Symbols and mental programs: a hypothesis about human singularity.

Area of Science:

• Cognitive Science
• Neuroscience
• Computational Linguistics

Background:

• Natural language is traditionally viewed as the sole driver of human cognitive uniqueness.
• Previous models often overlook the potential for non-linguistic symbolic systems in human cognition.

Purpose of the Study:

• To propose and investigate the existence of multiple internal "languages of thought" beyond natural language.
• To explore how these internal languages contribute to human cognitive abilities in various domains.

Main Methods:

• Developing discrete symbolic models inspired by computer languages.
• Analyzing human and non-human primate performance on perception tasks (e.g., shape, sequence).
• Comparing model predictions with behavioral data and brain activity.

Main Results:

• Minimum description length in proposed symbolic languages accurately predicts human behavior and neural data.
• Simpler, non-symbolic models fail to capture human performance but suffice for non-human primates.
• Identified distinct cortical circuits supporting these internal languages, separate from classical language areas.

Conclusions:

• Human cognition relies on multiple, discrete symbolic internal languages of thought.
• These languages are crucial for encoding and compressing complex structures in domains like mathematics and music.
• Supports a shift towards discrete symbolic models for understanding human thought processes.