Tracing 'driver' versus 'modulator' information flow throughout large-scale, task-related neural circuitry.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Local field potentials (LFPs) are commonly used for modeling brain network information flow.
• Mathematical techniques like Granger causality analysis are often applied to LFP data.
• The interpretation of LFP data in driving downstream neural activity can be problematic.

Purpose of the Study:

• To compare the utility of neural action potential (spike) data versus LFPs for modeling information flow in brain networks.
• To investigate whether LFP analysis can lead to spurious inferences about causal relationships between brain areas.
• To determine the most appropriate data type for modeling neural information processing.

Main Methods:

• Simultaneous recording of spikes and LFPs from the rat motor cortex (M1) and red nucleus (mRN) during a skilled reaching task.
• Application of Granger causality analysis to both LFP and spike data.
• Comparison of results obtained from LFP-based and spike-based analyses.

Main Results:

• Granger causality analysis on LFPs suggested that M1 drove mRN spiking.
• Analysis of spike data revealed no significant correlations between M1 and mRN activity.
• LFP analysis, reflecting dendritic currents, produced different interpretations than spike analysis, reflecting neural output.

Conclusions:

• LFP data, primarily reflecting dendritic activity, may be more indicative of intrinsic processing and inputs.
• Spike data provides a more direct measure of neural output and information flow.
• Mathematical models of information flow should differentiate between LFP and spike data to avoid spurious conclusions.