A Cognitive Modeling Approach to Strategy Formation in Dynamic Decision Making

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Psychology

Background:

• Decision-making is a complex cognitive process crucial for navigating real-world scenarios.
• Dynamic decision-making involves adapting to changing environments based on feedback.
• Understanding these processes is key to advancing artificial intelligence and human cognition research.

Purpose of the Study:

• To develop and test a cognitive model simulating dynamic decision-making in a rule-based category learning task.
• To investigate how cognitive processes like perception, attention, and memory contribute to decision-making.
• To account for individual differences in decision-making performance.

Main Methods:

• Applied cognitive modeling using the ACT-R framework to a complex rule-based category learning task.
• Implemented a hybrid approach combining exemplar-based and rule-based strategies.
• Simulated participants' adaptation to feedback contingency reversals.

Main Results:

• The ACT-R model successfully learned the target category by employing one-feature then two-feature strategies, mirroring human learning patterns.
• The model demonstrated effective initial learning and adaptation to feedback reversal, consistent with human performance.
• Model performance showed greater variance and lower overall accuracy compared to human data, suggesting the need to incorporate additional cognitive factors.

Conclusions:

• Cognitive modeling provides a valuable framework for understanding dynamic decision-making and category learning.
• The developed ACT-R model offers a generalizable account of rule-based decision-making, adaptable to new stimuli.
• Future research should integrate psychobiological and neurophysiological data to refine models and address interindividual differences in decision performance.