Related Experiment Videos
Cognitive Architecture with Evolutionary Dynamics Solves Insight Problem.
Anna Fedor1, István Zachar2, András Szilágyi3
1Parmenides Center for the Study of Thinking, Parmenides FoundationPullach am Isartal, Germany; MTA-ELTE Theoretical Biology and Evolutionary Ecology Research GroupBudapest, Hungary; Institute of Advanced Studies Kőszeg (iASK)Kőszeg, Hungary.
Frontiers in Psychology
|April 14, 2017
Summary
Darwinian Neurodynamics, a cognitive model, simulates insight problem solving as an evolutionary process. This model successfully solves the four-tree problem, mirroring human performance and learning effects.
Area of Science:
- Cognitive Science
- Computational Neuroscience
- Artificial Intelligence
Background:
- Insight problem solving is a complex cognitive function.
- Understanding the neural mechanisms underlying insight remains a challenge.
- Existing models often lack a dynamic, evolutionary component.
Purpose of the Study:
- To introduce and validate Darwinian Neurodynamics, a novel cognitive architecture.
- To model the unconscious, evolutionary processes in insight problem solving.
- To investigate the effects of pretraining and priming on problem-solving performance.
Main Methods:
- Developed a neurally implemented cognitive architecture (Darwinian Neurodynamics).
- Modeled problem-solving as the evolution of solution patterns via attractor networks.
- Used human data for benchmarking and conducted experiments on pretraining and priming effects.
Main Results:
- The model demonstrated human-comparable performance, improving with appropriate pretraining and priming.
- A 'beginner's luck' effect was observed, with priming alone yielding the highest solution rate.
- Reduced computational capacity and learning abilities negatively impacted model performance.
Conclusions:
- Darwinian Neurodynamics offers a promising framework for modeling human insight problem solving.
- The model's evolutionary dynamics provide a new perspective on unconscious problem-solving mechanisms.
- Further research is warranted to explore the full potential of this cognitive architecture.