A long-range, recurrent neuronal network linking the emotion regions with the somatic motor cortex

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• Recurrent neural networks (RNNs) excel at modeling sequential data in silico.
• Their existence and formation within native mammalian brain networks remain largely uncharacterized.

Purpose of the Study:

• To investigate the presence and structure of an innate RNN in the mammalian brain.
• To elucidate the neural pathways involved in processing emotion-related sequential information.

Main Methods:

• Tracing neural connections from emotion regions (basal lateral amygdala, insular cortex) to the medial prefrontal cortex (mPFC).
• Analyzing neuronal connectivity within the mPFC and its projections to the somatic motor cortex (sMO).
• Utilizing chemogenetic silencing of mPFC neurons in vivo to assess pathway function.

Main Results:

• Identified a three-unit RNN structure involving emotion regions, mPFC (hidden unit), and sMO (output units).
• Observed unidirectional connections from basal lateral amygdala and insular cortex to mPFC motor-cortex-projecting (MP) neurons.
• Found that MP neurons form a local self-feedback loop and project to the sMO, with stronger input to infragranular layers.
• Chemogenetic silencing of MP neurons abolished communication from emotion regions to the sMO.

Conclusions:

• Mammalian brains possess an innate RNN architecture.
• This neural circuit facilitates the transmission of emotion-related information to motor output regions.
• The medial prefrontal cortex plays a crucial role as a hidden unit in this innate RNN.