Local Search and the Evolution of World Models

Area of Science:

• Cognitive Science
• Evolutionary Psychology
• Computational Neuroscience

Background:

• Understanding how humans form world models is a key cognitive science question.
• The generation of novel ideas from infinite possibilities remains an open challenge.

Purpose of the Study:

• To explore the role of evolutionary mechanisms in generating new ideas for world models.
• To propose that cognitive search for theories is an incremental, locally guided process.

Main Methods:

• Analogy drawn between cognitive search and biological evolution (blind variation and selection).
• Proposal of algorithms from program synthesis as potential mechanisms for human cognitive processes.
• Discussion of objections and implications for understanding human cognition.

Main Results:

• Cognitive innovation, particularly for global world models, is argued to be incremental.
• Local mutations and recombinations of existing models facilitate exploration and discovery.
• This process avoids the need to search for a theoretical "global optimum."

Conclusions:

• Human learning and theory development are driven by incremental exploration, akin to evolutionary processes.
• Understanding this process offers insights into cognitive biases like anchoring and confirmation bias.
• Program synthesis algorithms may offer computational models for these cognitive mechanisms.