A pseudo-equilibrium thermodynamic model of information processing in nonlinear brain dynamics

Area of Science:

• Computational Neuroscience
• Thermodynamics
• Cognitive Science

Background:

• Current computational brain models lack speed and robustness in pattern recognition.
• Detecting subtle yet significant pattern fragments remains a challenge.

Purpose of the Study:

• To develop a novel computational model for brain dynamics.
• To improve speed and robustness in pattern recognition and knowledge retrieval.

Main Methods:

• Utilizing properties of thermodynamic systems operating far from equilibrium.
• Analyzing systems via linearization near adaptive operating points using root locus techniques.
• Employing reinforcement learning of conditioned stimuli to create attractor landscapes in sensory cortices.

Main Results:

• The model demonstrates enhanced pattern recognition by forming nerve cell assemblies and attractor landscapes.
• Knowledge retrieval is achieved through phase transitions induced by conditioned stimuli, leading to pattern self-organization.
• Near self-regulated criticality, cortical activity exhibits aperiodic null spikes, facilitating recognition and recall via a high signal-to-noise ratio.

Conclusions:

• The novel thermodynamic model offers improved speed and robustness in brain-inspired pattern recognition.
• Phase transitions and self-organized criticality are key mechanisms for efficient information processing and recall.
• The model's ability to capture weak signals suggests potential for understanding complex cognitive functions.