Discrete- and continuous-time probabilistic models and algorithms for inferring neuronal UP and DOWN states

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neuronal populations exhibit periodic UP and DOWN states, characterized by fluctuations in spiking activity.
• Understanding these dynamics is vital for comprehending neural information representation and transmission.
• Limited research has focused on the stochastic properties of UP-DOWN state dynamics.

Purpose of the Study:

• To develop and present novel Markov and semi-Markov probability models.
• To estimate UP and DOWN states from multiunit neural spiking activity.
• To provide a statistical characterization of UP-DOWN state dynamics.

Main Methods:

• Modeled multiunit neural spiking activity as a stochastic point process.
• Incorporated hidden UP and DOWN states and ensemble spiking history.
• Employed Expectation-Maximization (EM) and Monte Carlo EM algorithms for parameter and state estimation.

Main Results:

• Successfully applied models to simulated and in-vivo neural spiking data.
• Provided a statistical framework for analyzing UP-DOWN state dynamics.
• Enabled joint estimation of hidden states and model parameters.

Conclusions:

• The developed models offer a robust statistical characterization of UP-DOWN state dynamics.
• This approach can refine and verify mechanistic models of cortical circuit function.
• Facilitates deeper understanding of neural coding during different brain states.