Impact of network structure and cellular response on spike time correlations

Area of Science:

• Computational Neuroscience
• Neural Systems Dynamics
• Theoretical Neurobiology

Background:

• Experimental techniques increasingly allow observation of simultaneous activity in large neuronal populations.
• Understanding correlated neuronal activity is vital for deciphering the neural code.
• A key theoretical challenge is linking network architecture and single-cell properties to correlation structure.

Purpose of the Study:

• To develop a general theoretical approach for analyzing correlated activity in neural networks.
• To provide explicit expressions for cross-correlations between neurons in networks with arbitrary architecture.
• To investigate how neuronal operating points and network connectivity influence correlation magnitude and timescale.

Main Methods:

• Extension of linear response theory to analyze neural population dynamics.
• Modeling networks of general integrate-and-fire neurons with arbitrary connectivity.
• Derivation of approximate cross-correlation expressions and their expansion in terms of network architecture.

Main Results:

• Neural correlations are strongly dependent on the neurons' operating point (input mean and variance), even with fixed connectivity.
• Network architecture can be interpreted via an expansion related to paths between neurons.
• In excitatory-inhibitory networks, the balance of synaptic strengths and timescales critically shapes correlation structure.

Conclusions:

• The developed framework provides explicit expressions for average correlation structure in randomly connected networks.
• These findings identify key factors governing coordinated neural activity.
• The study offers insights into how network architecture and cell dynamics collectively shape neural population correlations.