Detection and estimation of neural connectivity based on crosscorrelation analysis.

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neural Network Modeling

Background:

• Crosscorrelation analysis is a standard technique for assessing functional neural connectivity by examining simultaneous neuronal activity.
• Understanding the limitations and conditions under which crosscorrelation accurately reflects neural connections is crucial for interpreting experimental data.

Purpose of the Study:

• To evaluate the adequacy of crosscorrelation methods for detecting and estimating neural connectivity.
• To investigate how factors like firing rate, postsynaptic potential duration, and network structure influence the accuracy of connectivity inference.

Main Methods:

• Computer simulations of small neural networks using realistic model neurons.
• Analysis of crosscorrelation functions under varying parameters, including mean firing intervals and postsynaptic potential durations.

Main Results:

• Excitatory connections are more easily detected when mean firing intervals are much larger than postsynaptic potential durations.
• Inhibitory connections are better revealed when mean firing intervals are smaller than postsynaptic potential durations.
• The 'shift predictor' method shows limited utility due to non-linear interactions between stimuli and connectivity, and potential neural timelock to stimuli.

Conclusions:

• Crosscorrelation analysis is sensitive to the relative timing of neural activity and postsynaptic potentials.
• The commonly used 'shift predictor' may not reliably distinguish between stimulus-evoked activity and true neural connectivity.
• In networks with parallel pathways, crosscorrelation estimates effective connectivity rather than direct synaptic connections.