Brain-consistent architecture for imagination.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Computational Neuroscience

Background:

• Imagination is key to human intelligence, crucial for developing human-like AI.
• Current brain reverse engineering methods are limited to small regions, inadequate for complex functions like imagination.
• Imagination involves diverse brain areas including the neocortex, basal ganglia, thalamus, and hippocampus.

Purpose of the Study:

• To propose an advanced Function-Oriented SCID method for reverse engineering broader brain areas.
• To design a hypothetical computational architecture for the brain's imaginative capabilities.
• To bridge the gap between neuroscience and AI by modeling cognitive functions.

Main Methods:

• Developed a four-step Function-Oriented SCID method, an improvement on the existing SCID.
• Defined human imagination and identified six critical requirements for its simulation.
• Established constraints based on neocortex properties and developed five distinct functions.

Main Results:

• Proposed a computational architecture for imagination, mapping functions to brain regions.
• Neocortex: distributed memory for representation/execution.
• Claustrum: imaginary zone maker for dynamic partitioning.
• Thalamus/basal ganglia: routing conductor for manipulation.
• Hippocampus: recorder for history management.

Conclusions:

• The proposed architecture provides a fundamental cognitive framework for brain imagination.
• This work advances AI development by offering a neuro-biologically plausible model of imagination.
• Highlights the potential of reverse engineering complex cognitive functions from brain structure and function.