Modeling Semantic Fluency Data as Search on a Semantic Network.

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Modeling

Background:

• The semantic fluency task is a long-standing tool for investigating memory retrieval processes.
• Optimal foraging theory, modeled by censored random walks on semantic networks, has been proposed to explain semantic fluency data.
• However, semantic fluency data exhibit complex structures beyond optimal foraging predictions.

Purpose of the Study:

• To test various memory search processes on a semantic network.
• To determine which search process best captures the rich structure of human semantic fluency data.
• To refine computational models of memory retrieval.

Main Methods:

• Representing memory as a semantic network.
• Simulating diverse memory search strategies, including global exploration, local clustering, and strategic vs. non-strategic approaches.
• Comparing simulation outputs to empirical semantic fluency data.

Main Results:

• A censored random walk model incorporating a priming component demonstrated the highest fidelity to human data.
• This model effectively captured both the frequency of retrieved items and the clustering of semantically related items.
• The findings highlight the importance of both exploration and exploitation in memory search.

Conclusions:

• A censored random walk with priming provides a superior computational model for semantic fluency.
• This model offers deeper insights into the cognitive mechanisms underlying memory retrieval and organization.
• Future research can build upon this model to explore semantic memory structure and search dynamics.