Dynamic tracking of objects in the macaque dorsomedial frontal cortex.

Area of Science:

• Cognitive Neuroscience
• Neurophysiology
• Computational Neuroscience

Background:

• Cognitive neuroscience posits internal world models and mental simulation for physical inferences.
• Human experiments support mental simulation theory in physical reasoning tasks.
• Direct neural evidence for population dynamics reflecting environmental simulations is limited.

Purpose of the Study:

• To test the mental simulation hypothesis by examining neural population dynamics.
• To investigate if neural activity in the frontal cortex reflects real-time simulation of environmental states.
• To provide evidence for a neural substrate of mental simulation.

Main Methods:

• Combined a naturalistic ball-interception task with non-human primates.
• Utilized large-scale electrophysiology to record neural activity.
• Employed recurrent neural network (RNN) modeling for data analysis.

Main Results:

• Neurons in the dorsomedial frontal cortex (DMFC) showed multiplexed representation of ball position.
• Population activity in DMFC formed a low-dimensional embedding tracking the ball's trajectory, visible or not.
• RNN models provided further support for the mental simulation hypothesis when compared to DMFC data.

Conclusions:

• Neural dynamics in the frontal cortex are consistent with internal simulations of external environmental states.
• The DMFC activity serves as a neural substrate for mental simulation during physical inference.
• Findings support the theory that the brain actively simulates the external world.