Capturing human data in a letter counting task: accessibility and action-centeredness in representing cognitive...

Area of Science:

• Cognitive Psychology
• Learning Sciences
• Computational Neuroscience

Background:

• Current skill learning models often conflate action-centeredness and knowledge accessibility.
• The distinction between procedural and declarative knowledge can be imprecise.
• A unified model is needed to capture diverse aspects of skill acquisition.

Purpose of the Study:

• To present an integrated skill learning model incorporating implicit/explicit and action-centered/non-action-centered knowledge.
• To refine the understanding of knowledge representation in skill learning.
• To simulate human data and evaluate model performance.

Main Methods:

• Developed an integrated computational model of skill learning.
• Separated action-centeredness and accessibility aspects of knowledge.
• Simulated human performance data from a letter counting task.
• Compared model fits using action-centered knowledge alone versus combined knowledge.

Main Results:

• The integrated model successfully simulated human data in a letter counting task.
• A model incorporating both action-centered and non-action-centered knowledge provided a superior fit to the data.
• Demonstrated the distinct contributions of different knowledge types to skill learning.

Conclusions:

• The proposed integrated model offers a more nuanced perspective on skill learning.
• Separating action-centeredness and accessibility clarifies knowledge distinctions.
• Combined action-centered and non-action-centered knowledge better explains complex skill acquisition.