Cellular-resolution optogenetics reveals attenuation-by-suppression in visual cortical neurons.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Machine Learning

Background:

• Understanding neural input-output (IO) functions is crucial for brain computation.
• Previous studies in vitro and anesthetized animals suggested nonlinearities in IO functions with increased network activity.
• How neurons transform inputs in vivo remains largely unclear.

Purpose of the Study:

• To characterize the activation functions of cortical principal neurons in awake mice.
• To investigate how neurons' IO functions change with varying network activity in vivo.
• To compare in vivo neural activation functions with those used in machine learning.

Main Methods:

• Utilized two-photon optogenetics to deliver fixed somatic inputs to cortical neurons in awake mice.
• Varied neuronal activity by presenting sensory stimuli.
• Measured neuronal responses to controlled optogenetic input under different activity states.

Main Results:

• Responses to fixed optogenetic input remained largely unchanged during neuronal excitation, indicating a linear response regime.
• Responses to fixed optogenetic input were significantly attenuated during neuronal suppression.
• This attenuation acts as a filtering mechanism, prioritizing inputs to excited neurons over suppressed ones.

Conclusions:

• Somatic neural activation functions in vivo are comparable to those used in modern machine learning systems.
• Neurons' IO functions dynamically filter sensory inputs, influencing how stimuli are processed.
• Sensory stimuli not only alter spiking outputs but also modulate responses to subsequent inputs, demonstrating a flexible information processing strategy.