| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Physiology, The Hebrew University–Hadassah Medical School; and 2Center for Neural Computation and 3Eric Roland Center for Neurodegenerative Diseases, The Hebrew University, Jerusalem, Israel
Submitted 19 February 2007; accepted in final form 14 August 2007
 |
ABSTRACT |
|---|
|
|
|
INTRODUCTION |
|---|
|
) and orientation movements (Basso and Wurtz 2002; Bayer et al. 2004; Handel and Glimcher 1999; Hikosaka et al. 2000, 2006). Physiological studies have indicated that as many as 58% of SNr neurons modulate their firing rate in relation to visual stimuli or saccades (Hikosaka and Wurtz 1983), 60% during the initiation and maintenance of smooth pursuit eye movements (Basso et al. 2005), and 16–25% to orofacial movements (DeLong et al. 1983; Schultz 1986; Wichmann and Kliem 2004). However, the SNr has been shown to be associated with other nonorofacial motor and sensory events (DeLong et al. 1983; Schultz 1986; Wichmann and Kliem 2004). In different studies the percentage of SNr neurons related to forelimb movements ranges between 5 and 46%. In addition, several studies report a significant percentage of SNr neurons with multimodality (Nagy et al. 2005) and more than one event (Schultz 1986; Wichmann and Kliem 2004) responses.
Such similar response properties appear to suggest that many SNr neurons share common functional inputs. This idea is supported by the anatomical characteristics of SNr afferents (Bevan et al. 1996; Francois et al. 1987; Kolomiets et al. 2003). Although neurons driven by common sources are expected to display correlated activity, as indicated by peaks or troughs in their spike-to-spike cross-correlation histograms (Brown et al. 2004; Eggermont 1990; Perkel et al. 1967), an early study of SNr neurons recorded in anesthetized rats reported that SNr spontaneous activity was uncorrelated (Wilson et al. 1977). Recent studies of the SNr of freely moving rats (Deransart et al. 2003) and behaving primates (Nevet et al. 2004) revealed a similar lack of spike-to-spike correlation in the control states.
Most early correlation studies focused on spike-timing (spike-to-spike) correlations (Abeles 1982; Eggermont 1990; Perkel et al. 1967), but other types of correlation between the spiking activity of two neurons can be evaluated. The predominant view today characterizing neuronal encoding by changes in firing rates has inspired new measures of correlation based on rates, or spike count. These were introduced recently and have been used in several brain areas (Averbeck et al. 2006; Bair et al. 2001; Gawne and Richmond 1993; Lee et al. 1998; Prut and Perlmutter 2003; van Kan et al. 1985). The spike-count correlation functions include the signal correlation (correlation between the average responses of two neurons to different events, indicating their tendency to respond similarly) and noise correlation (the correlation between intertrial fluctuations of the neuronal responses). These new measures of correlation also provide the basis for theoretical studies of information processing in the nervous system (Gawne and Richmond 1993; Oram et al. 1998; Panzeri and Schultz 2001; Pola et al. 2003; Schneidman et al. 2003, 2006).
Previous correlation studies of the SNr and other basal ganglia nuclei (Kimura et al. 2003; Levy et al. 2002; Nini et al. 1995; Raz et al. 2000) have concentrated on spike-timing correlations, whereas the novel measures of noise and signal (count) correlations have generally been neglected. In the presentstudy, we conducted multiple electrode recordings of SNr neurons in monkeys performing a probabilistic delayed visuomotor response task and studied the different forms of correlation between simultaneously recorded SNr neurons.
|
|
METHODS |
|---|
|
Two female Macaca fasicularis monkeys (E and G), weighing 2.5 and 3.5 kg, were used in this study. Care and surgical procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (1996) and with the Hebrew University guidelines for the use and care of laboratory animals in research, supervised by the Institutional Animal Care and Use Committee.
Behavioral paradigm
The monkeys were trained to perform a self-initiated probabilistic delayed visuomotor task, in which the probability of receiving reinforcement for correct performance depended on the visual cue presented (for details see Morris et al. 2004). Briefly, during all training and recording sessions, monkeys were seated facing a computer screen, with a panel consisting of three keys in front of them. Trials were initiated when the monkey touched the central key. After a variable delay of 1.5–2.5 s, a visual cue appeared for 350 ms on one side of the screen. The monkeys were overtrained with a set of four possible cues. Each cue was associated with a different probability of reward (0.25, 0.5, 0.75, and 1.0). The cue presentation was followed by a fixed hold period of 1.5 s, after which a Go signal appeared. The monkeys were required to press either the left or the right key to match the side of the visual cue. Correct performance was followed (after an interval of 700 ms) by liquid reward at the probability associated with the visual cue. All trials were followed by a variable intertrial interval (ITI) of 3–6 s, after which the color of the screen changed ("Pre-Cue"), and the central key was enabled. In most cases, the monkey had returned to the central key already within the ITI, and therefore the trial was initiated immediately. The same-duration ITI also followed all trials aborted due to any behavioral errors. All analyses presented here include only the correctly performed trials that met the electrophysiological stability criteria (see following text).
Surgical procedures
A square plastic recording chamber with a 27-mm (inner) side was attached to the skull to allow access to the SNr. In monkey E, the recording chamber was tilted 56° laterally in the coronal plane with its center targeted at the stereotaxic coordinates of the SNr (Martin and Bowden 2000; Szabo and Cowan 1984). In monkey G, the recording chamber was placed with its center dorsal to the stereotaxic coordinates of the SNr. The chamber coordinates were verified with MRI imaging (Biospec Bruker 4.7-Tesla animal system, fast-spin echo sequence; effective TE = 80 ms and TR = 2.5 s, 13 coronal slices 1 or 2 mm wide) by alignment of the two-dimensional MRI images with the sections from the atlas (Martin and Bowden 2000; Szabo and Cowan 1984). A cilux head holder was also attached to the monkey's head to allow for fixation during recording. At the end of the recording period, the chamber and head holder were removed, the skin was sutured, and following a recovery period the monkeys were sent to a primate sanctuary (www.ipsf.org.il). All surgical and MRI procedures were performed under general and deep anesthesia.
Neuronal recording
During recording sessions the monkey's head was fixed by connecting the head holder to an external metal frame. Eight glass-coated tungsten microelectrodes (impedance 0.3–1.2 M
at 1,000 Hz), confined within a cylindrical guide (1.65-mm inner diameter), were individually advanced (EPS, Alpha-Omega Engineering, Nazareth, Israel) to the SNr. Signals from the electrodes were amplified with a gain of 10K and band-pass filtered with a 300- to 6,000-Hz four-pole Butterworth filter (MCP+, Alpha-Omega Engineering). Extracellular action potentials were detected and classified on-line using a template-matching algorithm (MSD, Alpha-Omega Engineering). Spike-detection signals (spike trains) and behavioral events were logged to a data acquisition system at 12 kHz (AlphaMap, Alpha-Omega Engineering). Cells were classified as SNr neurons if they were found at the expected stereotaxic coordinates and matched the physiological characteristics typical of SNr neurons: a relatively short-duration action potential and a distinctively high firing rate (DeLong et al. 1983; Schultz 1986). Firing characteristics of neighboring neurons and fibers were also used to guarantee the identification of the target cells as SNr neurons. For example, adjacent to the identified SNr, usually deeper, neurons of the oculomotor nucleus and fibers of the oculomotor nerve were often encountered, displaying a characteristic waveform shape with a dominant positive phase, and prominent changes in discharge rate related to eye movements.
Data analysis
PREPROCESSING. Spike trains were used for analysis only if their spike waveforms could be reliably separated from those of other units during the on-line spike sorting. A critical issue in studying neuronal activity by extracellular recording in behaving animals is to make sure that the spike train recorded is a reliable presentation of a single and noninjured neuron. Each of the spike trains was therefore analyzed for rate stability. In this analysis, the rate of each unit during all ITI periods was displayed graphically as a function of time for the entire recording session. The underlying assumption is that unnoticed movements of the electrode and recording of another neuron, as well as possible injury to the recorded neuron, would be reflected in a sudden or gradual change in the tonic discharge rate (as reflected in the ITI discharge rate) of the recorded neuron. Only periods of time judged as displaying a stable firing rate were further analyzed. Note that this discharge stability is tested during the ITIs and is therefore not affected by modulation of discharge rate within the trial.
RESPONSES TO BEHAVIORAL EVENTS AND RESPONSE-MATCHES. Neuronal responses to events were evaluated by comparing the (negative or positive) peak of the firing rate of the neurons in the 500-ms time intervals starting at the time of the event (response interval) to the average rate in the baseline interval, defined as a 500-ms epoch in the ITI starting 1,500 ms before the Pre-Cue. The rate responses were calculated in 1-ms bins, averaged over trials, and then smoothed with a Gaussian window (SD = 100 ms). Responses were included in the database only if recorded for at least 10 repetitions of the event. Due to the complexity of SNr response patterns and differences in span and delay of responses, we compared a few methods of evaluation of response significance of the peristimulus histogram (PSTH). A definition of a significant response as a deviation of the peak rate from the baseline exceeding ±20% was found to be closest to our subjective evaluation of a significant response in the PSTH and the raster plots; therefore we used this criterion. However, a more formal criterion, a Mann–Whitney test of the null hypothesis that the rates during the response interval are derived from a distribution similar to the rates during the baseline interval, was also used (significance level P < 0.05). In cases of biphasic responses, composed of both increase and decrease in discharge rate, the response with the maximal deviation from the baseline was considered as the representative response to this event.
To examine the overlap between responses of pairs of neurons we calculated the dot-product of the response vectors—the "response-match"—which is a measure of correlation for each time point of each pair of PSTHs. This was performed by scalar multiplication of normalized PSTH vectors. This enabled a pairwise analysis of the overlap between changes in average firing rate during the trials. PSTHs were calculated in 1-ms bins, averaged over trials, and smoothed with a 25- or 100-ms Gaussian window. For this purpose we used PSTH vectors calculated for the whole trial and merged the responses to the Pre-Cue signal, the Cue and the Go signal, the movements, and the Reward, yielding 8,200-ms vectors. Each PSTH vector was normalized so that the baseline would be equivalent to zero and increases and decreases relative to it would be positive and negative, respectively. The normalization was performed as follows: R(t) = [r(t) – b]/b, where R(t) represents the normalized PSTH, r is the raw PSTH, and b is the baseline mean (defined earlier). The response-match was defined as RM(t) = R1(t)·R2(t), with a single product calculated for each time bin. This way, RM(t) > 0 indicates that both neurons displayed an average activity at time t that was either higher or lower than their baseline; RM(t) < 0 indicates that one neuron had a higher and the other a lower average firing rate than its baseline at this time point; and RM(t) = 0 indicates that at least one of the neurons did not modulate its firing rate in this time period. For each pair, we calculated a reference response-match surrogate vector by randomly shuffling the bins in both PSTHs. A response-match was considered significant if more than three consecutive bins deviated from the mean of the reference vector by >2.5 SDs (SD calculated from the reference vector as well). This criterion was chosen because it yielded <5% correlated pairs when calculated for the surrogate vectors.
SPIKE-TIMING CORRELATIONS.
Spike-to-spike cross-correlation functions (Perkel et al. 1967) were calculated with a 1-ms bin size for a time window of ±500 ms. Correlation functions were calculated over the whole duration of stable recording (including all intertrial and trial segments) and for the Reward epoch alone (see following text). The confidence limits of each correlogram were estimated using a resampling (bootstrap) method. This was performed by the calculation of 10 surrogate vectors; in each, every spike was randomly moved 0 to 50 ms forward or backward (Rivlin-Etzion et al. 2006). A cross-correlogram was considered significant if more than three consecutive bins deviated from the mean of the bootstrap vectors by >1 SD (SD calculated from the surrogate vectors). This criterion was selected because it yielded <5% significant correlations in the surrogate vectors. Only neurons recorded from different electrodes were examined for correlation analysis, to avoid detection of false correlations due to shadowing of coincident spikes (Bar-Gad et al. 2001). In an analysis of three or more simultaneously recorded neurons for correlation, not all comparisons are independent (e.g., if there is a correlation between the neuron pairs AB and BC, one would probably find a correlation between the AC pair). However, although we included correlation analysis for all possible pairs, this did not affect the statistical significance of our findings because we report a negative result.
Cross-correlations in Reward epochs were analyzed only for correct (rewarded) trials following cues that indicated a reward with a probability of 1. These were calculated in the time interval starting 500 ms before the reward until 1,000 ms after the reward and averaged over trials. A shift predictor was computed as the average of 10 surrogate cross-correlations calculated after random shuffling of the trials of one neuron. The shift predictor was subtracted from the original (raw) correlogram to produce the normalized correlogram (Eggermont 1990; Stevens and Gerstein 1976). Confidence levels were calculated as for the raw cross-correlogram (see earlier text).
SPIKE-COUNT CORRELATIONS. Signal correlations are defined as the correlation coefficients between the average responses (defined here as the spike count in the 500-ms time intervals starting at the time of the event) to different events. Signal correlations were calculated for the four-point event vector composed of the Pre-Cue, Cue, Go, and Reward events. Formally, if the average rate in the Pre-Cue, Cue, Go, and Reward is represented by P, C, G, and R, respectively, then for two neurons (1 and 2) the signal correlation (SC) is defined as SC = COV ([P1; C1; G1; R1][P2; C2; G2; R2])/SQRT{VAR ([P1; C1; G1; R1])·VAR ([P2; C2; G2; R2])}, where COV stands for covariance, SQRT for square root, and VAR for variance. Signal correlation can be calculated for simultaneously and nonsimultaneously recoded neurons.
Noise correlations were defined as the correlation coefficients between the number of spikes generated by each of the simultaneously recorded neurons in response to a single type of event (Gawne and Richmond 1993). The correlation coefficient by definition measures the correlation between intertrial fluctuations around the mean response in a single stimulus condition. We calculated noise correlations for the same events used for the calculation of the signal correlations. Formally, if the discharge rate of a unit during event E in trial t is represented by E(t) (E is either Pre-Cue, Cue, Go, or Reward), then for n trials the noise correlation (NC) for two neurons (1 and 2) in event E is defined as NC(E) = COV {[E1(1);...; E1(n)][E2(1);...; E2(n)]}/SQRT{VAR[E1(1);...; E1(n)]·VAR[E2(1);...; E2(n)]}, where COV stands for covariance, SQRT for square root, and VAR for variance. Noise correlation can be calculated only for simultaneously recoded neurons.
Average signal and noise correlations were calculated using Z-transforms to avoid bias from the zero to one range of the correlation-coefficient values.
|
|
RESULTS |
|---|
|
580 spikes in the analyzed period.
|
Most (55%) of the SNr neurons (that passed the inclusion criteria for analysis of responses, n = 34 and 31 for monkeys E and G, respectively) responded to multiple events, with a greater tendency to respond to the reward and the instruction cue than to other events (Table 1). As in previous studies of the globus pallidus (Arkadir et al. 2004; DeLong 1971; Turner and Anderson 1997) and SNr (Handel and Glimcher 1999; Wichmann and Kliem 2004) we found that SNr neurons tend to fire at a high (mean = 58 spikes/s, SE = 3.7) tonic discharge rate and to display about equal proportions of responses with increases (48.5%) versus decreases (51.5%) in firing rates (Table 1). Using a formal Mann–Whitney yielded qualitatively similar results.
We considered a change in firing rate as a task-related response if the average change in discharge rate was statistically significant. These changes in the discharge rate may be associated with changes in the variability of the discharge. As in other neuronal structures (Lee et al. 1998) we found that the relationship between the mean and the variability of discharge rates can be fitted well by power functions for both ITI and Reward epochs (Fig. 1A). The power functions that fit the actual data best and the corresponding r2 were: ITI: monkey E, SD = 0.35M0.38 (r2 = 0.37) and monkey G, SD = 0.66M0.49 (r2 = 0.33); Reward: monkey E, SD = 0.44M0.50 (r2 = 0.65) and monkey G, SD = 0.83M0.61 (r2 = 0.66). In these equations, M represents the mean and SD represents the SD of the discharge rates, respectively. Because unlike other studied areas, a significant fraction of SNr neurons responded by decreasing their discharge rate (Table 1 and Fig. 2), we also calculated the variability in the discharge rate during the reward period as a function of its deviation from the baseline (absolute value, normalized by the baseline rate). Figure 1B shows that the variability in the SNr firing rate was proportional to the magnitude of the positive (higher than baseline) responses. However, the variability of the negative (lower than baseline) responses was smaller than that of the positive responses and less affected by the amount of deviation from the background rate. Thus decoding (read-out by the next stations of the basal ganglia networks) of the neuronal responses of negative SNr responses was enhanced by the reduced variability of lower discharge rate of a Poisson-like firing (Fig. 1A), as well as by additional processes leading to stable variability of these responses (Fig. 1B).
|
|
|
|
Spike-timing correlations
In contrast to the similar responses of SNr neurons, spike-to-spike correlations between neurons recorded simultaneously (Fig. 5) were significant in only 2.1% of the cases (monkey E: 6/70; monkey G: 11/285), even less than the 3.8% significant pairs found in surrogate vectors. To isolate possible event-related correlations we also calculated the cross-correlation on Reward epochs (the time interval starting 500 ms before reward until 1,000 ms after reward). Although 63% of the neurons responded to the reward event, and therefore the expected percentage of correlated pairs was 0.632 
40%, only 10% of the reward epoch correlograms displayed significant peaks or troughs. All of the significant peaks and troughs vanished after normalization with a shift predictor (Fig. 6), indicating that these correlations were not due to precise time correlations in spiking activity, but rather were a consequence of simultaneous rate changes.
|
|
Signal correlation values were broadly distributed, showing both positive and negative values. The correlation values were subjected to Z-transformation for analysis of their nontruncated distribution (Fig. 7E). The average Z-transform of SNr signal correlations was –0.16 in monkey E and 0.09 in monkey G, both nonsignificantly different from zero (P < 0.05, two-tailed t-test). The 95% confidence interval calculated for the average signal correlation of both monkeys together was between –0.164 and 0.212.
|
; Brody 1999).
|
; Oram et al. 1998). A direct analysis of information coding (Pola et al. 2003; Schneidman et al. 2003) should be performed to further evaluate the mutual information relationships between SNr neurons.
|
|
|
|
DISCUSSION |
|---|
|
Previous studies have indicated that under conditions of normal dopaminergic activity, SNr and pallidal neurons lack spike-to-spike cross-correlations (Nevet et al. 2004; Nini et al. 1995; Raz et al. 2000; Wilson et al. 1977). However, correlations of the response properties of basal ganglia neurons have not been previously examined. The data presented here show analysis of average signal and noise correlations of SNr neurons. Most of the neurons we examined responded to the instruction cue and the delivery of reward, and their response-match vectors were rarely flat. Traditionally, such overlapping responses are expected to lead to correlated spiking activity (Eggermont 1990; Perkel et al. 1967). However, all the average measures of spike-count and spike-timing correlations we calculated were close to zero. To the best of our knowledge this is the first CNS structure where the average values of spike-count correlations are not significantly (at P < 0.05) different from zero. Although this unique finding could be due to several mechanisms, it still narrows the scope of possible modes of information coding in the basal ganglia and points to specific models of these structures (e.g., sparse connectivity, active decorrelation).
Lack of correlation can improve information transfer
The lack of average signal or noise correlations (compare with Lee et al. 1998) in a small nucleus such as the SNr that receives funneling inputs from much larger structures, including the external segment of the globus pallidus, subthalamic nucleus, and striatum, is surprising. Because most of the SNr neurons are responsive to the same events, the lack of correlation probably serves a purpose and is not fortuitous. The amount of information carried by a nucleus can be bounded by the signal correlation between its neurons (Gawne and Richmond 1993; Samengo and Treves 2001). Therefore keeping low values of the average signal correlations may improve the efficiency of information transfer by the SNr. The tonic background activity of SNr neurons—enabling both increases and decreases in firing rate in responses to behavioral events—probably plays a critical role in creating this equal (around zero) distribution of correlation values resulting in an information-efficient mode of work.
A low level of noise correlation improves the signal-to-noise ratio achieved when averaging multiple inputs. The need to avoid the consequences of correlated noise and redundancy of information (Oram et al. 1998) may favor a selection of connection weights that keeps the responses of adjacent neurons nearly independent for both signal and noise. Neurons sharing common inputs can be decorrelated by different means, including sparse afferent connectivity (Jaeger 2003), various temporal filtering properties of different neurons (Gawne and Richmond 1993), and local Hebbian and anti-Hebbian learning rules as in principal component neural networks (Bar-Gad et al. 2003; Diamantaras and Kung 1996; Foldiak 1990; Oja 1982). Most of these would not result in synergy (a state where the population activity provides more information than the sum of their individual members). The independence of the signal and noise correlation in our study indicates that the basal ganglia networks achieve maximal reduction of redundancy, yet synergy is not observed. Further studies should examine whether this is a property of our limited behavioral paradigm, whether the neuronal machinery of the basal ganglia was not stretched to its maximal capabilities, or whether the neuronal machinery necessary for creating and reading a synergetic code does not exist in the basal ganglia.
The basal high firing rate of SNr neurons is intrinsic
The lack of spike-to-spike correlations (compare with Bair et al. 2001; Gawne and Richmond 1993) supports the hypothesis that the basal high-rate firing of these neurons is intrinsically driven, and thus does not depend on the SNr inputs (Atherton and Bevan 2005; Nakanishi et al. 1987; Stanford 2003; Windels and Kiyatkin 2006). The overlapping responses are superimposed on the uncorrelated basal activity. If the input-driven modulations are rare and weak relative to the basal activity, their influence on the spike-to-spike correlogram will not be significant. In this case no peaks or troughs will be found even without using a shift predictor. A simulation we conducted demonstrates this (Fig. 10) and highlights the differences between correlation studies of neurons with high spontaneous discharge rates versus neurons with low background discharge (Stevens and Gerstein 1976). Therefore we argue that the spontaneous activity of SNr neurons is self-initiated and thus independent, whereas SNr responses are driven by overlapping input sources. The finding that the variability of the positive responses and the ITI discharge rate, but not the negative responses, is correlated with the mean firing rate supports the conjecture that the negative responses of SNr neurons are driven by different mechanisms than the spontaneous activity and positive responses (Gulley et al. 2002; Sato and Hikosaka 2002).
|
Pallidal neurons demonstrate overlapping responses (Anderson and Horak 1985; Arkadir et al. 2004; Brotchie et al. 1991; DeLong 1972; Nambu et al. 1990) without spike-to-spike correlations, like the SNr. The suggestion of independent spontaneous activity modulated by overlapping inputs can therefore be applied to the pallidum as well. However, significant spike-to-spike correlations appear in tremulous Parkinson's disease (PD) patients (Hurtado et al. 1999; Levy et al. 2002) and in primate models of this disease (Heimer et al. 2006; Raz et al. 2000). This phenomenon, not yet tested in the SNr, may reflect a higher dominance of input-driven rate modulations relative to the basal activity of output basal-ganglia neurons in this disease. Signal and noise correlations in the output structures of the basal ganglia of parkinsonian monkeys have yet to be evaluated.
|
|
GRANTS |
|---|
|
|
|
ACKNOWLEDGMENTS |
|---|
|
|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: A. Nevet, Department of Physiology, The Hebrew University–Hadassah Medical School, P.O. Box 12272, Jerusalem, Israel 91120 (E-mail: alonne{at}ekmd.huji.ac.il)
|
|
REFERENCES |
|---|
|
Anderson ME, Horak FB. Influence of the globus pallidus on arm movements in monkeys. III. Timing of movement-related information. J Neurophysiol 54: 433–448, 1985.
Arkadir D, Morris G, Vaadia E, Bergman H. Independent coding of movement direction and reward prediction by single pallidal neurons. J Neurosci 24: 10047–10056, 2004.
Atherton JF, Bevan MD. Ionic mechanisms underlying autonomous action potential generation in the somata and dendrites of GABAergic substantia nigra pars reticulata neurons in vitro. J Neurosci 25: 8272–8281, 2005.
Averbeck BB, Latham PE, Pouget A. Neural correlations, population coding and computation. Nat Rev Neurosci 7: 358–366, 2006.[CrossRef][Web of Science][Medline]
Bair W, Zohary E, Newsome WT. Correlated firing in macaque visual area MT: time scales and relationship to behavior. J Neurosci 21: 1676–1697, 2001.
Bar-Gad I, Morris G, Bergman H. Information processing, dimensionality reduction and reinforcement learning in the basal ganglia. Prog Neurobiol 71: 439–473, 2003.[CrossRef][Web of Science][Medline]
Bar-Gad I, Ritov Y, Vaadia E, Bergman H. Failure in identification of overlapping spikes from multiple neuron activity causes artificial correlations. J Neurosci Methods 107: 1–13, 2001.[CrossRef][Web of Science][Medline]
Basso MA, Pokorny JJ, Liu P. Activity of substantia nigra pars reticulata neurons during smooth pursuit eye movements in monkeys. Eur J Neurosci 22: 448–464, 2005.[CrossRef][Web of Science][Medline]
Basso MA, Wurtz RH. Neuronal activity in substantia nigra pars reticulata during target selection. J Neurosci 22: 1883–1894, 2002.
Bayer HM, Handel A, Glimcher PW. Eye position and memory saccade related responses in substantia nigra pars reticulata. Exp Brain Res 154: 428–441, 2004.[CrossRef][Web of Science][Medline]
Ben Shaul Y, Bergman H, Ritov Y, Abeles M. Trial to trial variability in either stimulus or action causes apparent correlation and synchrony in neuronal activity. J Neurosci Methods 111: 99–110, 2001.[CrossRef][Web of Science][Medline]
Bevan MD, Smith AD, Bolam JP. The substantia nigra as a site of synaptic integration of functionally diverse information arising from the ventral pallidum and the globus pallidus in the rat. Neuroscience 75: 5–12, 1996.[CrossRef][Web of Science][Medline]
Brody CD. Correlations without synchrony. Neural Comput 11: 1537–1551, 1999.[CrossRef][Web of Science][Medline]
Brotchie P, Iansek R, Horne MK. Motor function of the monkey globus pallidus. 1. Neuronal discharge and parameters of movement. Brain 114: 1667–1683, 1991.
Brown EN, Kass RE, Mitra PP. Multiple neural spike train data analysis: state-of-the-art and future challenges. Nat Neurosci 7: 456–461, 2004.[CrossRef][Web of Science][Medline]
DeLong MR. Activity of pallidal neurons during movement. J Neurophysiol 34: 414–427, 1971.
DeLong MR. Activity of basal ganglia neurons during movement. Brain Res 40: 127–135, 1972.[CrossRef][Web of Science][Medline]
DeLong MR, Crutcher MD, Georgopoulos AP. Relations between movement and single cell discharge in the substantia nigra of the behaving monkey. J Neurosci 3: 1599–1606, 1983.[Web of Science][Medline]
Deransart C, Hellwig B, Heupel-Reuter M, Leger JF, Heck D, Lucking CH. Single-unit analysis of substantia nigra pars reticulata neurons in freely behaving rats with genetic absence epilepsy. Epilepsia 44: 1513–1520, 2003.[CrossRef][Web of Science][Medline]
Diamantaras KI, Kung SY. Principal Component Neural Networks: Theory and Applications. New York: Wiley, 1996.
Eggermont JJ. The Correlative Brain. Berlin: Springer-Verlag, 1990.
Foldiak P. Forming sparse representations by local anti-Hebbian learning. Biol Cybern 64: 165–170, 1990.[CrossRef][Web of Science][Medline]
Francois C, Yelnik J, Percheron G. Golgi study of the primate substantia nigra. II. Spatial organization of dendritic arborizations in relation to the cytoarchitectonic boundaries and to the striatonigral bundle. J Comp Neurol 265: 473–493, 1987.[CrossRef][Web of Science][Medline]
Gawne TJ, Richmond BJ. How independent are the messages carried by adjacent inferior temporal cortical neurons? J Neurosci 13: 2758–2771, 1993.[Abstract]
Gulley JM, Kosobud AE, Rebec GV. Behavior-related modulation of substantia nigra pars reticulata neurons in rats performing a conditioned reinforcement task. Neuroscience 111: 337–349, 2002.[CrossRef][Web of Science][Medline]
Handel A, Glimcher PW. Quantitative analysis of substantia nigra pars reticulata activity during a visually guided saccade task. J Neurophysiol 82: 3458–3475, 1999.
Heimer G, Rivlin M, Israel Z, Bergman H. Synchronizing activity of basal ganglia and pathophysiology of Parkinson's disease. J Neural Transm Suppl 70: 17–20, 2006.[CrossRef][Medline]
Hikosaka O, Nakamura K, Nakahara H. Basal ganglia orient eyes to reward. J Neurophysiol 95: 567–584, 2006.
Hikosaka O, Takikawa Y, Kawagoe R. Role of the basal ganglia in the control of purposive saccadic eye movements. Physiol Rev 80: 953–978, 2000.
Hikosaka O, Wurtz RH. Visual and oculomotor functions of monkey substantia nigra pars reticulata. I. Relation of visual and auditory responses to saccades. J Neurophysiol 49: 1230–1253, 1983.
Hurtado JM, Gray CM, Tamas LB, Sigvardt KA. Dynamics of tremor-related oscillations in the human globus pallidus: a single case study. Proc Natl Acad Sci USA 96: 1674–1679, 1999.
Inchul P, Amano N, Satoda T, Murata T, Kawagishi S, Yoshino K, Tanaka K. Control of oro-facio-lingual movements by the substantia nigra pars reticulata: high-frequency electrical microstimulation and GABA microinjection findings in rats. Neuroscience 134: 677–689, 2005.[CrossRef][Web of Science][Medline]
Jaeger D. No parallel fiber volleys in the cerebellar cortex: evidence from cross-correlation analysis between Purkinje cells in a computer model and in recordings from anesthetized rats. J Comput Neurosci 14: 311–327, 2003.[CrossRef][Web of Science][Medline]
Kimura M, Matsumoto N, Okahashi K, Ueda Y, Satoh T, Minamimoto T, Sakamoto M, Yamada H. Goal-directed, serial and synchronous activation of neurons in the primate striatum. Neuroreport 14: 799–802, 2003.[CrossRef][Web of Science][Medline]
Kolomiets BP, Deniau JM, Glowinski J, Thierry AM. Basal ganglia and processing of cortical information: functional interactions between trans-striatal and trans-subthalamic circuits in the substantia nigra pars reticulata. Neuroscience 117: 931–938, 2003.[CrossRef][Web of Science][Medline]
Lee D, Port NL, Kruse W, Georgopoulos AP. Variability and correlated noise in the discharge of neurons in motor and parietal areas of the primate cortex. J Neurosci 18: 1161–1170, 1998.
Levy R, Hutchison WD, Lozano AM, Dostrovsky JO. Synchronized neuronal discharge in the basal ganglia of parkinsonian patients is limited to oscillatory activity. J Neurosci 22: 2855–2861, 2002.
Martin RF, Bowden DM. Primate Brain Maps: Structure of the Macaque Brain. Amsterdam: Elsevier Science, 2000.
Morris G, Arkadir D, Nevet A, Vaadia E, Bergman H. Coincident but distinct messages of midbrain dopamine and striatal tonically active neurons. Neuron 43: 133–143, 2004.[CrossRef][Web of Science][Medline]
Nagy A, Paroczy Z, Norita M, Benedek G. Multisensory responses and receptive field properties of neurons in the substantia nigra and in the caudate nucleus. Eur J Neurosci 22: 419–424, 2005.[CrossRef][Web of Science][Medline]
Nakanishi H, Kita H, Kitai ST. Intracellular study of rat substantia nigra pars reticulata neurons in an in vitro slice preparation: electrical membrane properties and response characteristics to subthalamic stimulation. Brain Res 437: 45–55, 1987.[CrossRef][Web of Science][Medline]
Nambu A, Yoshida S, Jinnai K. Discharge patterns of pallidal neurons with input from various cortical areas during movement in the monkey. Brain Res 519: 183–191, 1990.[CrossRef][Web of Science][Medline]
Nevet A, Morris G, Saban G, Fainstein N, Bergman H. Discharge rate of substantia nigra pars reticulata neurons is reduced in non-parkinsonian monkeys with apomorphine-induced orofacial dyskinesia. J Neurophysiol 92: 1973–1981, 2004.
Nini A, Feingold A, Slovin H, Bergman H. Neurons in the globus pallidus do not show correlated activity in the normal monkey, but phase-locked oscillations appear in the MPTP model of parkinsonism. J Neurophysiol 74: 1800–1805, 1995.
Oja E. A simplified neuron model as a principal component analyzer. J Math Biol 15: 267–273, 1982.[CrossRef][Web of Science][Medline]
Oram MW, Foldiak P, Perrett DI, Sengpiel F. The "Ideal Homunculus": decoding neural population signals. Trends Neurosci 21: 259–265, 1998.[CrossRef][Web of Science][Medline]
Panzeri S, Schultz SR. A unified approach to the study of temporal, correlational, and rate coding. Neural Comput 13: 1311–1349, 2001.[CrossRef][Web of Science][Medline]
Perkel DH, Gerstein GL, Moore GP. Neuronal spike trains and stochastic point processes. II. Simultaneous spike trains. Biophys J 7: 419–440, 1967.[Web of Science][Medline]
Pola G, Thiele A, Hoffmann KP, Panzeri S. An exact method to quantify the information transmitted by different mechanisms of correlational coding. Network 14: 35–60, 2003.[Web of Science][Medline]
Prut Y, Perlmutter SI. Firing properties of spinal interneurons during voluntary movement. II. Interactions between spinal neurons. J Neurosci 23: 9611–9619, 2003.
Raz A, Vaadia E, Bergman H. Firing patterns and correlations of spontaneous discharge of pallidal neurons in the normal and the tremulous 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine vervet model of parkinsonism. J Neurosci 20: 8559–8571, 2000.
Rivlin-Etzion M, Ritov Y, Heimer G, Bergman H, Bar-Gad I. Local shuffling of spike trains boosts the accuracy of spike train spectral analysis. J Neurophysiol 95: 3245–3256, 2006.
Samengo I, Treves A. Representational capacity of a set of independent neurons. Phys Rev E Stat Nonlin Soft Matter Phys 63: 011910, 2001.[Medline]
Sato M, Hikosaka O. Role of primate substantia nigra pars reticulata in reward-oriented saccadic eye movement. J Neurosci 22: 2363–2373, 2002.
Schneidman E, Berry MJ, Segev R, Bialek W. Weak pairwise correlations imply strongly correlated network states in a neural population. Nature 440: 1007–1012, 2006.[CrossRef][Medline]
Schneidman E, Bialek W, Berry MJ. Synergy, redundancy, and independence in population codes. J Neurosci 23: 11539–11553, 2003.
Schultz W. Activity of pars reticulata neurons of monkey substantia nigra in relation to motor, sensory, and complex events. J Neurophysiol 55: 660–677, 1986.
Stanford IM. Independent neuronal oscillators of the rat globus pallidus. J Neurophysiol 89: 1713–1717, 2003.
Stevens JK, Gerstein GL. Interactions between cat lateral geniculate neurons. J Neurophysiol 39: 239–256, 1976.
Szabo J, Cowan WM. A stereotaxic atlas of the brain of the cynomolgus monkey (Macaca fascicularis). J Comp Neurol 222: 265–300, 1984.[CrossRef][Web of Science][Medline]
Turner RS, Anderson ME. Pallidal discharge related to the kinematics of reaching movements in two dimensions. J Neurophysiol 77: 1051–1074, 1997.
van Kan PL, Scobey RP, Gabor AJ. Response covariance in cat visual cortex. Exp Brain Res 60: 559–563, 1985.[Web of Science][Medline]
Wichmann T, Kliem MA. Neuronal activity in the primate substantia nigra pars reticulata during the performance of simple and memory-guided elbow movements. J Neurophysiol 91: 815–827, 2004.
Wilson CJ, Young SJ, Groves PM. Statistical properties of neuronal spike trains in the substantia nigra: cell types and their interactions. Brain Res 136: 243–260, 1977.[CrossRef][Web of Science][Medline]
Windels F, Kiyatkin EA. GABAergic mechanisms in regulating the activity state of substantia nigra pars reticulata neurons. Neuroscience 140: 1289–1299, 2006.[CrossRef][Web of Science][Medline]
Zohary E, Shadlen MN, Newsome WT. Correlated neuronal discharge rate and its implications for psychophysical performance. Nature 370: 140–143, 1994.[CrossRef][Medline]
This article has been cited by other articles:
|
|
 |
|
  M. Joshua, A. Adler, B. Rosin, E. Vaadia, and H. Bergman Encoding of Probabilistic Rewarding and Aversive Events by Pallidal and Nigral Neurons J Neurophysiol, February 1, 2009; 101(2): 758 - 772. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Elias, Y. Ritov, and H. Bergman Balance of Increases and Decreases in Firing Rate of the Spontaneous Activity of Basal Ganglia High-Frequency Discharge Neurons J Neurophysiol, December 1, 2008; 100(6): 3086 - 3104. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. L. Bodor, K. Giber, Z. Rovo, I. Ulbert, and L. Acsady Structural Correlates of Efficient GABAergic Transmission in the Basal Ganglia-Thalamus Pathway J. Neurosci., March 19, 2008; 28(12): 3090 - 3102. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
TOP
ABSTRACT
INTRODUCTION
