A Computational Model of Basic Addition Solving.

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Educational Psychology

Background:

• Learning arithmetic involves developing strategies like counting and memory retrieval.
• The interplay between procedural execution and memory access is crucial for skill acquisition.
• Previous models have not fully captured the dynamic co-development of these strategies.

Purpose of the Study:

• To present a computational learning model where procedural execution and memory retrieval co-develop.
• To investigate how simple arithmetic learning, specifically single-digit addition, is modeled.
• To account for strategy selection based on expected duration and practice effects.

Main Methods:

• Developed a computational model dynamically selecting between counting and memory retrieval based on expected duration.
• Incorporated a mechanism for accelerating counting through repeated practice.
• Tested the model on empirical data from adults learning alphabet arithmetic problems over 3 weeks.
• Simulated experiments manipulating problem structure (contiguous vs. noncontiguous sequences).

Main Results:

• The model successfully replicated the finding that larger problems are memorized earlier than smaller ones.
• It reproduced the observed effect of problem structure on strategy transition, with noncontiguous sequences favoring retrieval.
• Demonstrated that counting and retrieval strategies coexist post-learning, influenced by problem size and structure.

Conclusions:

• The developed model offers a unified account of arithmetic learning, integrating procedural and retrieval processes.
• The model's counting acceleration mechanism aligns with automated counting theory.
• Findings suggest a dynamic coexistence of strategies rather than a single dominant method in arithmetic learning.