Limits...
Input dependent cell assembly dynamics in a model of the striatal medium spiny neuron network.

Ponzi A, Wickens J - Front Syst Neurosci (2012)

Bottom Line: However if cortical excitation strength is increased more regularly firing and completely quiescent cells are found, which depend on the cortical stimulation.Model cell population PSTH display both stimulus and temporal specificity, with large population firing rate modulations locked to elapsed time from task events.We suggest the MSN network is well suited to the generation of such slow coherent task dependent response which could be utilized by the animal in behavior.

View Article: PubMed Central - PubMed

Affiliation: Neurobiology Research Unit, Okinawa Institute of Science and Technology Okinawa, Japan.

ABSTRACT
The striatal medium spiny neuron (MSN) network is sparsely connected with fairly weak GABAergic collaterals receiving an excitatory glutamatergic cortical projection. Peri-stimulus time histograms (PSTH) of MSN population response investigated in various experimental studies display strong firing rate modulations distributed throughout behavioral task epochs. In previous work we have shown by numerical simulation that sparse random networks of inhibitory spiking neurons with characteristics appropriate for UP state MSNs form cell assemblies which fire together coherently in sequences on long behaviorally relevant timescales when the network receives a fixed pattern of constant input excitation. Here we first extend that model to the case where cortical excitation is composed of many independent noisy Poisson processes and demonstrate that cell assembly dynamics is still observed when the input is sufficiently weak. However if cortical excitation strength is increased more regularly firing and completely quiescent cells are found, which depend on the cortical stimulation. Subsequently we further extend previous work to consider what happens when the excitatory input varies as it would when the animal is engaged in behavior. We investigate how sudden switches in excitation interact with network generated patterned activity. We show that sequences of cell assembly activations can be locked to the excitatory input sequence and outline the range of parameters where this behavior is shown. Model cell population PSTH display both stimulus and temporal specificity, with large population firing rate modulations locked to elapsed time from task events. Thus the random network can generate a large diversity of temporally evolving stimulus dependent responses even though the input is fixed between switches. We suggest the MSN network is well suited to the generation of such slow coherent task dependent response which could be utilized by the animal in behavior.

No MeSH data available.


500 cell network simulations under variation of connectivity for weak input rC = 0.0085, bC = 0.001, α = 5, τg = 50. Each point is calculated from a 50-s time series from a different network simulation. Black solid circles - mean CV over all cells. Red solid squares – mean CV of k-means generated clusters. Green solid diamonds – mean CV of randomly generated clusters. Blue dashed triangles - proportion of cells which fire at least one spike.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3306002&req=5

Figure 1: 500 cell network simulations under variation of connectivity for weak input rC = 0.0085, bC = 0.001, α = 5, τg = 50. Each point is calculated from a 50-s time series from a different network simulation. Black solid circles - mean CV over all cells. Red solid squares – mean CV of k-means generated clusters. Green solid diamonds – mean CV of randomly generated clusters. Blue dashed triangles - proportion of cells which fire at least one spike.

Mentions: In Figure 1 we describe how the spiking activity of 500 cell networks depends on connectivity ρ at low input rates rC ∼ 10 Hz and uniform input distribution α = 5. Throughout the whole connectivity range the proportion of active cells (which fire at least one spike during the simulation period) is usually 100% (Figure 1, blue dashed triangles) and cells firing completely regularly with constant inter-spike-interval (ISI) are absent. The whole network is therefore engaged in dynamical activity. To investigate the structure of this dynamical activity we use the coefficient of variation (CV) of the cells’ ISI distributions (Tomko and Crapper, 1974). This defined to be the ISI standard deviation divided by the ISI mean for each cell, then averaged over all active cells. A high value, larger than unity, indicates spiking which is burstier (more clustered) than Poisson, while as the CV approaches zero firing becomes more regular. As connectivity increases the coefficient of variation (Figure 1, black solid circles) decreases from a high value near 2.5 to a value near 1. This means that at low connectivity many cells are firing in a bursty way with episodes of high frequency spiking separated by long silent periods, while at high connectivity most cells are firing in a Poisson like way without high frequency bursts and pauses.


Input dependent cell assembly dynamics in a model of the striatal medium spiny neuron network.

Ponzi A, Wickens J - Front Syst Neurosci (2012)

500 cell network simulations under variation of connectivity for weak input rC = 0.0085, bC = 0.001, α = 5, τg = 50. Each point is calculated from a 50-s time series from a different network simulation. Black solid circles - mean CV over all cells. Red solid squares – mean CV of k-means generated clusters. Green solid diamonds – mean CV of randomly generated clusters. Blue dashed triangles - proportion of cells which fire at least one spike.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3306002&req=5

Figure 1: 500 cell network simulations under variation of connectivity for weak input rC = 0.0085, bC = 0.001, α = 5, τg = 50. Each point is calculated from a 50-s time series from a different network simulation. Black solid circles - mean CV over all cells. Red solid squares – mean CV of k-means generated clusters. Green solid diamonds – mean CV of randomly generated clusters. Blue dashed triangles - proportion of cells which fire at least one spike.
Mentions: In Figure 1 we describe how the spiking activity of 500 cell networks depends on connectivity ρ at low input rates rC ∼ 10 Hz and uniform input distribution α = 5. Throughout the whole connectivity range the proportion of active cells (which fire at least one spike during the simulation period) is usually 100% (Figure 1, blue dashed triangles) and cells firing completely regularly with constant inter-spike-interval (ISI) are absent. The whole network is therefore engaged in dynamical activity. To investigate the structure of this dynamical activity we use the coefficient of variation (CV) of the cells’ ISI distributions (Tomko and Crapper, 1974). This defined to be the ISI standard deviation divided by the ISI mean for each cell, then averaged over all active cells. A high value, larger than unity, indicates spiking which is burstier (more clustered) than Poisson, while as the CV approaches zero firing becomes more regular. As connectivity increases the coefficient of variation (Figure 1, black solid circles) decreases from a high value near 2.5 to a value near 1. This means that at low connectivity many cells are firing in a bursty way with episodes of high frequency spiking separated by long silent periods, while at high connectivity most cells are firing in a Poisson like way without high frequency bursts and pauses.

Bottom Line: However if cortical excitation strength is increased more regularly firing and completely quiescent cells are found, which depend on the cortical stimulation.Model cell population PSTH display both stimulus and temporal specificity, with large population firing rate modulations locked to elapsed time from task events.We suggest the MSN network is well suited to the generation of such slow coherent task dependent response which could be utilized by the animal in behavior.

View Article: PubMed Central - PubMed

Affiliation: Neurobiology Research Unit, Okinawa Institute of Science and Technology Okinawa, Japan.

ABSTRACT
The striatal medium spiny neuron (MSN) network is sparsely connected with fairly weak GABAergic collaterals receiving an excitatory glutamatergic cortical projection. Peri-stimulus time histograms (PSTH) of MSN population response investigated in various experimental studies display strong firing rate modulations distributed throughout behavioral task epochs. In previous work we have shown by numerical simulation that sparse random networks of inhibitory spiking neurons with characteristics appropriate for UP state MSNs form cell assemblies which fire together coherently in sequences on long behaviorally relevant timescales when the network receives a fixed pattern of constant input excitation. Here we first extend that model to the case where cortical excitation is composed of many independent noisy Poisson processes and demonstrate that cell assembly dynamics is still observed when the input is sufficiently weak. However if cortical excitation strength is increased more regularly firing and completely quiescent cells are found, which depend on the cortical stimulation. Subsequently we further extend previous work to consider what happens when the excitatory input varies as it would when the animal is engaged in behavior. We investigate how sudden switches in excitation interact with network generated patterned activity. We show that sequences of cell assembly activations can be locked to the excitatory input sequence and outline the range of parameters where this behavior is shown. Model cell population PSTH display both stimulus and temporal specificity, with large population firing rate modulations locked to elapsed time from task events. Thus the random network can generate a large diversity of temporally evolving stimulus dependent responses even though the input is fixed between switches. We suggest the MSN network is well suited to the generation of such slow coherent task dependent response which could be utilized by the animal in behavior.

No MeSH data available.