Models of Neuronal Stimulus-Response Functions: Elaboration, Estimation, and Evaluation

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neurons encode complex sensory stimuli through time-varying activity.
• Understanding neural encoding requires characterizing the function linking neuronal firing to sensory input history.

Purpose of the Study:

• To provide a unifying and critical survey of techniques for modeling neural responses to sensory stimuli.
• To make key concepts accessible to researchers outside the field.

Main Methods:

• Review of classical linear receptive field models.
• Examination of modern approaches incorporating normalization and nonlinearities.
• Discussion of model structure, parameter estimation, and regularization techniques.

Main Results:

• Comparison of benefits and drawbacks of various modeling approaches.
• Identification of scenarios where different methods succeed or fail.
• Quantification of model-neuron response agreement.

Conclusions:

• A comprehensive overview of neural encoding models is presented.
• Accessible explanations of complex methods are provided.
• Unified code for implemented methods is available to the research community.