Bidirectional optimization of firing rate in a mouse neuronal brain-machine interface

Area of Science:

• Neuroscience
• Computational Neuroscience
• Neuroplasticity

Background:

• Neuroplasticity allows the brain to adapt neural activity.
• It remains unclear if neuroplasticity can be used for abstract optimization tasks, such as finding curve extrema.
• Understanding this could enhance brain-machine interface (BMI) applications.

Purpose of the Study:

• To investigate if motor cortex neurons in mice can optimize their activity along a unimodal curve using a BMI.
• To determine if neuroplasticity underlies the ability to learn abstract optimization rules through feedback.

Main Methods:

• A brain-machine interface (BMI) was used in mice.
• Auditory feedback of neuronal firing rate was paired with water rewards.
• Mice were trained to modulate firing rates along a curve mapping firing rate to sound frequency increase speed, seeking the reward-accelerating extremum.

Main Results:

• Mice successfully learned to modulate firing rates towards the curve's peak.
• Reward acquisition time decreased significantly (from 18.64 ± 7.30 s to 11.59 ± 4.38 s).
• High-response events increased (from 66 to 104 occurrences), and neurons prioritized high-response intervals.

Conclusions:

• Cortical neurons can dynamically optimize activity along non-monotonic reward landscapes.
• Neuroplasticity serves as a substrate for adaptive self-optimization in abstract tasks.
• This research expands understanding of learning abstract rules via feedback and has implications for neuroprosthetic design.