Consciousness: converging insights from connectionist modeling and neuroscience

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Neuroscience of Consciousness

Background:

• Recent findings suggest selective attention, working memory, and cognitive control rely on competition among distributed neural representations.
• Top-down projections, particularly from the prefrontal cortex, modulate this competition by enhancing specific representations.

Purpose of the Study:

• To review connectionist models implementing the competition-based view of cognitive functions.
• To highlight the relevance of these models for understanding consciousness.
• To explore similarities between these models and those implementing global workspace theory.

Main Methods:

• Review of existing connectionist models in cognitive neuroscience.
• Analysis of model architectures and principles.
• Comparison with global workspace theory implementations.

Main Results:

• Connectionist models successfully implement the view of competitive, top-down modulated representations in cognitive functions.
• These models share fundamental principles with global workspace theory.
• A core principle identified is global constraint satisfaction, prevalent in connectionist modeling for over two decades.

Conclusions:

• Connectionist models provide a valuable framework for understanding the neural basis of consciousness.
• The principle of global constraint satisfaction is key to integrating cognitive functions and consciousness.
• Further research integrating computational models and global workspace theory can advance our understanding of the mind.