The outcome is thought to be important for selection of behavioral responses appropriate to the animal's environment.
Vocal learning in songbirds provides an excellent model for studying human speech learning and, more generally, sensorimotor learning in vertebrates. A set of discrete, interconnected forebrain nuclei, collectively termed the song system, mediates song learning and production. These nuclei form two major pathways, one essential for song production, the other essential for song learning. We have provided anatomical and electrophysiological evidence that the pathway essential for learning is a basal ganglia circuit with the major cell types and connections of both "direct" and "indirect" mammalian basal ganglia pathways. More recently, we have found a set of dopamine actions on cellular excitability, on synaptic transmission, and on activity-dependent synaptic plasticity, that could play an important role in song learning.
Systems composed of many non-linear units interacting locally can exhibit complex emergent properties that extend over a wide range of spatial and temporal scales. For example, avalanches, earthquakes, forest fires, and nuclear chain reactions all emerge from systems organized into a critical state where event sizes show no characteristic scale and are described by power laws. Theory has predicted that neuronal networks, likewise composed of locally interacting, non-linear units, should exhibit similarly complex emergent behavior. Here we provide the experimental support for this prediction that so far has been lacking as we describe the propagation of spontaneous activity in cultured networks of cortical neurons. As predicted by theory for a critical branching process, this propagation obeys a power law with an exponent of -3/2 for event sizes, with a branching parameter close to the critical value of 1. We use neural network simulations to show that a branching parameter at this value will optimize information transmission. Our findings suggest that "neuronal avalanches" may be a generic property of excitatory neural networks, and represent a mode of activity that differs profoundly from more commonly described oscillatory, synchronized, or wave-like network states. At this state characterized by a critical branching process, the network may satisfy the competing demands of information transmission and network stability.
Much is known about the biophysics and anatomy of basal ganglia motor circuits, their normal and pathological neuronal firing patterns, the characteristics of normal and pathological motor behavior and the effects of various treatment strategies on motor symptoms of basal ganglia disorders such as Parkinson's disease. This experimental basis together with the previous theoretical results is used in this work to build a biophysically-based network model of pallido-subthalamic circuits of basal ganglia, which are involved in movement control. We construct a model of basal ganglia network for control of motor programs. The network organization corresponds to the experimentally supported hypothesis that the basal ganglia facilitate the desired motor program and inhibit competing motor programs that interfere with the desired movement. The network consists of subthalamic and pallidal (both external and internal segments) units, with inputs from the cortex and the striatum. Network organization includes functional units within the basal ganglia nuclei that correspond to the desired motor program and the unwanted motor programs. A single compartment conductance-based model represents each unit. Dynamics of the model network is examined and the relationship between these dynamics and the motor behavior observed in normal subjects and the hypokinetic behavior in parkinsonian patients is considered.
The mechanism underlying the effectiveness of deep brain stimulation of the subthalamic nucleus (STN) or globus pallidus (GPi) for alleviating motor symptoms, such as those associated with Parkinson's disease, remains unknown. I will present joint work with David Terman in which we use a computational model to explore the hypothesis that DBS works by replacing pathologically rhythmic basal ganglia output with tonic, high frequency firing. This will include simulation results and a mathematical phase plane analysis of the mechanisms producing these results.
Midbrain dopaminergic neurons are crucially involved in a number of higher level cognitive and emotional functions in addition to their well-known involvement in voluntary motor behavior. Recent data linking the time-locked expression of a short burst of spikes in dopaminergic neurons in response to reward-related environmental stimuli have given added impetus to efforts to understand the mechanisms controlling the neuronal activity of these cells.
Dopaminergic neurons exhibit slow (averaging around 4Hz) spontaneous activity in vivo along a continuum of firing patterns ranging from a regular, pacemaker-like mode, through an irregular or random mode to a bursty firing mode in which a large fraction of the action potentials are fired within slow bursts consisting of 2-8 spikes at instantaneous firing rates of 15-20 Hz that often exhibit marked spike frequency adaptation and decreasing amplitude. Overall, there is not much difference in the mean firing rate of these three modes and it is rare to encounter cells firing more slowly than 1 Hz or more rapidly than 8 Hz. Often neurons exhibit spontaneous activity that is a mixture of all three of these modes. Interestingly, these three patterns of firing obtain only in vivo; it is difficult or impossible to evoke the random or bursty patterns of firing by conventional means in vitro, suggesting that they depend on some afferent input or inputs interacting with intrinsic membrane properties for their expression.
In vivo, burst firing that is qualitatively and quantitatively similar or identical to endogenous burst activity can be reliably evoked by blocking GABAA input to dopaminergic neurons, either by direct application of GABAA receptor antagonists (Tepper et al, 1995; Paladini and Tepper, 1999) or by inhibiting the firing of a normal source of GABAergic input to dopaminergic neurons (Celada et al., 1999). This suggests that at least one of the important afferents controlling dopaminergic neuron firing pattern is GABAergic. Recent experiments have revealed several aspects of this GABAergic regulation that may be of value in constructing biologically realistic system-level models of substantia nigra dopaminergic neurons.
The template hypothesis for birdsong learned first posited by Konishi has served as an organizing framework for the field. However, this hypothesis says relatively little about the nature of the various representations of song and how they interact during learning. I will review some basic models for how these representations might interact during the sensorimotor phase of learning. I will also discuss several questions regarding the notion of temporal hierarchy and its importance during song learning.
The subthalamic nucleus (STN) and external globus pallidus (GP) form a recurrent excitatory-inhibitory interaction within the basal ganglia. Through a computational model of these interactions we show that, under the influence of appropriate external input, the two nuclei can be switched between states of high and low activity or can generate oscillations consisting of bursts of high-frequency activity repeated at a low rate. It is further demonstrated from the model that the generation of the repetitive burst behaviour is favoured by increased inhibition of the GP, which is a condition indicated in Parkinson's disease. Paradoxically, increased striatal inhibition of the GP is predicted to cause an increase rather than a decrease in its mean firing rate. These behaviours, arising from a biologically inspired computational model of the STN-GP interaction, have important consequences for basal ganglia function and dysfunction.
Keywords: basal ganglia; subthalamic nucleus; globus pallidus; bifurcation analysis; oscillatory behaviour.
Cholinergic interneurons play an important but indirect role in synaptic integration in the neostriatum. Their synapses do not evoke fast synaptic potentials in the projection neurons of the striatum, but act as neuromodulators of voltage sensitive ion channels at several pre- and post-synaptic sites. Stimulation of cholinergic m1 receptors is also essential for induction of long term potentiation at the corticostriatal synapse. Background levels of acetylcholine are maintained in the neostriatum by the tonic spiking activity of cholinergic interneuons. Recent studies of cholinergic interneurons in striatal slices have shown that these cells are autonomous pacemakers, meaning no synaptic input is required to maintain their firing. Two qualitatively different autonomous firing patterns are expressed by the cells. One, a pattern of rhythmic single spiking, relies upon the interaction between subthreshold-activated non-inactivating sodium current, hyperpolarization-activated cation current (IH) and apamin-sensitive calcium-dependent potassium current (SK). The other pattern consists of rhythmic bursting, in which bursts consist of 5-15 action potentials and are generated at a rate of less than 1/second. That pattern is associated with a slow membrane potential oscillation that is not dependent upon sodium current or on apamin-sensitive calcium dependent potassium current. The depolarizing current for the underlying oscillation is a calcium current, and the bursts are phased by a slow, apamin-insensitive calcium-dependent potassium current.