Prototype or Exemplar Representations in the 5/5 Category Learning Task

Area of Science:

• Cognitive psychology
• Computational neuroscience
• Machine learning

Background:

• Category learning theories often focus on structure, not temporal dynamics.
• Limited research explores how representations evolve during learning.
• Understanding representation shifts is crucial for cognitive models.

Purpose of the Study:

• Investigate factors influencing category representation acquisition and utilization.
• Examine how representations change over the time course of learning.
• Determine if exemplar or prototype strategies dominate in a novel task.

Main Methods:

• Developed a novel "5/5" categorization task.
• Employed computational modeling to analyze categorization strategies.
• Utilized schematic robot-like stimuli for controlled learning.

Main Results:

• Overall category learning was best described by prototype representations.
• A significant shift from exemplar to prototype strategies was observed.
• Exemplar representation use decreased, while prototype use increased across learning blocks.

Conclusions:

• Category learning is dynamic, with strategy shifts occurring over time.
• Prototype strategies become more dominant with extended learning.
• This highlights the adaptive nature of cognitive representations in learning.