Global and Multiplexed Dendritic Computations under In Vivo-like Conditions

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• Dendrites exhibit nonlinear integration of synaptic inputs, but their precise contribution to neuronal output remains unclear.
• Understanding neuronal input-output transformations is crucial for deciphering cortical circuit computations.

Purpose of the Study:

• To develop a statistically principled model for characterizing neuronal input-output transformations under in vivo-like conditions.
• To investigate the role of dendritic nonlinearities in shaping somatic responses to complex spatiotemporal synaptic input patterns.

Main Methods:

• Development of a hierarchical linear-nonlinear (hLN) subunit model.
• Application of the hLN model to predict the somatic membrane potential of a detailed biophysical model of a L2/3 pyramidal cell.
• Validation of model predictions using in vivo data.

Main Results:

• A linear-nonlinear model with a single global dendritic nonlinearity achieved over 90% prediction accuracy for somatic membrane potential.
• A novel hLN motif, input multiplexing, enhanced prediction accuracy comparably to additional layers of local nonlinearities.
• Similar results were observed in two other neuronal cell types.

Conclusions:

• The hLN modeling approach provides a data-driven characterization of neuronal input-output transformations during in vivo-like activity.
• Dendritic nonlinearities play a significant role in shaping neuronal responses, and their integration can be effectively modeled.
• Input multiplexing represents a key computational motif in neuronal processing.