Inferring causal connectivity from pairwise recordings and optogenetics.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Econometrics

Background:

• Understanding neural mechanisms requires quantifying causal interactions between neurons.
• Optogenetics allows specific neural perturbation but widefield interventions cause confounds.
• Discerning specific neuronal contributions is challenging due to simultaneous stimulation.

Purpose of the Study:

• To address confounds in optogenetic interventions.
• To introduce econometrics-based methods for estimating causal effective connectivity.
• To improve the accuracy of neural interaction analysis.

Main Methods:

• Combining instrumental variables (IV) and difference in differences (DiD) techniques.
• Exploiting the interaction between stimulation and the absolute refractory period.
• Applying methods to simulated neural networks.

Main Results:

• Identified confounds in widefield optogenetics lead to biased interpretations.
• Demonstrated that IV and DiD methods reduce confounding biases.
• Showcased superior performance of IV/DiD over naive techniques in simulations.

Conclusions:

• Econometric methods like IV and DiD can effectively disentangle neural interactions.
• These causal inference techniques offer a powerful approach for analyzing brain connectivity.
• The study highlights the utility of interdisciplinary methods in neuroscience research.