Limits...
Effects of dopamine depletion on network entropy in the external globus pallidus.

Cruz AV, Mallet N, Magill PJ, Brown P, Averbeck BB - J. Neurophysiol. (2009)

Bottom Line: Dopamine depletion led to decreases in the firing rates of GPe neurons and increases in synchronized network oscillations in the beta frequency (13-30 Hz) band.Changes in autocorrelations tended to offset these effects because autocorrelations decreased entropy more in the control animals.Thus it is possible that reduced information coding capacity within basal ganglia networks may contribute to the behavioral deficits accompanying PD.

View Article: PubMed Central - PubMed

Affiliation: Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College London, London WC1N 3BG, United Kingdom.

ABSTRACT
Dopamine depletion in cortical-basal ganglia circuits in Parkinson's disease (PD) grossly disturbs movement and cognition. Classic models relate Parkinsonian dysfunction to changes in firing rates of basal ganglia neurons. However, disturbances in other dynamics of neural activity are also common. Taking both inappropriate firing rates and other dynamics into account and determining how changes in the properties of these neural circuits that occur during PD impact on information coding are thus imperative. Here, we examined in vivo network dynamics in the external globus pallidus (GPe) of rats before and after chronic dopamine depletion. Dopamine depletion led to decreases in the firing rates of GPe neurons and increases in synchronized network oscillations in the beta frequency (13-30 Hz) band. Using logistic regression models, we determined the combined and separate impacts of these factors on network entropy, a measure of the upper bound of information coding capacity. Importantly, changes in these features in dopamine-depleted rats led to a significant decrease in GPe network entropy. Changes in firing rates had the largest impact on entropy, with changes in synchrony also decreasing entropy at the network level. Changes in autocorrelations tended to offset these effects because autocorrelations decreased entropy more in the control animals. Thus it is possible that reduced information coding capacity within basal ganglia networks may contribute to the behavioral deficits accompanying PD.

Show MeSH

Related in: MedlinePlus

Optimal number of lags for auto- (K1) and cross-(K2) correlations. A: optimal number of lagged time bins (5 ms each), selected with Bayesian information criteria (BIC) for autocorrelations in lesioned and control ensembles. B: same for cross-correlations. C: total of auto- and cross-correlations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2724349&req=5

f2: Optimal number of lags for auto- (K1) and cross-(K2) correlations. A: optimal number of lagged time bins (5 ms each), selected with Bayesian information criteria (BIC) for autocorrelations in lesioned and control ensembles. B: same for cross-correlations. C: total of auto- and cross-correlations.

Mentions: We began by determining the number of lagged time bins that were necessary in the model. This is an estimate of the time interval over which a spike train is correlated with itself (oscillations) or between pairs of neurons (synchrony), and time bins that were not necessary are bins that had a statistically negligible contribution. We found that the effect of the autocorrelations rarely extended beyond ∼15 bins (75 ms) for either the control or the lesioned animals (Fig. 2A). The effect of the cross-correlations, however, rarely went beyond about six bins (30 ms) for the lesioned animals, and never went beyond one bin (0 time lag, synchronous spikes) for the normal animals (Fig. 2B). In fact, for the control animals, 99.3% of the pairs did not have a significant synchronous term and for the lesioned animals >80% did not (Fig. 2B). Therefore the correlation time was much shorter between neurons than within the spike trains of a single neuron, and this effect was more dramatic for the control animals than it was for the lesioned animals. The number of parameters for the full model was then the sum of the number for the auto and cross models (Fig. 2C).


Effects of dopamine depletion on network entropy in the external globus pallidus.

Cruz AV, Mallet N, Magill PJ, Brown P, Averbeck BB - J. Neurophysiol. (2009)

Optimal number of lags for auto- (K1) and cross-(K2) correlations. A: optimal number of lagged time bins (5 ms each), selected with Bayesian information criteria (BIC) for autocorrelations in lesioned and control ensembles. B: same for cross-correlations. C: total of auto- and cross-correlations.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2724349&req=5

f2: Optimal number of lags for auto- (K1) and cross-(K2) correlations. A: optimal number of lagged time bins (5 ms each), selected with Bayesian information criteria (BIC) for autocorrelations in lesioned and control ensembles. B: same for cross-correlations. C: total of auto- and cross-correlations.
Mentions: We began by determining the number of lagged time bins that were necessary in the model. This is an estimate of the time interval over which a spike train is correlated with itself (oscillations) or between pairs of neurons (synchrony), and time bins that were not necessary are bins that had a statistically negligible contribution. We found that the effect of the autocorrelations rarely extended beyond ∼15 bins (75 ms) for either the control or the lesioned animals (Fig. 2A). The effect of the cross-correlations, however, rarely went beyond about six bins (30 ms) for the lesioned animals, and never went beyond one bin (0 time lag, synchronous spikes) for the normal animals (Fig. 2B). In fact, for the control animals, 99.3% of the pairs did not have a significant synchronous term and for the lesioned animals >80% did not (Fig. 2B). Therefore the correlation time was much shorter between neurons than within the spike trains of a single neuron, and this effect was more dramatic for the control animals than it was for the lesioned animals. The number of parameters for the full model was then the sum of the number for the auto and cross models (Fig. 2C).

Bottom Line: Dopamine depletion led to decreases in the firing rates of GPe neurons and increases in synchronized network oscillations in the beta frequency (13-30 Hz) band.Changes in autocorrelations tended to offset these effects because autocorrelations decreased entropy more in the control animals.Thus it is possible that reduced information coding capacity within basal ganglia networks may contribute to the behavioral deficits accompanying PD.

View Article: PubMed Central - PubMed

Affiliation: Sobell Department of Motor Neuroscience and Movement Disorders, Institute of Neurology, University College London, London WC1N 3BG, United Kingdom.

ABSTRACT
Dopamine depletion in cortical-basal ganglia circuits in Parkinson's disease (PD) grossly disturbs movement and cognition. Classic models relate Parkinsonian dysfunction to changes in firing rates of basal ganglia neurons. However, disturbances in other dynamics of neural activity are also common. Taking both inappropriate firing rates and other dynamics into account and determining how changes in the properties of these neural circuits that occur during PD impact on information coding are thus imperative. Here, we examined in vivo network dynamics in the external globus pallidus (GPe) of rats before and after chronic dopamine depletion. Dopamine depletion led to decreases in the firing rates of GPe neurons and increases in synchronized network oscillations in the beta frequency (13-30 Hz) band. Using logistic regression models, we determined the combined and separate impacts of these factors on network entropy, a measure of the upper bound of information coding capacity. Importantly, changes in these features in dopamine-depleted rats led to a significant decrease in GPe network entropy. Changes in firing rates had the largest impact on entropy, with changes in synchrony also decreasing entropy at the network level. Changes in autocorrelations tended to offset these effects because autocorrelations decreased entropy more in the control animals. Thus it is possible that reduced information coding capacity within basal ganglia networks may contribute to the behavioral deficits accompanying PD.

Show MeSH
Related in: MedlinePlus