A geometrical solution underlies general neural principle for serial ordering.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• A mathematical framework for sequential knowledge encoding in the brain is lacking.
• Understanding how the brain learns and recalls ordered information is a fundamental challenge.

Purpose of the Study:

• To propose a novel linear mathematical model for serial learning tasks.
• To explain the neural mechanisms underlying sequence encoding and transitive inference.

Main Methods:

• Developing a linear solution for serial learning based on mixed selectivity in neural state spaces.
• Utilizing classical conditioning to learn a "geometric" mental line for item representation.
• Applying the model to recurrent neural networks for motor decision tasks.

Main Results:

• The model successfully solves serial position tasks.
• It explains observed behaviors in human and animal transitive inference tasks, even with noisy data.
• The geometric mental line correlates with motor plans in recurrent neural networks.

Conclusions:

• Serial ordering may emerge from a monotonic mapping between sensory input and behavioral output.
• Motor-related associative cortices could play a key role in transitive inference.
• The proposed linear framework offers a new perspective on neural sequence encoding.